AI-Driven Innovation in Intangible Cultural Heritage:A Semiotic Analysis of Door-God Woodblock Prints Using Diffusion Models

AI-Driven Innovation in Intangible Cultural Heritage:A Semiotic Analysis of Door-God Woodblock Prints Using Diffusion Models | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article AI-Driven Innovation in Intangible Cultural Heritage:A Semiotic Analysis of Door-God Woodblock Prints Using Diffusion Models junjun li, Sainan Zhang, Jiajie Li, Pengcheng Ju This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8747381/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract As a form of China’s intangible cultural heritage, woodblock New Year prints face challenges, including insufficient innovative expression and a weakening resonance with contemporary audiences. Using the “Qin Qiong and Jing De” Door-God prints as a case study, this research constructs a Peircean triadic semiotic model and, under controlled conditions, conducts two types of image-innovation studies: traditional manual design and diffusion-model-assisted design. Quantitative indicators are established using AHP and FCE to compare visual appeal, cultural identity, and emotional resonance. The results show that AI-assisted design has advantages in visual appeal and creative diversity, while traditional design performs better in cultural identity and semantic stability. Experiments with general audiences confirm that AI outputs are readable but still fall short of manual design in terms of stability with respect to cultural conventions. This study proposes a “diffusion models × semiotics × quantitative evaluation” framework that provides a heritage-science-oriented analytical framework for evaluating the semantic sustainability of AI-generated cultural heritage imagery. Humanities/Cultural and media studies Social science/Cultural and media studies Physical sciences/Mathematics and computing Intangible Cultural Heritage Diffusion Models AI-Assisted Design Peircean Semiotics Symbolic Reconstruction Door-God Woodblock Prints Figures Figure 1 Figure 2 Figure 3 1. Introduction In 1987, the World Commission on Environment and Development proposed the concept of sustainable development [ 1 ]. Culture is essential to the sustainable development of modern society [ 2 ]. In 2003, UNESCO launched the Convention for the Safeguarding of the Intangible Cultural Heritage, formally introducing the concept of “intangible cultural heritage”[ 3 ]. As an important form of China’s intangible cultural heritage, Yangjiabu woodblock New Year prints constitute an intergenerational visual-symbol system through their forms, colours, and narrative structures, playing a central role in maintaining cultural diversity and social memory [ 4 – 6 ]. Among them, Door-God imagery combines ritual and customary functions with cultural symbolic meaning, serving as a key visual carrier of traditional narratives and folk beliefs [ 6 ]. In today’s media environment, the readability of cultural-heritage symbols and the stability of their cultural semantics are insufficient [ 7 ]. Younger audiences often struggle to accurately interpret the meanings of traditional cultural symbols, leading to difficulties in the transmission of such symbols and a consequent decline in cultural reproduction capacity [ 8 , 9 ]. In addition, artificial design tends to be patterned in visual form, lacks innovative expression, and is difficult to adapt to new carriers such as digital media, which weakens the expression and contemporary value of cultural heritage symbols[ 10 ]. Against this background, how to overcome the limitations of traditional artificial design methods through the innovation of cultural heritage symbols has become an important issue for cultural inheritance and development [ 11 ]. With the rise of generative artificial intelligence, these technologies have attracted widespread social attention [ 12 ]. Diffusion models demonstrate a high capacity to fit visual styles and structural characteristics, enabling the generation of diverse image variants while preserving the basic semantic outline [ 12 , 13 ]. Their application potential in cultural-heritage visualisation has therefore drawn increasing interest [ 14 ]. Wang Shaofeng et al. examined the feasibility of generative design methods for Chinese woodblock New Year prints, outlined a blueprint for derivative works, and, within a methodological framework, explored three-dimensional design concepts to promote the dynamic safeguarding and wide dissemination of intangible cultural heritage[ 15 ]. To address the declining influence of traditional cultural symbols, Lin and other researchers investigated the application of traditional cultural symbols in art and design in the context of artificial intelligence[ 16 ]. Guo Mengyao et al. carried out innovation and preservation in the generative design of Yi ethnic embroidery patterns for cultural heritage based on LoRA [ 17 ]. Wang Yidan et al. developed a LoRA model embodying the blue clamp resist-dyeing technique, enabling the digital preservation of traditional patterns and their innovative reinterpretation [ 18 ]. AI technologies have made notable progress in the preservation and reproduction of intangible cultural heritage images [ 19 ], yet they still struggle to address the challenge of weakened cultural semantics in traditional imagery[ 20 , 21 ], as well as broader issues such as authenticity, subjectivity, and interpretive bias in cultural heritage [ 22 ]. At present, the following research gaps remain: (1) the extent to which diffusion models can preserve the structural logic of traditional visual symbols; (2) the differences between AI-assisted design and manual design at the semiotic level (representamen–object–interpretant); and (3) how to quantitatively evaluate AI’s innovativeness in cultural symbols and its degree of cultural-semantic preservation. These gaps constrain the authenticity and practical usability of AI in the preservation, interpretation, and dissemination of cultural-heritage imagery. To explore the mechanism through which diffusion models operate in the semiotic reproduction of cultural-heritage symbols, this study introduces Peirce’s triadic semiotic framework (representamen–object–interpretant) to describe the sign chain linking visual form, cultural reference, and meaning interpretation [ 23 ]. This framework enables the construction of an interpretable analytical system for AI-generated cultural-heritage imagery, but it has not yet been sufficiently validated in diffusion-model image research. Building on the limitations of existing research, this study makes the following contributions: (1) It proposes an analytical framework that integrates diffusion models with Peircean triadic semiotics, enabling a structured description of the generative mechanisms of traditional visual symbols and the interpretive biases that may arise in AI generation. (2) It develops a quantitative evaluation system combining AHP and FCE, translating symbolic innovativeness and cultural-semantic preservation into comparable and reusable metrics. (3) It designs a controlled comparative experimental procedure between manual design and AI-assisted design, systematically testing semiotic differences in symbol generation across the two design pathways and providing methodological support for the redesign of cultural-heritage symbols. 2. Related research As an element of China’s intangible cultural heritage, woodblock New Year prints have a long history and serve as an important vehicle of traditional visual art [ 24 ]. Characterised by concise yet exaggerated imagery, bright and auspicious colours, and romantic modes of expression, they function as cultural symbols that carry social norms, ethical values, and collective emotions [ 25 , 26 ]. Door-God New Year prints display distinct semiotic features, such as paired figures, symmetrical compositions, highly saturated colours, and signifying elements including weapons, armour, and decorative patterns [ 27 ]. These decomposable and quantifiable symbolic elements make Door-God prints an ideal subject for analysing the visual structure and semiotic logic of cultural heritage. With the development of society and the transformation of media, cultural heritage has been continuously impacted in terms of inheritance, production mechanisms, and social needs. Its living space in modern society has been compressed, and it faces the risk of disappearance[ 28 ]. In particular, New Year pictures, which are rooted in the lifestyle of traditional agrarian society, have been weakened in the process of social structural transformation and life scene change, and their original usage context and dissemination channels have been continuously weakened, making them more vulnerable[ 29 ].Therefore, the traditional symbol system of New Year pictures not only needs to be preserved, but also needs to be reactivated and contemporary translated through design language so that its symbolic meaning can be continued and regenerated in the contemporary communication context[ 30 ]. In recent years, high-precision scanning, 3D modelling, and digital archiving systems have significantly improved the preservation quality and accessibility of traditional images, enabling the retention of surface features such as colour and texture and expanding the dissemination of intangible cultural heritage through online platforms [ 31 – 33 ]. However, digital technologies primarily focus on recording visual appearance and lack explanatory capacity for deeper semiotic semantics such as character relationships, symbolic objects, and ritual logic [ 34 ]. Relying on digitisation alone is therefore insufficient to sustain the cultural continuity of the symbolic system, which creates both the necessity and research space for AI to participate in the preservation and enhancement of cultural heritage [ 35 ]. Diffusion models have demonstrated innovative capabilities in image generation, texture synthesis, and style transfer [ 36 , 37 ], and have gradually been applied to tasks such as restoration, inpainting, and stylistic reproduction of cultural-heritage imagery [ 38 , 39 ]. They can learn the artistic style of traditional images and, guided by textual or exemplar prompts, generate consistent outputs that reconstruct traditional New Year print imagery at the visual level [ 40 ]. Whether diffusion models can simultaneously preserve the structural logic of traditional symbols, their ritual references, and their cultural-semantic relationships during image generation still lacks systematic empirical verification [ 41 ]. Peirce’s triadic semiotic model (representamen–object–interpretant) provides a theoretical framework for analyzing the generation, reference, and interpretation of visual signs [ 42 , 43 ]. Sign meaning is not a static correspondence; rather, it is dynamically constructed through a “triadic semiosis” process jointly shaped by visual form, cultural reference, and audience interpretation [ 44 ]. The triadic semiotic model enables researchers to decompose complex images into operable sign elements, thereby identifying differences in semantic preservation and semantic drift across different generative mechanisms [ 45 , 46 ]. In cultural-heritage image research, Peircean semiotics can serve as an important tool for examining whether AI follows traditional semiotic logic. However, there remains a lack of studies that systematically apply this framework to diffusion-model image analysis, and systematic evidence is still insufficient regarding issues such as referential misplacement and the weakening of symbolic structures that may arise during AI-based sign generation. With the development of generative artificial intelligence, design activities have gradually shifted from one-way creation to a human–AI co-creation model [ 47 – 50 ]. Diffusion models have advantages in image generation and stylistic diversification, while designers remain irreplaceable in semantic judgment, contextual understanding, and value trade-offs [ 51 ]. For the visual systems of intangible cultural heritage—where ritual norms and symbolic referential relationships are clearly defined—symbol generation involves not only formal innovation but also relies heavily on accurate cultural reference and consistency in interpretation. In this context, human–AI collaboration is no longer merely a strategy for improving design efficiency; it may also function as an important mechanism for mitigating the risks of semantic drift and symbol misplacement during AI generation. However, existing research on human–AI collaboration has largely focused on general design or commercial creative tasks. For intangible cultural heritage, a highly semantically constrained symbolic system, the specific ways in which human–AI collaboration operates at the level of sign generation still lack systematic empirical investigation. In summary, existing studies—approaching the topic from perspectives such as digital preservation, generative artificial intelligence, semiotic analysis, and human–AI collaboration—have provided important theoretical and technical foundations for the reproduction and innovation of cultural-heritage imagery. However, much of this research focuses on the reproduction of visual style, evaluation of generative outcomes, or general collaboration mechanisms, while systematic examination remains insufficient regarding whether generative models follow traditional semiotic structures, ritual references, and cultural-semantic relationships at the level of cultural-heritage symbols. In particular, within intangible cultural-heritage visual systems that are highly symbolic and narrative in function, there is still a lack of empirical research that conducts semiotic-level comparative analyses between diffusion-model outputs and manual designs. There is also a lack of a quantitative methodological framework capable of simultaneously evaluating symbolic innovativeness and the degree of cultural-semantic preservation. Therefore, it is necessary to introduce a research approach that balances interpretability of the symbol-generation mechanism with comparability of evaluation results, in order to systematically analyse the role and mechanism of AI-assisted design in the semiotic reproduction of intangible cultural heritage. This research provides the theoretical basis for the methodological choices and research design developed in this study. 3. Methods The study employs a parallel controlled mixed experimental design [ 52 ]. It examines the role of diffusion models in the reproduction of traditional New Year print symbols from two levels: cultural-heritage symbol production and symbol perception (see Fig. 1 ). Study 1 compares the generation pathways of traditional manual design and diffusion-model-assisted design, constructs a semiotic variable structure based on Peirce’s triadic semiotics, and combines the Analytic Hierarchy Process (AHP) and Fuzzy Comprehensive Evaluation (FCE) to quantitatively evaluate differences between the two design approaches across visual, cultural, and emotional dimensions. Study 2 employs a controlled behavioural experiment and questionnaire measures to test perceptual differences among general audiences in symbol decoding, cultural-semantic understanding, and emotional responses. Together, the two studies constitute an integrated analytical framework spanning generative mechanisms, expert judgment, and audience experience. Study 1: Symbolic innovation mechanisms from expert and designer perspectives In order to ensure the statistical effectiveness of design samples, this study uses G*Power for pre-performance analysis[ 53 ] and recruits 8 design practitioners with different experience backgrounds (6 graduate students in the visual communication direction and 2 professional designers with more than 5 years of experience), all of whom have experience in traditional image creation or AIGC tools. Another 10 expert group, including 3 non-heritage research scholars and 7 design theory or practice experts, was set up to fill in the AHP judgement matrix and complete the FCE score. This study selected the most representative “Qin Qiong and Jing De” Door-God prints from the Yangjiabu woodblock New Year prints as the experimental materials (see Fig. 2 ). The imagery contains representative symbolic elements—such as figure styling, costume patterns, and weapon types [ 6 , 54 ]—which facilitates the extraction of sign elements and subsequent quantitative modelling [ 27 ]. Two design pathways were established—manual design (Group A) and AI-assisted design (Group B)—to compare differences in symbol generation, semantic preservation, and formal innovation across different design mechanisms. Both groups were based on the same “Qin Qiong and Jing De” Door-God theme and used an identical system of sign elements to ensure consistent control conditions. The independent variable was the type of design pathway; the control variables were the cultural theme and symbolic features; and the dependent variables were the evaluation outcomes constructed from Peirce’s semiotic structure. The corresponding relationships are shown in Table 1 . Table 1 Experimental Variables and Their Semiotic Logic Semiotic Level Variable Type Variable Content Representamen Independent Variable Design Pathway: Traditional vs. AI-assisted Object Controlled Variable Unified Door God Theme (Qin Qiong & Yuchi Gong) Interpretant Dependent Variable Audience/Expert Evaluations(C1–C3 + Innovativeness) Group A adopted a linear manual workflow centred on sketching, line drawing, and digital colouring, in which the designer reconstructed symbols based on their own cultural knowledge and modelling experience. Group B used the DALL·E 3 diffusion model to generate initial images; the designer controlled the theme and composition through prompts and then performed localised revisions and semantic calibration of the outputs. The two pathways respectively represent an experience-driven traditional creation process and a human–AI collaborative mechanism for symbol generation. Their key differences and technical characteristics are summarised in Table 2 . Table 2 Experimental Group Configuration Group Design Pathway Description Technical Approach & Features Mode of Semiotic Reproduction Group A: Traditional Manual Design Designer-led manual creation following sketch → line drawing → final rendering Hand drawing + digital coloring; relies on designers’ cultural knowledge and visual experience Subjective, experience-based re-encoding of traditional symbols Group B: AI-Assisted Design DALL·E 3 diffusion model used to generate drafts; designers iteratively refine prompts and edit outputs editing AI-generated candidates. Prompt engineering + image editing; emphasizes AI’s efficiency in form generation and semantic variation diffusion, and form generation. Human–AI collaborative encoding with algorithmic mediation between prompt language and algorithmic output. To ensure comparability between the two pathways, the design procedures for Groups A and B were standardised: both groups were required to produce three final images, all developed around the same Door-God theme. In Group A, the designer generated and integrated the main symbolic elements. In Group B, the design was completed through a collaborative process of prompt-based control, diffusion-based generation, and manual revision. All design processes were documented for subsequent semiotic analysis, and the final images were used for expert review and audience experiments. The relevant workflow and output structure are shown in Fig. 3 . Based on Peirce’s triadic semiotics, this study constructs a symbolic innovativeness evaluation structure consisting of three criterion layers (C1–C3) and nine sub-indicators (S1–S9) (see Table 3 ), decomposing the visual form and cultural semantics of traditional New Year prints into quantifiable evaluation dimensions. Experts conducted pairwise comparisons of the indicators using a 1–9 ratio scale to construct judgment matrices and calculate weights (see Table 4 ). All matrices passed the consistency test. On this basis, FCE was used to build the membership matrices for the nine sub-indicators and, combined with the overall weights, to obtain innovativeness indices for the two design pathways [ 55 ]. The complete definitions of the evaluation sets, membership construction, and calculation procedures are provided in the Appendix. Table 3 AHP Hierarchical Structure for Symbolic Cognition Optimisation in Intangible Heritage Goal (G) Criterion (C) Sub-Criterion (S) Symbolic Cognition Optimization in Intangible Heritage Design (G) C1 Visual Attraction S1 Qualisign (sensory attributes) S2 Sinsign (stylistic coherence) S3 Legisign (compositional structure) C2 Cultural Identity S4 Iconic Object (heroic prototype) S5 Indexical Object (ritual reference) S6 Symbolic Object (cultural convention) C3 Emotional Resonance S7 Rheme (emotional impression) S8 Dicent (functional cognition) S9 Argument (value and belief) Table 4 Comprehensive Weight Results of Design Elements (AHP Final Output) Goal (G) Criterion (C) Primary Weight Sub-Criterion (S) Secondary Weight Composite Weight (Wi) Rank Symbolic Cognition Optimization in Intangible Heritage Design C1 Visual Attraction 0.4013 S1 Qualisign 0.4344 0.1743 1 S2 Sinsign 0.2432 0.0976 5 S3 Legisign 0.3224 0.1294 3 C2 Cultural Identity 0.3361 S4 Iconic Object 0.3141 0.1056 4 S5 Indexical Object 0.4711 0.1583 2 S6 Symbolic Object 0.2148 0.0722 9 C3 Emotional Resonance 0.2626 S7 Rheme 0.3428 0.0900 7 S8 Dicent 0.3660 0.0961 6 S9 Argument 0.2912 0.0765 8 Study 2: Audience perception experiment Study 2 aims to examine whether Groups A and B from Study 1 differ in symbol readability, cultural-semantic understanding, and emotional responses among general audiences. This experiment employs a controlled behavioural experiment and questionnaire measures to compare perceptual differences between the two image sets across three levels: visual representation (Representamen), cultural semantics (Object), and emotional interpretation (Interpretant). A total of 114 general participants were recruited for this study. Among them, 56.14% were female, and 43.86% were male; 42.98% were from design-related majors, and 57.02% were from non-design majors. All participants completed the questionnaire anonymously and provided informed consent, and the research procedure complied with ethical guidelines. The experimental stimuli were drawn from the final design outputs of Study 1. Each group included three images, and all images were presented in a randomised order to reduce order effects. This study set up three types of evaluation tasks to cover the form, reference, and meaning structure of the images. The representamen task assessed the clarity and coherence of color, form, and composition. The object task assessed the balance of character identity, cultural meaning, and cultural conventions. The interpretant task assessed emotional impression, degree of meaning understanding, and cultural pride. All items used a five-point Likert scale to rate the two image sets separately, producing paired data. 4. Results 4.1 Study 1: Generation Outcomes of the Two Design Pathways Group A presented a manual design approach using sketches, line drawings, and digital coloring. The overall visual style was stable, and the cultural symbol structure remained intact. Group B, on the other hand, employed a collaborative model based on diffusion model image generation and manual correction. While exhibiting significantly stronger visual novelty, it showed some instability in certain details. The two design models demonstrated differences in stability and novelty regarding their symbol generation mechanisms. 4.2 Study 1: Results of the Three-Level Semiotic Analysis At the Representame level of semiotics, Group A maintains the visual elements of traditional New Year paintings relatively well; Group B exhibits greater variation in visual style but is weaker in expressing proportions and details in local image areas. At the Object level, Group A accurately presents figures, weapons, and ritualistic connotations; while Group B maintains thematic consistency, it is prone to issues such as mixed use of weapons and misplaced patterns. At the Interpretant level, Group A coherently presents cultural semantics; Group B excels in visual novelty but is slightly weaker in emotional depth and symbolic meaning. 4.3 AHP Weighting Model Results All judgment matrices passed the consistency test (CR < 0.10). The weights of the first-level criteria indicate that visual appeal was the highest (C1 = 0.4013), followed by cultural identity (C2 = 0.3361) and emotional resonance (C3 = 0.2626). Among the sub-indicators, qualisign (S1), indexical object (S5), and legisign (S3) ranked in the top three, suggesting that visual presentation, traditional belief references, and compositional style are key factors influencing innovativeness. 4.4 FCE Results The overall innovativeness scores were 92.38 for Group A and 91.94 for Group B. Both groups reached a high level, and the difference was minimal, indicating that diffusion models have achieved a level comparable to manual design in terms of formal innovation and style generation, while their cultural-semantic consistency remains slightly weaker than that of manual design. 4.5 Study 2: Results of the General-Audience Experiment Data analysis employed descriptive statistics and paired-samples t-tests to compare the differences between the two groups of images across various symbol dimensions. The scale demonstrated good reliability and validity (Cronbach’s α = 0.967, KMO = 0.908, Bartlett p < .001). Scores across the three semiotic dimensions (C1–C3) were all in a relatively high range (4.0–4.3), and the differences between Groups A and B in overall visual performance and cultural readability were small (see Table 5 ). In the subsequent paired-samples t-tests, most indicators showed no significant differences (see Table 6 ). Among them, A3 (composition clarity) was close to significance, while A6 (balance of cultural conventions) showed a significant difference (p = .034), indicating that participants tended to favour manual design in judgments related to cultural semantics, whereas the two groups performed largely similarly at the visual level. Table 5 Descriptive Statistics of the Three Semiotic Dimensions (C1–C3) in Study 2 Dimension Group A Mean Group A SD Group B Mean Group B SD Visual appeal (C1) 4.27 0.82 4.15 0.94 Cultural identity (C2) 4.32 0.88 4.20 0.94 Emotional resonance (C3) 4.12 0.96 4.14 0.99 Table 6 Paired-Samples t-Test Results (Group A vs. Group B) Dimension Indicator Number Indicator Content Group A Mean Group B Mean t value p value Significance Visual appeal (C1) A1 Novelty of color and shape 4.175 4.088 1.043 0.299 — A2 Style coordination 4.333 4.202 1.565 0.120 — A3 Compositional clarity 4.298 4.149 1.706 0.091 Marginal significance Cultural identity (C2) A4 Character recognition 4.360 4.237 1.221 0.225 — A5 Cultural connotation expression 4.263 4.211 0.551 0.583 — A6 Cultural norms balance 4.325 4.158 2.144 0.034 Significant Emotional resonance (C3) A7 First impression and emotional reaction 3.930 4.009 0.904 0.368 — A8 Meaning comprehension 4.167 4.184 0.187 0.852 — A9 Cultural pride 4.254 4.228 0.274 0.785 — 4.6 Summary Synthesising the findings from both studies, Group A performed better in cultural-semantic stability and semiotic consistency, whereas Group B showed advantages in image generation and stylistic divergence. The AHP–FCE composite innovativeness index indicated that the two groups achieved similar scores (92.38 vs. 91.94), suggesting that AI’s capacity for visual innovation has approached that of manual creation. The audience experiment further showed no significant difference between the two groups in overall visual performance; however, regarding the balance between traditional and modern semantics, participants’ evaluations tended to favour manual design. 5. Discussion This study examines the expressive power of the diffusion model in the redesign of New Year pictures from two aspects: the generation mechanism of cultural heritage symbols and the audience perception results. The results of Study 1 and Study 2 show that the diffusion model has significant advantages in the generation of visual forms, but has certain limitations in terms of cultural semantic consistency and the stability of symbolic meaning interpretation. This phenomenon of enhanced visual form innovation but weakened cultural semantic stability echoes the relevant discussions in the field of heritage science on the quality of visual representation and cultural adaptability [ 56 ]. From a semiotic perspective, the image generated by the diffusion model based on the AHP weighting analysis results achieved the highest weight in terms of visual appeal, and the audience also gave positive feedback on the visual novelty of the generated image. These results suggest that diffusion models have strong generative capability at the level of the representamen, effectively learning and recombining the colour styles, compositional features, and formal language of traditional New Year prints. This characteristic aligns with prior studies arguing that diffusion models tend to prioritise learning visual representations and, due to a lack of inherent constraints regarding semantic meaning and semiotic rules, are prone to producing spurious correlations [ 57 ]. However, both the FCE innovativeness evaluation and the audience symbol-decoding tasks showed that diffusion-model-generated images still exhibited varying degrees of semantic drift from New Year prints, including weakened logical coherence in local details, misplacement of symbolic objects, and instability in figure-identity reference. This indicates that while diffusion models can reproduce traditional visual features, they have not yet formed a stable structure of cultural conventions to support meaning construction at the object and interpretant levels. Further comparison between the manual design and AI-assisted design pathways shows that human–AI collaboration produces a clear complementary effect in the reproduction of traditional symbols. Diffusion models excel in exploring visual forms and diversifying styles, whereas manual design demonstrates greater stability in object reference, recognition of ritual conventions, and grasp of narrative value. In particular, with respect to the expression of Door-God figure identity, the combination of symbolic objects and their cultural implications, manual design performed significantly better than diffusion-model outputs at the interpretant level. This finding supports prior research emphasising the crucial role of designers in cognitive decision-making and value trade-offs within human–AI collaboration [ 58 ]. The results suggest that, in intangible cultural heritage visual systems characterised by strong ritual norms and symbolic functions, human–AI collaboration can help mitigate the risk of semantic drift during generative-model-based symbol production. From the perspective of symbolic sustainability, the findings of this study can be further interpreted. The diverse visual forms generated by diffusion models can facilitate the re-dissemination of cultural-heritage symbols across media contexts and, to some extent, increase young audiences’ visual interest, thereby alleviating the decline in symbol readability in contemporary communication environments. However, Study 2 revealed significant differences in its cultural normative balance indicators. When the evaluation involved cultural significance, customary rationality, and symbolic consistency, the audience tended to favor artificially designed images. This aligns with the argument in cultural heritage research that digital transformation should not only focus on formal innovation and protection but also on sustainability at the social, environmental, and economic levels [ 58 , 59 ]. This suggests that AI-generated images are suitable as a medium for visual communication and formal exploration in the field of cultural heritage, rather than as the subject of the symbolic meaning of cultural heritage. The quantitative results of AHP–FCE reveal the difference between the innovativeness and semantic stability of cultural heritage symbols. The highest-weighted prime signifier (S1), indicator object (S5), and type signifier (S3) are key elements in maintaining the visual form and cultural semantics of traditional New Year paintings. Theoretically, diffusion models can learn the visual form rules of symbols through large-scale data learning. However, in traditional cultural systems that are highly dependent on context, customs, and collective memory, the cultural semantics of symbols are still profoundly influenced by traditional customs and social contracts. This has important implications for the research on the digitization of cultural heritage and cultural sustainability. The introduction of generative artificial intelligence requires a corresponding semantic supervision mechanism and expert knowledge calibration to avoid semantic deviation in the reproduction of cultural information. This finding is consistent with the relevant discussions in the current research on the authenticity of AI-generated content [ 60 , 61 ]. From the methodological perspective of heritage science, the "diffusion model × semiotics × quantitative evaluation" framework proposed in this study analyzes cultural heritage semiotically and transforms the degree of semantic shift of symbols into quantifiable indicators. This allows conservation agencies and design teams to more accurately digitally reorganize the symbolic elements of cultural heritage, thereby establishing an adjustable balance mechanism between visual innovation and cultural semantic protection. 6. Conclusion This study takes Yangjiabu door god New Year paintings as a case study to investigate the innovative expression of generative artificial intelligence in the redesign of visual symbols of intangible cultural heritage. The results show that the diffusion model can effectively expand visual expression forms in the redesign of visual symbols of cultural heritage, but it is weak in terms of cultural reference and semantic interpretation stability, thus clarifying the positioning of generative artificial intelligence in the context of cultural heritage. Building upon this foundation, this paper proposes and validates an analytical framework integrating a diffusion model, Peircean's triadic semiotics, and a quantitative evaluation method. This framework is used to analyze the performance characteristics and potential risks of the diffusion model in the redesign of cultural heritage symbols. It provides a structured tool for distinguishing different levels of symbols, such as visual representation, cultural reference, and meaning interpretation, and offers a reusable methodology for related research. This study does not advocate replacing human design with artificial intelligence, but rather emphasizes the role of human-machine collaboration in designer-led design processes. Through explicit semantic oversight and symbol calibration, generative AI can support visual innovation while avoiding cultural semantic shifts. Overall, this study provides methodological and practical insights for the participation of generative artificial intelligence in the visual redesign of intangible cultural heritage, and lays an analytical foundation for discussions on the sustainability of cultural heritage symbols in the age of artificial intelligence. Future research can further test the applicability of this framework in different heritage types and cultural contexts, and explore more refined human-machine collaboration mechanisms to promote a long-term balance between innovative expression and semantic preservation in the digital practice of cultural heritage. Declarations Ethics Approval and Consent to Participate This study was conducted in accordance with the academic research ethics guidelines of the authors’ affiliated institutions. Given that the research did not involve medical experiments or high-risk interventions with human participants, formal approval from an Institutional Review Board (IRB) was not required. Informed Consent All human participants involved in this study were fully informed about the research objectives and procedures prior to participation and voluntarily provided their informed consent. The study ensured participant anonymity and data security throughout the research process. Funding This work was supported by the Talent Start-up Project of the Scientific Research and Development Fund of Zhejiang A&F University (Grant No. 2024FR043). Data Availability The data generated or analyzed during the current study are available from the corresponding author upon reasonable request. Author Contribution Conceptualization, Junjun Li and Sainan Zhang; Methodology, Junjun Li; Formal Analysis, Jiajie Li; Investigation, Pengcheng Ju; Writing—original draft, Junjun Li; Writing—review and editing, Sainan Zhang; Supervision, Sainan Zhang; Funding acquisition, Sainan Zhang. References Environment, W.C.o., Our common future . 1987: Peterson's. Lisitza, A. and G. Wolbring, Sustainability within the academic EcoHealth literature: Existing engagement and future prospects. Sustainability, 2016. 8 (3): p. 202. Teke, S.G., Living Heritages and Fixed Traditions in the Lists of the Convention for the Safeguarding of Intangible Cultural Heritage. Milli Folklor, 2018(120): p. 19-31. Shen, H., Classic Chinese New Year Paintings . 2015, Haitian Publishing House: Shenzhen, China. Xu, Y.W., Y. Tao, and B. Smith, China's emerging legislative and policy framework for safeguarding intangible cultural heritage. International Journal of Cultural Policy, 2022. 28 (5): p. 566-580. Liang, J.Y., R.T. Guo, and Y.Q. He, A Psychological Analysis of the Imagery of Chinese Menshen. Culture & Psychology, 2022. 28 (4): p. 527-549. Baojie, J., Exploring the Integration of Chinese Traditional Elements in Visual Communication and Its Influence on Cultural Identity and Public Perception: A review. Peta International Journal of Social Science and Humanity, 2025. 4 (3): p. 1-28. Cao, H. and S. Champadaeng, The Art of Chinese Calligraphy: Educational Protection and Literacy Study of Cultural Heritage. International Journal of Education and Literacy Studies, 2024. 12 (3): p. 160-171. Cai, Z., et al., Spatial distribution characteristics and sustainable inheritance strategies of national traditional fine arts intangible cultural heritage in China. Sustainability, 2024. 16 (11): p. 4488. Shen, Y., Y.-c. Cai, and X.-y. Wang, Intangible cultural heritage in industrial design. Digital Scholarship in the Humanities, 2024. 39 (1): p. 354-372. Xie, J., Innovative design of artificial intelligence in intangible cultural heritage. Scientific Programming, 2022. 2022 (1): p. 6913046. Cao, Y., et al., A survey of ai-generated content (aigc). ACM Computing Surveys, 2025. 57 (5): p. 1-38. Oyedokun, G.E. Cultural Heritage and Intellectual Property Rights in the Digital Age: A UNESCO 2005 Convention Perspective . in Cultural Odyssey: 20 Years of Implementation of UNESCO's 2005 Convention in Nigeria . 2025. Zhou, Y., et al., Fine-tuning diffusion model to generate new kite designs for the revitalization and innovation of intangible cultural heritage. Scientific Reports, 2025. 15 (1): p. 7519. Wang, S.F. and C.C. Chen, Revitalizing Intangible cultural heritage via derivative design: A focus on chinese woodblock printing. Plos One, 2025. 20 (4). Lin, C., Application of traditional cultural symbols in art design under the background of artificial intelligence. Mathematical Problems in Engineering, 2021. 2021 (1): p. 1258080. Guo, M.Y., et al. LoRA-Based Pattern Generation for Yi Ethnic Embroidery Heritage Preservation . in 17th Conference on Creativity and Cognition-C&C . 2025. null, ELECTR NETWORK. Wang, Y.D. and Y.X. Zhou, Artificial Intelligence-Driven Interactive Experience for Intangible Cultural Heritage: Sustainable Innovation of Blue Clamp-Resist Dyeing. Sustainability, 2025. 17 (3). Liu, Y., Application of Digital Technology in Intangible Cultural Heritage Protection. Mobile Information Systems, 2022. 2022 . Siliutina, I., et al., Cultural preservation and digital heritage: challenges and opportunities. Amazonia Investiga, 2024. 13 (75): p. 262-273. Khan, A.S., The preservation of cultural heritage: challenges and opportunities in the digital age. Al-Behishat Research Archive, 2023. 1 (01): p. 16-24. Tiribelli, S., et al., Ethics of Artificial Intelligence for Cultural Heritage: Opportunities and Challenges. Ieee Transactions on Technology and Society, 2024. 5 (3): p. 293-305. Fredin-Ortiz, E., Charles S. Peirce's Semiotics and the Concept of Mental Representation in Cognitive Science. Designis, 2025(43): p. 271-279. Nie, H. and M.Y. Ran. Study on the Woodblock New Year Pictures of Zhuxianzhen Town . in 2nd International Conference on Art Studies: Science, Experience, Education (ICASSEE) . 2018. Moscow, RUSSIA. Kryzhko, O., National and Cultural Symbolism of Zoo-Anthropomorphic Creatures in Ukrainian Myphological Picture of the World. LOGOS-A Journal of Religion, Philosophy, Comparative Cultural Studies and Art, 2019(101): p. 159-168. Gao, Y., Y. Zhang, and I. Destech Publicat. The Fusion of Traditional New Year Pictures and Modern Decoration Development-The Case of Hebei Wuqiang New Year Pictures . in International Conference on Education and Educational Research and Environmental Studies (EERES) . 2016. Hong Kong, PEOPLES R CHINA. Su, T.C., et al., Style Recognition of Door God Paintings by Hypothesis Testing for Texture Features of Painting Patterns. Applied Sciences-Basel, 2022. 12 (5). Zhang, L., et al., Mastersu: the sustainable development of su embroidery based on digital technology. Sustainability, 2022. 14 (12): p. 7094. Zhang, B., et al., Can AI-generated art stimulate the sustainability of intangible cultural heritage? A quantitative research on cultural and creative products of New Year Prints generated by AI. Heliyon, 2023. 9 (10). Que, Y.X., T. de la Hera, and J. Jansz, Dive Into the Past: Chinese Intangible Cultural Heritage Represented in Chinese Mobile Games. Media and Communication, 2025. 13 . Triviño-Tarradas, P., D.F. García-Molina, and J.I. Rojas-Sola, Impact of 3D digitising technologies and their implementation. Technologies, 2024. 12 (12): p. 260. Banfi, F., et al., Narrating Ancient Roman Heritage through Drawings and Digital Architectural Representation: From Historical Archives, UAV and LIDAR to Virtual-Visual Storytelling and HBIM Projects. Drones, 2023. 7 (1). Zhang, C.Y., et al., Integrated high-precision real scene 3D modeling of karst cave landscape based on laser scanning and photogrammetry. Scientific Reports, 2024. 14 (1). Jiang, Y., et al., National representations in digital heritage exposure: exploring young people’s in-group/out-group categorisation on cultural artefact visuals and related evaluations. Humanities and Social Sciences Communications, 2025. 12 (1): p. 1743. Colace, F., et al., New AI challenges for cultural heritage protection: A general overview. Journal of Cultural Heritage, 2025. 75 : p. 168-193. Cao, T., et al. Texfusion: Synthesizing 3d textures with text-guided image diffusion models . in Proceedings of the IEEE/CVF international conference on computer vision . 2023. Chen, H., et al., Comprehensive exploration of diffusion models in image generation: a survey. Artificial Intelligence Review, 2025. 58 (4): p. 99. Hsieh, K., et al. Cultural Heritage Meets AI: Advanced Text-to-Image Models for Digital Reconstruction and Preservation . in 2024 6th International Conference on Control and Robotics (ICCR) . 2024. IEEE. Yu, T., et al., Artificial intelligence for Dunhuang cultural heritage protection: the project and the dataset. International Journal of Computer Vision, 2022. 130 (11): p. 2646-2673. Zhang, C., et al., Text-to-image diffusion models in generative ai: A survey. arXiv preprint arXiv:2303.07909, 2023. Santoro, A., et al., Symbolic behaviour in artificial intelligence. arXiv preprint arXiv:2102.03406, 2021. Ortiz, E.F., La semiótica de Charles S. Peirce y el concepto de representación mental en la ciencia cognitiva. DeSignis: Publicación de la Federación Latinoamericana de Semiótica (FELS), 2025(43): p. 271-279. Lorino, P., Charles Sanders Peirce (1839–1914). Oxford handbook of process philosophy and organization studies, 2014: p. 143-165. Peirce, C.S., Peirce on signs. 1991. Yunhee Lee, Narrative cognition and modeling in new media communication from Peirce's semiotic perspective. Semiotica, 2012. 2012 (189): p. 181-195. Luo, Q., Understanding artworks from Danto’s philosophy of art: a Peircean semiotic approach. Chinese Semiotic Studies, 2023. 19 (4): p. 665-685. Fui-Hoon Nah, F., et al., Generative AI and ChatGPT: Applications, challenges, and AI-human collaboration . 2023, Taylor & Francis. p. 277-304. Puerta-Beldarrain, M., et al., A multifaceted vision of the Human-AI collaboration: a comprehensive review. IEEE Access, 2025. Napoleão, E. and A.D. Silva, Design Practice with Generative Artificial Intelligence for Sustainability: An Experimental Study with Undergraduate Designers , in Artificial Intelligence-Aided Design for Sustainability . 2025, Springer. p. 75-93. Li, W., et al., Exploring human-machine collaboration paths in the context of AI-generation content creation: a case study in product styling design. Journal of Engineering Design, 2025. 36 (2): p. 298-324. Saldana Ochoa, K., Can Artificial Intelligence Mark the Next Architectural Revolution? Design Exploration in the Realm of Generative Algorithms and Search Engines , in Decoding Cultural Heritage: A Critical Dissection and Taxonomy of Human Creativity through Digital Tools . 2024, Springer. p. 3-22. Schoonenboom, J. and R.B. Johnson, How to construct a mixed methods research design. Kolner Zeitschrift fur Soziologie und Sozialpsychologie, 2017. 69 (Suppl 2): p. 107. Faul, F., et al., Statistical power analyses using G* Power 3.1: Tests for correlation and regression analyses. Behavior research methods, 2009. 41 (4): p. 1149-1160. Peng, X. Study on the Application and Research of Visual Symbols of Traditional New Year Pictures in Modern Design . in The 6th International Conference on Arts, Design and Contemporary Education (ICADCE 2020) . 2021. Atlantis Press. Ho, W. and X. Ma, The state-of-the-art integrations and applications of the analytic hierarchy process. European Journal of Operational Research, 2018. 267 (2): p. 399-414. Lin, Y.H., H.L. Liu, and N. Yang. The Application Strategies of AIGC Technology Empowering Chinese Traditional Cultural Elements in Graphic Design . in 27th International Conference on Human-Computer Interaction, HCII-HCII . 2025. Goeteborg, SWEDEN. Huang, S.S., et al. Visual Representation Learning through Causal Intervention for Controllable Image Editing . in 2025 Conference on Computer Vision and Pattern Recognition-CVPR-Annual . 2025. Nashville, TN. Ren, M.L., N.Y. Chen, and H. Qiu, Human-machine Collaborative Decision-making: An Evolutionary Roadmap Based on Cognitive Intelligence. International Journal of Social Robotics, 2023. 15 (7): p. 1101-1114. Loureiro, S.M.C., et al., Culture, heritage looting, and tourism: A text mining review approach. Frontiers in Psychology, 2022. 13 . Ghiurau, D. and D.E. Popescu, Distinguishing Reality from AI: Approaches for Detecting Synthetic Content. Computers, 2025. 14 (1). Farooq, A. and C. de Vreese, Deciphering authenticity in the age of AI: how AI-generated disinformation images and AI detection tools influence judgements of authenticity. Ai & Society, 2025. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8747381","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":588210046,"identity":"a4a6c94b-fd56-41df-a5e6-19a6e5a20513","order_by":0,"name":"junjun li","email":"","orcid":"","institution":"Hanyang University","correspondingAuthor":false,"prefix":"","firstName":"junjun","middleName":"","lastName":"li","suffix":""},{"id":588210047,"identity":"26ffc075-154a-4131-bd1d-931d604f2393","order_by":1,"name":"Sainan Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAzElEQVRIiWNgGAWjYDACCSBOYLBhYGA+AGSxEa8lDag6gRQtDAyHSdAiP7v52YMHNeft+dt4DBg+lB1m4J/dgF+LwZ1j5gYJx24zSxzjMWCcce4wg8SdAwS0SCSYSSSw3WZjuN9jwMzbdhgkQsBhM9K/SST8O8cjD7SF+S8xWhhu5JhJJLYdkDAAaWEkRovBjZwyicS+ZAPDY2wFB3vOpfNI3CDssG2SP77Z2csdY9744EeZtRz/DEIOQwYHgJiHBPWjYBSMglEwCnABAMD5PtV1Fr3eAAAAAElFTkSuQmCC","orcid":"","institution":"Zhejiang A\u0026F University","correspondingAuthor":true,"prefix":"","firstName":"Sainan","middleName":"","lastName":"Zhang","suffix":""},{"id":588210048,"identity":"286bb007-9135-4a1b-a373-793dd8dc2aa4","order_by":2,"name":"Jiajie Li","email":"","orcid":"","institution":"Hanyang University","correspondingAuthor":false,"prefix":"","firstName":"Jiajie","middleName":"","lastName":"Li","suffix":""},{"id":588210049,"identity":"09bafe2c-2dac-4932-9719-ccebd9c91c9d","order_by":3,"name":"Pengcheng Ju","email":"","orcid":"","institution":"Hanyang University","correspondingAuthor":false,"prefix":"","firstName":"Pengcheng","middleName":"","lastName":"Ju","suffix":""}],"badges":[],"createdAt":"2026-01-31 07:09:48","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8747381/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8747381/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":102337356,"identity":"85e9b423-2c65-47ef-8c99-6d87d9dee18e","added_by":"auto","created_at":"2026-02-10 16:12:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":158889,"visible":true,"origin":"","legend":"\u003cp\u003eThe overall research design process of this study\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8747381/v1/4030479065ada59641e51996.png"},{"id":102337352,"identity":"0039931a-667b-47c2-b198-a3d2ede92e46","added_by":"auto","created_at":"2026-02-10 16:12:26","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2927894,"visible":true,"origin":"","legend":"\u003cp\u003eDoor God--Qin Qiong Jingde\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8747381/v1/5b641de33c5f2d125592299b.png"},{"id":102337323,"identity":"b3f4d63b-99dc-43dd-9cd7-b357769d2227","added_by":"auto","created_at":"2026-02-10 16:12:17","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":29419326,"visible":true,"origin":"","legend":"\u003cp\u003eDesign Process and Derivative Outputs of Groups A and B\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8747381/v1/b13b3d5d5f656b115cc0f616.png"},{"id":104780248,"identity":"deddefd6-f12d-48aa-8f06-caf4beab4309","added_by":"auto","created_at":"2026-03-17 07:51:41","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":39563283,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8747381/v1/929a18e7-d964-41a2-9ec4-96e7f2938fdf.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"AI-Driven Innovation in Intangible Cultural Heritage:A Semiotic Analysis of Door-God Woodblock Prints Using Diffusion Models","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eIn 1987, the World Commission on Environment and Development proposed the concept of sustainable development [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Culture is essential to the sustainable development of modern society [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e]. In 2003, UNESCO launched the Convention for the Safeguarding of the Intangible Cultural Heritage, formally introducing the concept of \u0026ldquo;intangible cultural heritage\u0026rdquo;[\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. As an important form of China\u0026rsquo;s intangible cultural heritage, Yangjiabu woodblock New Year prints constitute an intergenerational visual-symbol system through their forms, colours, and narrative structures, playing a central role in maintaining cultural diversity and social memory [\u003cspan additionalcitationids=\"CR5\" citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. Among them, Door-God imagery combines ritual and customary functions with cultural symbolic meaning, serving as a key visual carrier of traditional narratives and folk beliefs [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. In today\u0026rsquo;s media environment, the readability of cultural-heritage symbols and the stability of their cultural semantics are insufficient [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. Younger audiences often struggle to accurately interpret the meanings of traditional cultural symbols, leading to difficulties in the transmission of such symbols and a consequent decline in cultural reproduction capacity [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. In addition, artificial design tends to be patterned in visual form, lacks innovative expression, and is difficult to adapt to new carriers such as digital media, which weakens the expression and contemporary value of cultural heritage symbols[\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. Against this background, how to overcome the limitations of traditional artificial design methods through the innovation of cultural heritage symbols has become an important issue for cultural inheritance and development [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eWith the rise of generative artificial intelligence, these technologies have attracted widespread social attention [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. Diffusion models demonstrate a high capacity to fit visual styles and structural characteristics, enabling the generation of diverse image variants while preserving the basic semantic outline [\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e, \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e]. Their application potential in cultural-heritage visualisation has therefore drawn increasing interest [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e]. Wang Shaofeng et al. examined the feasibility of generative design methods for Chinese woodblock New Year prints, outlined a blueprint for derivative works, and, within a methodological framework, explored three-dimensional design concepts to promote the dynamic safeguarding and wide dissemination of intangible cultural heritage[\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. To address the declining influence of traditional cultural symbols, Lin and other researchers investigated the application of traditional cultural symbols in art and design in the context of artificial intelligence[\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e]. Guo Mengyao et al. carried out innovation and preservation in the generative design of Yi ethnic embroidery patterns for cultural heritage based on LoRA [\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e]. Wang Yidan et al. developed a LoRA model embodying the blue clamp resist-dyeing technique, enabling the digital preservation of traditional patterns and their innovative reinterpretation [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e]. AI technologies have made notable progress in the preservation and reproduction of intangible cultural heritage images [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e], yet they still struggle to address the challenge of weakened cultural semantics in traditional imagery[\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e], as well as broader issues such as authenticity, subjectivity, and interpretive bias in cultural heritage [\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eAt present, the following research gaps remain: (1) the extent to which diffusion models can preserve the structural logic of traditional visual symbols; (2) the differences between AI-assisted design and manual design at the semiotic level (representamen\u0026ndash;object\u0026ndash;interpretant); and (3) how to quantitatively evaluate AI\u0026rsquo;s innovativeness in cultural symbols and its degree of cultural-semantic preservation. These gaps constrain the authenticity and practical usability of AI in the preservation, interpretation, and dissemination of cultural-heritage imagery.\u003c/p\u003e \u003cp\u003eTo explore the mechanism through which diffusion models operate in the semiotic reproduction of cultural-heritage symbols, this study introduces Peirce\u0026rsquo;s triadic semiotic framework (representamen\u0026ndash;object\u0026ndash;interpretant) to describe the sign chain linking visual form, cultural reference, and meaning interpretation [\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e]. This framework enables the construction of an interpretable analytical system for AI-generated cultural-heritage imagery, but it has not yet been sufficiently validated in diffusion-model image research.\u003c/p\u003e \u003cp\u003eBuilding on the limitations of existing research, this study makes the following contributions: (1) It proposes an analytical framework that integrates diffusion models with Peircean triadic semiotics, enabling a structured description of the generative mechanisms of traditional visual symbols and the interpretive biases that may arise in AI generation. (2) It develops a quantitative evaluation system combining AHP and FCE, translating symbolic innovativeness and cultural-semantic preservation into comparable and reusable metrics. (3) It designs a controlled comparative experimental procedure between manual design and AI-assisted design, systematically testing semiotic differences in symbol generation across the two design pathways and providing methodological support for the redesign of cultural-heritage symbols.\u003c/p\u003e"},{"header":"2. Related research","content":"\u003cp\u003eAs an element of China\u0026rsquo;s intangible cultural heritage, woodblock New Year prints have a long history and serve as an important vehicle of traditional visual art [\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. Characterised by concise yet exaggerated imagery, bright and auspicious colours, and romantic modes of expression, they function as cultural symbols that carry social norms, ethical values, and collective emotions [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e, \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. Door-God New Year prints display distinct semiotic features, such as paired figures, symmetrical compositions, highly saturated colours, and signifying elements including weapons, armour, and decorative patterns [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e]. These decomposable and quantifiable symbolic elements make Door-God prints an ideal subject for analysing the visual structure and semiotic logic of cultural heritage. With the development of society and the transformation of media, cultural heritage has been continuously impacted in terms of inheritance, production mechanisms, and social needs. Its living space in modern society has been compressed, and it faces the risk of disappearance[\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e]. In particular, New Year pictures, which are rooted in the lifestyle of traditional agrarian society, have been weakened in the process of social structural transformation and life scene change, and their original usage context and dissemination channels have been continuously weakened, making them more vulnerable[\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e].Therefore, the traditional symbol system of New Year pictures not only needs to be preserved, but also needs to be reactivated and contemporary translated through design language so that its symbolic meaning can be continued and regenerated in the contemporary communication context[\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eIn recent years, high-precision scanning, 3D modelling, and digital archiving systems have significantly improved the preservation quality and accessibility of traditional images, enabling the retention of surface features such as colour and texture and expanding the dissemination of intangible cultural heritage through online platforms [\u003cspan additionalcitationids=\"CR32\" citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e]. However, digital technologies primarily focus on recording visual appearance and lack explanatory capacity for deeper semiotic semantics such as character relationships, symbolic objects, and ritual logic [\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e]. Relying on digitisation alone is therefore insufficient to sustain the cultural continuity of the symbolic system, which creates both the necessity and research space for AI to participate in the preservation and enhancement of cultural heritage [\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e]. Diffusion models have demonstrated innovative capabilities in image generation, texture synthesis, and style transfer [\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e], and have gradually been applied to tasks such as restoration, inpainting, and stylistic reproduction of cultural-heritage imagery [\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e, \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e]. They can learn the artistic style of traditional images and, guided by textual or exemplar prompts, generate consistent outputs that reconstruct traditional New Year print imagery at the visual level [\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e]. Whether diffusion models can simultaneously preserve the structural logic of traditional symbols, their ritual references, and their cultural-semantic relationships during image generation still lacks systematic empirical verification [\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e].\u003c/p\u003e \u003cp\u003ePeirce\u0026rsquo;s triadic semiotic model (representamen\u0026ndash;object\u0026ndash;interpretant) provides a theoretical framework for analyzing the generation, reference, and interpretation of visual signs [\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e]. Sign meaning is not a static correspondence; rather, it is dynamically constructed through a \u0026ldquo;triadic semiosis\u0026rdquo; process jointly shaped by visual form, cultural reference, and audience interpretation [\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e]. The triadic semiotic model enables researchers to decompose complex images into operable sign elements, thereby identifying differences in semantic preservation and semantic drift across different generative mechanisms [\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e, \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e]. In cultural-heritage image research, Peircean semiotics can serve as an important tool for examining whether AI follows traditional semiotic logic. However, there remains a lack of studies that systematically apply this framework to diffusion-model image analysis, and systematic evidence is still insufficient regarding issues such as referential misplacement and the weakening of symbolic structures that may arise during AI-based sign generation.\u003c/p\u003e \u003cp\u003eWith the development of generative artificial intelligence, design activities have gradually shifted from one-way creation to a human\u0026ndash;AI co-creation model [\u003cspan additionalcitationids=\"CR48 CR49\" citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e]. Diffusion models have advantages in image generation and stylistic diversification, while designers remain irreplaceable in semantic judgment, contextual understanding, and value trade-offs [\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e]. For the visual systems of intangible cultural heritage\u0026mdash;where ritual norms and symbolic referential relationships are clearly defined\u0026mdash;symbol generation involves not only formal innovation but also relies heavily on accurate cultural reference and consistency in interpretation. In this context, human\u0026ndash;AI collaboration is no longer merely a strategy for improving design efficiency; it may also function as an important mechanism for mitigating the risks of semantic drift and symbol misplacement during AI generation. However, existing research on human\u0026ndash;AI collaboration has largely focused on general design or commercial creative tasks. For intangible cultural heritage, a highly semantically constrained symbolic system, the specific ways in which human\u0026ndash;AI collaboration operates at the level of sign generation still lack systematic empirical investigation.\u003c/p\u003e \u003cp\u003eIn summary, existing studies\u0026mdash;approaching the topic from perspectives such as digital preservation, generative artificial intelligence, semiotic analysis, and human\u0026ndash;AI collaboration\u0026mdash;have provided important theoretical and technical foundations for the reproduction and innovation of cultural-heritage imagery. However, much of this research focuses on the reproduction of visual style, evaluation of generative outcomes, or general collaboration mechanisms, while systematic examination remains insufficient regarding whether generative models follow traditional semiotic structures, ritual references, and cultural-semantic relationships at the level of cultural-heritage symbols. In particular, within intangible cultural-heritage visual systems that are highly symbolic and narrative in function, there is still a lack of empirical research that conducts semiotic-level comparative analyses between diffusion-model outputs and manual designs. There is also a lack of a quantitative methodological framework capable of simultaneously evaluating symbolic innovativeness and the degree of cultural-semantic preservation.\u003c/p\u003e \u003cp\u003eTherefore, it is necessary to introduce a research approach that balances interpretability of the symbol-generation mechanism with comparability of evaluation results, in order to systematically analyse the role and mechanism of AI-assisted design in the semiotic reproduction of intangible cultural heritage. This research provides the theoretical basis for the methodological choices and research design developed in this study.\u003c/p\u003e"},{"header":"3. Methods","content":"\u003cp\u003eThe study employs a parallel controlled mixed experimental design [\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e]. It examines the role of diffusion models in the reproduction of traditional New Year print symbols from two levels: cultural-heritage symbol production and symbol perception (see Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). Study 1 compares the generation pathways of traditional manual design and diffusion-model-assisted design, constructs a semiotic variable structure based on Peirce\u0026rsquo;s triadic semiotics, and combines the Analytic Hierarchy Process (AHP) and Fuzzy Comprehensive Evaluation (FCE) to quantitatively evaluate differences between the two design approaches across visual, cultural, and emotional dimensions. Study 2 employs a controlled behavioural experiment and questionnaire measures to test perceptual differences among general audiences in symbol decoding, cultural-semantic understanding, and emotional responses. Together, the two studies constitute an integrated analytical framework spanning generative mechanisms, expert judgment, and audience experience.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eStudy 1: Symbolic innovation mechanisms from expert and designer perspectives\u003c/p\u003e \u003cp\u003eIn order to ensure the statistical effectiveness of design samples, this study uses G*Power for pre-performance analysis[\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e] and recruits 8 design practitioners with different experience backgrounds (6 graduate students in the visual communication direction and 2 professional designers with more than 5 years of experience), all of whom have experience in traditional image creation or AIGC tools. Another 10 expert group, including 3 non-heritage research scholars and 7 design theory or practice experts, was set up to fill in the AHP judgement matrix and complete the FCE score.\u003c/p\u003e \u003cp\u003eThis study selected the most representative \u0026ldquo;Qin Qiong and Jing De\u0026rdquo; Door-God prints from the Yangjiabu woodblock New Year prints as the experimental materials (see Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). The imagery contains representative symbolic elements\u0026mdash;such as figure styling, costume patterns, and weapon types [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e, \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e54\u003c/span\u003e]\u0026mdash;which facilitates the extraction of sign elements and subsequent quantitative modelling [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTwo design pathways were established\u0026mdash;manual design (Group A) and AI-assisted design (Group B)\u0026mdash;to compare differences in symbol generation, semantic preservation, and formal innovation across different design mechanisms. Both groups were based on the same \u0026ldquo;Qin Qiong and Jing De\u0026rdquo; Door-God theme and used an identical system of sign elements to ensure consistent control conditions. The independent variable was the type of design pathway; the control variables were the cultural theme and symbolic features; and the dependent variables were the evaluation outcomes constructed from Peirce\u0026rsquo;s semiotic structure. The corresponding relationships are shown in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eExperimental Variables and Their Semiotic Logic\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSemiotic Level\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVariable Type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eVariable Content\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRepresentamen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIndependent Variable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDesign Pathway: Traditional vs. AI-assisted\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eObject\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eControlled Variable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eUnified Door God Theme (Qin Qiong \u0026amp; Yuchi Gong)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eInterpretant\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDependent Variable\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eAudience/Expert Evaluations(C1\u0026ndash;C3\u0026thinsp;+\u0026thinsp;Innovativeness)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eGroup A adopted a linear manual workflow centred on sketching, line drawing, and digital colouring, in which the designer reconstructed symbols based on their own cultural knowledge and modelling experience. Group B used the DALL\u0026middot;E 3 diffusion model to generate initial images; the designer controlled the theme and composition through prompts and then performed localised revisions and semantic calibration of the outputs. The two pathways respectively represent an experience-driven traditional creation process and a human\u0026ndash;AI collaborative mechanism for symbol generation. Their key differences and technical characteristics are summarised in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eExperimental Group Configuration\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGroup\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDesign Pathway Description\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eTechnical Approach \u0026amp; Features\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMode of Semiotic Reproduction\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGroup A: Traditional Manual Design\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDesigner-led manual creation following \u003cem\u003esketch \u0026rarr; line drawing \u0026rarr; final rendering\u003c/em\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eHand drawing\u0026thinsp;+\u0026thinsp;digital coloring; relies on designers\u0026rsquo; cultural knowledge and visual experience\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSubjective, experience-based re-encoding of traditional symbols\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGroup B: AI-Assisted Design\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDALL\u0026middot;E 3 diffusion model used to generate drafts; designers iteratively refine prompts and edit outputs editing AI-generated candidates.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrompt engineering\u0026thinsp;+\u0026thinsp;image editing; emphasizes AI\u0026rsquo;s efficiency in form generation and semantic variation diffusion, and form generation.\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHuman\u0026ndash;AI collaborative encoding with algorithmic mediation between prompt language and algorithmic output.\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo ensure comparability between the two pathways, the design procedures for Groups A and B were standardised: both groups were required to produce three final images, all developed around the same Door-God theme. In Group A, the designer generated and integrated the main symbolic elements. In Group B, the design was completed through a collaborative process of prompt-based control, diffusion-based generation, and manual revision. All design processes were documented for subsequent semiotic analysis, and the final images were used for expert review and audience experiments. The relevant workflow and output structure are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eBased on Peirce\u0026rsquo;s triadic semiotics, this study constructs a symbolic innovativeness evaluation structure consisting of three criterion layers (C1\u0026ndash;C3) and nine sub-indicators (S1\u0026ndash;S9) (see Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e), decomposing the visual form and cultural semantics of traditional New Year prints into quantifiable evaluation dimensions. Experts conducted pairwise comparisons of the indicators using a 1\u0026ndash;9 ratio scale to construct judgment matrices and calculate weights (see Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e). All matrices passed the consistency test. On this basis, FCE was used to build the membership matrices for the nine sub-indicators and, combined with the overall weights, to obtain innovativeness indices for the two design pathways [\u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e55\u003c/span\u003e]. The complete definitions of the evaluation sets, membership construction, and calculation procedures are provided in the Appendix.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAHP Hierarchical Structure for Symbolic Cognition Optimisation in Intangible Heritage\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"3\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGoal (G)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCriterion (C)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSub-Criterion (S)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"8\" rowspan=\"9\"\u003e \u003cp\u003eSymbolic Cognition Optimization in Intangible Heritage Design (G)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eC1 Visual Attraction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS1 Qualisign (sensory attributes)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS2 Sinsign (stylistic coherence)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS3 Legisign (compositional structure)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eC2 Cultural Identity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS4 Iconic Object (heroic prototype)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS5 Indexical Object (ritual reference)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS6 Symbolic Object (cultural convention)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eC3 Emotional Resonance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS7 Rheme (emotional impression)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS8 Dicent (functional cognition)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eS9 Argument (value and belief)\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComprehensive Weight Results of Design Elements (AHP Final Output)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGoal (G)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eCriterion (C)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePrimary Weight\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSub-Criterion (S)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSecondary Weight\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eComposite Weight (Wi)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eRank\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"8\" rowspan=\"9\"\u003e \u003cp\u003eSymbolic Cognition Optimization in Intangible Heritage Design\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eC1 Visual Attraction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e0.4013\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS1 Qualisign\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.4344\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.1743\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS2 Sinsign\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.2432\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.0976\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS3 Legisign\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.3224\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.1294\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eC2 Cultural Identity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e0.3361\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS4 Iconic Object\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.3141\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.1056\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS5 Indexical Object\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.4711\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.1583\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS6 Symbolic Object\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.2148\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.0722\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eC3 Emotional Resonance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003e0.2626\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS7 Rheme\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.3428\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.0900\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS8 Dicent\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.3660\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.0961\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eS9 Argument\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.2912\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.0765\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eStudy 2: Audience perception experiment\u003c/p\u003e \u003cp\u003eStudy 2 aims to examine whether Groups A and B from Study 1 differ in symbol readability, cultural-semantic understanding, and emotional responses among general audiences. This experiment employs a controlled behavioural experiment and questionnaire measures to compare perceptual differences between the two image sets across three levels: visual representation (Representamen), cultural semantics (Object), and emotional interpretation (Interpretant).\u003c/p\u003e \u003cp\u003eA total of 114 general participants were recruited for this study. Among them, 56.14% were female, and 43.86% were male; 42.98% were from design-related majors, and 57.02% were from non-design majors. All participants completed the questionnaire anonymously and provided informed consent, and the research procedure complied with ethical guidelines.\u003c/p\u003e \u003cp\u003eThe experimental stimuli were drawn from the final design outputs of Study 1. Each group included three images, and all images were presented in a randomised order to reduce order effects.\u003c/p\u003e \u003cp\u003eThis study set up three types of evaluation tasks to cover the form, reference, and meaning structure of the images. The representamen task assessed the clarity and coherence of color, form, and composition. The object task assessed the balance of character identity, cultural meaning, and cultural conventions. The interpretant task assessed emotional impression, degree of meaning understanding, and cultural pride. All items used a five-point Likert scale to rate the two image sets separately, producing paired data.\u003c/p\u003e"},{"header":"4. Results","content":"\u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Study 1: Generation Outcomes of the Two Design Pathways\u003c/h2\u003e \u003cp\u003eGroup A presented a manual design approach using sketches, line drawings, and digital coloring. The overall visual style was stable, and the cultural symbol structure remained intact. Group B, on the other hand, employed a collaborative model based on diffusion model image generation and manual correction. While exhibiting significantly stronger visual novelty, it showed some instability in certain details. The two design models demonstrated differences in stability and novelty regarding their symbol generation mechanisms.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Study 1: Results of the Three-Level Semiotic Analysis\u003c/h2\u003e \u003cp\u003eAt the Representame level of semiotics, Group A maintains the visual elements of traditional New Year paintings relatively well; Group B exhibits greater variation in visual style but is weaker in expressing proportions and details in local image areas. At the Object level, Group A accurately presents figures, weapons, and ritualistic connotations; while Group B maintains thematic consistency, it is prone to issues such as mixed use of weapons and misplaced patterns. At the Interpretant level, Group A coherently presents cultural semantics; Group B excels in visual novelty but is slightly weaker in emotional depth and symbolic meaning.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e4.3 AHP Weighting Model Results\u003c/h2\u003e \u003cp\u003eAll judgment matrices passed the consistency test (CR\u0026thinsp;\u0026lt;\u0026thinsp;0.10). The weights of the first-level criteria indicate that visual appeal was the highest (C1\u0026thinsp;=\u0026thinsp;0.4013), followed by cultural identity (C2\u0026thinsp;=\u0026thinsp;0.3361) and emotional resonance (C3\u0026thinsp;=\u0026thinsp;0.2626). Among the sub-indicators, qualisign (S1), indexical object (S5), and legisign (S3) ranked in the top three, suggesting that visual presentation, traditional belief references, and compositional style are key factors influencing innovativeness.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e4.4 FCE Results\u003c/h2\u003e \u003cp\u003eThe overall innovativeness scores were 92.38 for Group A and 91.94 for Group B. Both groups reached a high level, and the difference was minimal, indicating that diffusion models have achieved a level comparable to manual design in terms of formal innovation and style generation, while their cultural-semantic consistency remains slightly weaker than that of manual design.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e4.5 Study 2: Results of the General-Audience Experiment\u003c/h2\u003e \u003cp\u003eData analysis employed descriptive statistics and paired-samples t-tests to compare the differences between the two groups of images across various symbol dimensions. The scale demonstrated good reliability and validity (Cronbach\u0026rsquo;s α\u0026thinsp;=\u0026thinsp;0.967, KMO\u0026thinsp;=\u0026thinsp;0.908, Bartlett p \u0026lt; .001).\u003c/p\u003e \u003cp\u003eScores across the three semiotic dimensions (C1\u0026ndash;C3) were all in a relatively high range (4.0\u0026ndash;4.3), and the differences between Groups A and B in overall visual performance and cultural readability were small (see Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). In the subsequent paired-samples t-tests, most indicators showed no significant differences (see Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Among them, A3 (composition clarity) was close to significance, while A6 (balance of cultural conventions) showed a significant difference (p = .034), indicating that participants tended to favour manual design in judgments related to cultural semantics, whereas the two groups performed largely similarly at the visual level.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDescriptive Statistics of the Three Semiotic Dimensions (C1\u0026ndash;C3) in Study 2\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDimension\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGroup A Mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eGroup A SD\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGroup B Mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eGroup B SD\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVisual appeal (C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.15\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eCultural identity (C2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.94\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eEmotional resonance (C3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e4.12\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.96\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.99\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePaired-Samples t-Test Results (Group A vs. Group B)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"8\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDimension\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIndicator Number\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eIndicator Content\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eGroup A Mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eGroup B Mean\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003et value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003ep value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003eSignificance\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eVisual appeal (C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eNovelty of color and shape\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.175\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.088\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.043\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.299\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eStyle coordination\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.333\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.202\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.565\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.120\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCompositional clarity\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.298\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.149\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.706\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.091\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eMarginal significance\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eCultural identity (C2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCharacter recognition\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.360\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.237\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e1.221\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.225\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCultural connotation expression\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.263\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.211\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.551\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.583\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCultural norms balance\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.325\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.158\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2.144\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.034\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003eSignificant\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\" morerows=\"2\" rowspan=\"3\"\u003e \u003cp\u003eEmotional resonance (C3)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFirst impression and emotional reaction\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e3.930\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.009\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.904\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.368\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeaning comprehension\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.167\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.184\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.187\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.852\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eA9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eCultural pride\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e4.254\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4.228\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.274\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.785\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e\u0026mdash;\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e4.6 Summary\u003c/h2\u003e \u003cp\u003eSynthesising the findings from both studies, Group A performed better in cultural-semantic stability and semiotic consistency, whereas Group B showed advantages in image generation and stylistic divergence. The AHP\u0026ndash;FCE composite innovativeness index indicated that the two groups achieved similar scores (92.38 vs. 91.94), suggesting that AI\u0026rsquo;s capacity for visual innovation has approached that of manual creation. The audience experiment further showed no significant difference between the two groups in overall visual performance; however, regarding the balance between traditional and modern semantics, participants\u0026rsquo; evaluations tended to favour manual design.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eThis study examines the expressive power of the diffusion model in the redesign of New Year pictures from two aspects: the generation mechanism of cultural heritage symbols and the audience perception results. The results of Study 1 and Study 2 show that the diffusion model has significant advantages in the generation of visual forms, but has certain limitations in terms of cultural semantic consistency and the stability of symbolic meaning interpretation. This phenomenon of enhanced visual form innovation but weakened cultural semantic stability echoes the relevant discussions in the field of heritage science on the quality of visual representation and cultural adaptability [\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFrom a semiotic perspective, the image generated by the diffusion model based on the AHP weighting analysis results achieved the highest weight in terms of visual appeal, and the audience also gave positive feedback on the visual novelty of the generated image. These results suggest that diffusion models have strong generative capability at the level of the representamen, effectively learning and recombining the colour styles, compositional features, and formal language of traditional New Year prints. This characteristic aligns with prior studies arguing that diffusion models tend to prioritise learning visual representations and, due to a lack of inherent constraints regarding semantic meaning and semiotic rules, are prone to producing spurious correlations [\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e]. However, both the FCE innovativeness evaluation and the audience symbol-decoding tasks showed that diffusion-model-generated images still exhibited varying degrees of semantic drift from New Year prints, including weakened logical coherence in local details, misplacement of symbolic objects, and instability in figure-identity reference. This indicates that while diffusion models can reproduce traditional visual features, they have not yet formed a stable structure of cultural conventions to support meaning construction at the object and interpretant levels.\u003c/p\u003e \u003cp\u003eFurther comparison between the manual design and AI-assisted design pathways shows that human\u0026ndash;AI collaboration produces a clear complementary effect in the reproduction of traditional symbols. Diffusion models excel in exploring visual forms and diversifying styles, whereas manual design demonstrates greater stability in object reference, recognition of ritual conventions, and grasp of narrative value. In particular, with respect to the expression of Door-God figure identity, the combination of symbolic objects and their cultural implications, manual design performed significantly better than diffusion-model outputs at the interpretant level. This finding supports prior research emphasising the crucial role of designers in cognitive decision-making and value trade-offs within human\u0026ndash;AI collaboration [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e]. The results suggest that, in intangible cultural heritage visual systems characterised by strong ritual norms and symbolic functions, human\u0026ndash;AI collaboration can help mitigate the risk of semantic drift during generative-model-based symbol production.\u003c/p\u003e \u003cp\u003eFrom the perspective of symbolic sustainability, the findings of this study can be further interpreted. The diverse visual forms generated by diffusion models can facilitate the re-dissemination of cultural-heritage symbols across media contexts and, to some extent, increase young audiences\u0026rsquo; visual interest, thereby alleviating the decline in symbol readability in contemporary communication environments. However, Study 2 revealed significant differences in its cultural normative balance indicators. When the evaluation involved cultural significance, customary rationality, and symbolic consistency, the audience tended to favor artificially designed images. This aligns with the argument in cultural heritage research that digital transformation should not only focus on formal innovation and protection but also on sustainability at the social, environmental, and economic levels [\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e, \u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e]. This suggests that AI-generated images are suitable as a medium for visual communication and formal exploration in the field of cultural heritage, rather than as the subject of the symbolic meaning of cultural heritage.\u003c/p\u003e \u003cp\u003eThe quantitative results of AHP\u0026ndash;FCE reveal the difference between the innovativeness and semantic stability of cultural heritage symbols. The highest-weighted prime signifier (S1), indicator object (S5), and type signifier (S3) are key elements in maintaining the visual form and cultural semantics of traditional New Year paintings. Theoretically, diffusion models can learn the visual form rules of symbols through large-scale data learning. However, in traditional cultural systems that are highly dependent on context, customs, and collective memory, the cultural semantics of symbols are still profoundly influenced by traditional customs and social contracts. This has important implications for the research on the digitization of cultural heritage and cultural sustainability. The introduction of generative artificial intelligence requires a corresponding semantic supervision mechanism and expert knowledge calibration to avoid semantic deviation in the reproduction of cultural information. This finding is consistent with the relevant discussions in the current research on the authenticity of AI-generated content [\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e, \u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eFrom the methodological perspective of heritage science, the \"diffusion model \u0026times; semiotics \u0026times; quantitative evaluation\" framework proposed in this study analyzes cultural heritage semiotically and transforms the degree of semantic shift of symbols into quantifiable indicators. This allows conservation agencies and design teams to more accurately digitally reorganize the symbolic elements of cultural heritage, thereby establishing an adjustable balance mechanism between visual innovation and cultural semantic protection.\u003c/p\u003e"},{"header":"6. Conclusion","content":"\u003cp\u003eThis study takes Yangjiabu door god New Year paintings as a case study to investigate the innovative expression of generative artificial intelligence in the redesign of visual symbols of intangible cultural heritage. The results show that the diffusion model can effectively expand visual expression forms in the redesign of visual symbols of cultural heritage, but it is weak in terms of cultural reference and semantic interpretation stability, thus clarifying the positioning of generative artificial intelligence in the context of cultural heritage.\u003c/p\u003e \u003cp\u003eBuilding upon this foundation, this paper proposes and validates an analytical framework integrating a diffusion model, Peircean's triadic semiotics, and a quantitative evaluation method. This framework is used to analyze the performance characteristics and potential risks of the diffusion model in the redesign of cultural heritage symbols. It provides a structured tool for distinguishing different levels of symbols, such as visual representation, cultural reference, and meaning interpretation, and offers a reusable methodology for related research.\u003c/p\u003e \u003cp\u003eThis study does not advocate replacing human design with artificial intelligence, but rather emphasizes the role of human-machine collaboration in designer-led design processes. Through explicit semantic oversight and symbol calibration, generative AI can support visual innovation while avoiding cultural semantic shifts.\u003c/p\u003e \u003cp\u003eOverall, this study provides methodological and practical insights for the participation of generative artificial intelligence in the visual redesign of intangible cultural heritage, and lays an analytical foundation for discussions on the sustainability of cultural heritage symbols in the age of artificial intelligence. Future research can further test the applicability of this framework in different heritage types and cultural contexts, and explore more refined human-machine collaboration mechanisms to promote a long-term balance between innovative expression and semantic preservation in the digital practice of cultural heritage.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eEthics Approval and Consent to Participate \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study was conducted in accordance with the academic research ethics guidelines of the authors\u0026rsquo; affiliated institutions. Given that the research did not involve medical experiments or high-risk interventions with human participants, formal approval from an Institutional Review Board (IRB) was not required.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eInformed Consent \u0026nbsp;\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAll human participants involved in this study were fully informed about the research objectives and procedures prior to participation and voluntarily provided their informed consent. The study ensured participant anonymity and data security throughout the research process.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis work was supported by the Talent Start-up Project of the Scientific Research and Development Fund of Zhejiang A\u0026amp;F University (Grant No. 2024FR043).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability \u0026nbsp; \u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data generated or analyzed during the current study are available from the corresponding author upon reasonable request.\u003c/p\u003e\n\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\n\u003cp\u003eConceptualization, Junjun Li and Sainan Zhang; Methodology, Junjun Li; Formal Analysis, Jiajie Li; Investigation, Pengcheng Ju; Writing—original draft, Junjun Li; Writing—review and editing, Sainan Zhang; Supervision, Sainan Zhang; Funding acquisition, Sainan Zhang.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eEnvironment, W.C.o., \u003cem\u003eOur common future\u003c/em\u003e. 1987: Peterson\u0026apos;s.\u003c/li\u003e\n\u003cli\u003eLisitza, A. and G. Wolbring, \u003cem\u003eSustainability within the academic EcoHealth literature: Existing engagement and future prospects.\u003c/em\u003e Sustainability, 2016. \u003cstrong\u003e8\u003c/strong\u003e(3): p. 202.\u003c/li\u003e\n\u003cli\u003eTeke, S.G., \u003cem\u003eLiving Heritages and Fixed Traditions in the Lists of the Convention for the Safeguarding of Intangible Cultural Heritage.\u003c/em\u003e Milli Folklor, 2018(120): p. 19-31.\u003c/li\u003e\n\u003cli\u003eShen, H., \u003cem\u003eClassic Chinese New Year Paintings\u003c/em\u003e. 2015, Haitian Publishing House: Shenzhen, China.\u003c/li\u003e\n\u003cli\u003eXu, Y.W., Y. Tao, and B. Smith, \u003cem\u003eChina\u0026apos;s emerging legislative and policy framework for safeguarding intangible cultural heritage.\u003c/em\u003e International Journal of Cultural Policy, 2022. \u003cstrong\u003e28\u003c/strong\u003e(5): p. 566-580.\u003c/li\u003e\n\u003cli\u003eLiang, J.Y., R.T. Guo, and Y.Q. He, \u003cem\u003eA Psychological Analysis of the Imagery of Chinese Menshen.\u003c/em\u003e Culture \u0026amp; Psychology, 2022. \u003cstrong\u003e28\u003c/strong\u003e(4): p. 527-549.\u003c/li\u003e\n\u003cli\u003eBaojie, J., \u003cem\u003eExploring the Integration of Chinese Traditional Elements in Visual Communication and Its Influence on Cultural Identity and Public Perception: A review.\u003c/em\u003e Peta International Journal of Social Science and Humanity, 2025. \u003cstrong\u003e4\u003c/strong\u003e(3): p. 1-28.\u003c/li\u003e\n\u003cli\u003eCao, H. and S. Champadaeng, \u003cem\u003eThe Art of Chinese Calligraphy: Educational Protection and Literacy Study of Cultural Heritage.\u003c/em\u003e International Journal of Education and Literacy Studies, 2024. \u003cstrong\u003e12\u003c/strong\u003e(3): p. 160-171.\u003c/li\u003e\n\u003cli\u003eCai, Z., et al., \u003cem\u003eSpatial distribution characteristics and sustainable inheritance strategies of national traditional fine arts intangible cultural heritage in China.\u003c/em\u003e Sustainability, 2024. \u003cstrong\u003e16\u003c/strong\u003e(11): p. 4488.\u003c/li\u003e\n\u003cli\u003eShen, Y., Y.-c. Cai, and X.-y. Wang, \u003cem\u003eIntangible cultural heritage in industrial design.\u003c/em\u003e Digital Scholarship in the Humanities, 2024. \u003cstrong\u003e39\u003c/strong\u003e(1): p. 354-372.\u003c/li\u003e\n\u003cli\u003eXie, J., \u003cem\u003eInnovative design of artificial intelligence in intangible cultural heritage.\u003c/em\u003e Scientific Programming, 2022. \u003cstrong\u003e2022\u003c/strong\u003e(1): p. 6913046.\u003c/li\u003e\n\u003cli\u003eCao, Y., et al., \u003cem\u003eA survey of ai-generated content (aigc).\u003c/em\u003e ACM Computing Surveys, 2025. \u003cstrong\u003e57\u003c/strong\u003e(5): p. 1-38.\u003c/li\u003e\n\u003cli\u003eOyedokun, G.E. \u003cem\u003eCultural Heritage and Intellectual Property Rights in the Digital Age: A UNESCO 2005 Convention Perspective\u003c/em\u003e. in \u003cem\u003eCultural Odyssey: 20 Years of Implementation of UNESCO\u0026apos;s 2005 Convention in Nigeria\u003c/em\u003e. 2025.\u003c/li\u003e\n\u003cli\u003eZhou, Y., et al., \u003cem\u003eFine-tuning diffusion model to generate new kite designs for the revitalization and innovation of intangible cultural heritage.\u003c/em\u003e Scientific Reports, 2025. \u003cstrong\u003e15\u003c/strong\u003e(1): p. 7519.\u003c/li\u003e\n\u003cli\u003eWang, S.F. and C.C. Chen, \u003cem\u003eRevitalizing Intangible cultural heritage via derivative design: A focus on chinese woodblock printing.\u003c/em\u003e Plos One, 2025. \u003cstrong\u003e20\u003c/strong\u003e(4).\u003c/li\u003e\n\u003cli\u003eLin, C., \u003cem\u003eApplication of traditional cultural symbols in art design under the background of artificial intelligence.\u003c/em\u003e Mathematical Problems in Engineering, 2021. \u003cstrong\u003e2021\u003c/strong\u003e(1): p. 1258080.\u003c/li\u003e\n\u003cli\u003eGuo, M.Y., et al. \u003cem\u003eLoRA-Based Pattern Generation for Yi Ethnic Embroidery Heritage Preservation\u003c/em\u003e. in \u003cem\u003e17th Conference on Creativity and Cognition-C\u0026amp;C\u003c/em\u003e. 2025. null, ELECTR NETWORK.\u003c/li\u003e\n\u003cli\u003eWang, Y.D. and Y.X. Zhou, \u003cem\u003eArtificial Intelligence-Driven Interactive Experience for Intangible Cultural Heritage: Sustainable Innovation of Blue Clamp-Resist Dyeing.\u003c/em\u003e Sustainability, 2025. \u003cstrong\u003e17\u003c/strong\u003e(3).\u003c/li\u003e\n\u003cli\u003eLiu, Y., \u003cem\u003eApplication of Digital Technology in Intangible Cultural Heritage Protection.\u003c/em\u003e Mobile Information Systems, 2022. \u003cstrong\u003e2022\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eSiliutina, I., et al., \u003cem\u003eCultural preservation and digital heritage: challenges and opportunities.\u003c/em\u003e Amazonia Investiga, 2024. \u003cstrong\u003e13\u003c/strong\u003e(75): p. 262-273.\u003c/li\u003e\n\u003cli\u003eKhan, A.S., \u003cem\u003eThe preservation of cultural heritage: challenges and opportunities in the digital age.\u003c/em\u003e Al-Behishat Research Archive, 2023. \u003cstrong\u003e1\u003c/strong\u003e(01): p. 16-24.\u003c/li\u003e\n\u003cli\u003eTiribelli, S., et al., \u003cem\u003eEthics of Artificial Intelligence for Cultural Heritage: Opportunities and Challenges.\u003c/em\u003e Ieee Transactions on Technology and Society, 2024. \u003cstrong\u003e5\u003c/strong\u003e(3): p. 293-305.\u003c/li\u003e\n\u003cli\u003eFredin-Ortiz, E., \u003cem\u003eCharles S. Peirce\u0026apos;s Semiotics and the Concept of Mental Representation in Cognitive Science.\u003c/em\u003e Designis, 2025(43): p. 271-279.\u003c/li\u003e\n\u003cli\u003eNie, H. and M.Y. Ran. \u003cem\u003eStudy on the Woodblock New Year Pictures of Zhuxianzhen Town\u003c/em\u003e. in \u003cem\u003e2nd International Conference on Art Studies: Science, Experience, Education (ICASSEE)\u003c/em\u003e. 2018. Moscow, RUSSIA.\u003c/li\u003e\n\u003cli\u003eKryzhko, O., \u003cem\u003eNational and Cultural Symbolism of Zoo-Anthropomorphic Creatures in Ukrainian Myphological Picture of the World.\u003c/em\u003e LOGOS-A Journal of Religion, Philosophy, Comparative Cultural Studies and Art, 2019(101): p. 159-168.\u003c/li\u003e\n\u003cli\u003eGao, Y., Y. Zhang, and I. Destech Publicat. \u003cem\u003eThe Fusion of Traditional New Year Pictures and Modern Decoration Development-The Case of Hebei Wuqiang New Year Pictures\u003c/em\u003e. in \u003cem\u003eInternational Conference on Education and Educational Research and Environmental Studies (EERES)\u003c/em\u003e. 2016. Hong Kong, PEOPLES R CHINA.\u003c/li\u003e\n\u003cli\u003eSu, T.C., et al., \u003cem\u003eStyle Recognition of Door God Paintings by Hypothesis Testing for Texture Features of Painting Patterns.\u003c/em\u003e Applied Sciences-Basel, 2022. \u003cstrong\u003e12\u003c/strong\u003e(5).\u003c/li\u003e\n\u003cli\u003eZhang, L., et al., \u003cem\u003eMastersu: the sustainable development of su embroidery based on digital technology.\u003c/em\u003e Sustainability, 2022. \u003cstrong\u003e14\u003c/strong\u003e(12): p. 7094.\u003c/li\u003e\n\u003cli\u003eZhang, B., et al., \u003cem\u003eCan AI-generated art stimulate the sustainability of intangible cultural heritage? A quantitative research on cultural and creative products of New Year Prints generated by AI.\u003c/em\u003e Heliyon, 2023. \u003cstrong\u003e9\u003c/strong\u003e(10).\u003c/li\u003e\n\u003cli\u003eQue, Y.X., T. de la Hera, and J. Jansz, \u003cem\u003eDive Into the Past: Chinese Intangible Cultural Heritage Represented in Chinese Mobile Games.\u003c/em\u003e Media and Communication, 2025. \u003cstrong\u003e13\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eTrivi\u0026ntilde;o-Tarradas, P., D.F. Garc\u0026iacute;a-Molina, and J.I. Rojas-Sola, \u003cem\u003eImpact of 3D digitising technologies and their implementation.\u003c/em\u003e Technologies, 2024. \u003cstrong\u003e12\u003c/strong\u003e(12): p. 260.\u003c/li\u003e\n\u003cli\u003eBanfi, F., et al., \u003cem\u003eNarrating Ancient Roman Heritage through Drawings and Digital Architectural Representation: From Historical Archives, UAV and LIDAR to Virtual-Visual Storytelling and HBIM Projects.\u003c/em\u003e Drones, 2023. \u003cstrong\u003e7\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eZhang, C.Y., et al., \u003cem\u003eIntegrated high-precision real scene 3D modeling of karst cave landscape based on laser scanning and photogrammetry.\u003c/em\u003e Scientific Reports, 2024. \u003cstrong\u003e14\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eJiang, Y., et al., \u003cem\u003eNational representations in digital heritage exposure: exploring young people\u0026rsquo;s in-group/out-group categorisation on cultural artefact visuals and related evaluations.\u003c/em\u003e Humanities and Social Sciences Communications, 2025. \u003cstrong\u003e12\u003c/strong\u003e(1): p. 1743.\u003c/li\u003e\n\u003cli\u003eColace, F., et al., \u003cem\u003eNew AI challenges for cultural heritage protection: A general overview.\u003c/em\u003e Journal of Cultural Heritage, 2025. \u003cstrong\u003e75\u003c/strong\u003e: p. 168-193.\u003c/li\u003e\n\u003cli\u003eCao, T., et al. \u003cem\u003eTexfusion: Synthesizing 3d textures with text-guided image diffusion models\u003c/em\u003e. in \u003cem\u003eProceedings of the IEEE/CVF international conference on computer vision\u003c/em\u003e. 2023.\u003c/li\u003e\n\u003cli\u003eChen, H., et al., \u003cem\u003eComprehensive exploration of diffusion models in image generation: a survey.\u003c/em\u003e Artificial Intelligence Review, 2025. \u003cstrong\u003e58\u003c/strong\u003e(4): p. 99.\u003c/li\u003e\n\u003cli\u003eHsieh, K., et al. \u003cem\u003eCultural Heritage Meets AI: Advanced Text-to-Image Models for Digital Reconstruction and Preservation\u003c/em\u003e. in \u003cem\u003e2024 6th International Conference on Control and Robotics (ICCR)\u003c/em\u003e. 2024. IEEE.\u003c/li\u003e\n\u003cli\u003eYu, T., et al., \u003cem\u003eArtificial intelligence for Dunhuang cultural heritage protection: the project and the dataset.\u003c/em\u003e International Journal of Computer Vision, 2022. \u003cstrong\u003e130\u003c/strong\u003e(11): p. 2646-2673.\u003c/li\u003e\n\u003cli\u003eZhang, C., et al., \u003cem\u003eText-to-image diffusion models in generative ai: A survey.\u003c/em\u003e arXiv preprint arXiv:2303.07909, 2023.\u003c/li\u003e\n\u003cli\u003eSantoro, A., et al., \u003cem\u003eSymbolic behaviour in artificial intelligence.\u003c/em\u003e arXiv preprint arXiv:2102.03406, 2021.\u003c/li\u003e\n\u003cli\u003eOrtiz, E.F., \u003cem\u003eLa semi\u0026oacute;tica de Charles S. Peirce y el concepto de representaci\u0026oacute;n mental en la ciencia cognitiva.\u003c/em\u003e DeSignis: Publicaci\u0026oacute;n de la Federaci\u0026oacute;n Latinoamericana de Semi\u0026oacute;tica (FELS), 2025(43): p. 271-279.\u003c/li\u003e\n\u003cli\u003eLorino, P., \u003cem\u003eCharles Sanders Peirce (1839\u0026ndash;1914).\u003c/em\u003e Oxford handbook of process philosophy and organization studies, 2014: p. 143-165.\u003c/li\u003e\n\u003cli\u003ePeirce, C.S., \u003cem\u003ePeirce on signs.\u003c/em\u003e 1991.\u003c/li\u003e\n\u003cli\u003eYunhee Lee, \u003cem\u003eNarrative cognition and modeling in new media communication from Peirce\u0026apos;s semiotic perspective.\u003c/em\u003e Semiotica, 2012. \u003cstrong\u003e2012\u003c/strong\u003e(189): p. 181-195.\u003c/li\u003e\n\u003cli\u003eLuo, Q., \u003cem\u003eUnderstanding artworks from Danto\u0026rsquo;s philosophy of art: a Peircean semiotic approach.\u003c/em\u003e Chinese Semiotic Studies, 2023. \u003cstrong\u003e19\u003c/strong\u003e(4): p. 665-685.\u003c/li\u003e\n\u003cli\u003eFui-Hoon Nah, F., et al., \u003cem\u003eGenerative AI and ChatGPT: Applications, challenges, and AI-human collaboration\u003c/em\u003e. 2023, Taylor \u0026amp; Francis. p. 277-304.\u003c/li\u003e\n\u003cli\u003ePuerta-Beldarrain, M., et al., \u003cem\u003eA multifaceted vision of the Human-AI collaboration: a comprehensive review.\u003c/em\u003e IEEE Access, 2025.\u003c/li\u003e\n\u003cli\u003eNapole\u0026atilde;o, E. and A.D. Silva, \u003cem\u003eDesign Practice with Generative Artificial Intelligence for Sustainability: An Experimental Study with Undergraduate Designers\u003c/em\u003e, in \u003cem\u003eArtificial Intelligence-Aided Design for Sustainability\u003c/em\u003e. 2025, Springer. p. 75-93.\u003c/li\u003e\n\u003cli\u003eLi, W., et al., \u003cem\u003eExploring human-machine collaboration paths in the context of AI-generation content creation: a case study in product styling design.\u003c/em\u003e Journal of Engineering Design, 2025. \u003cstrong\u003e36\u003c/strong\u003e(2): p. 298-324.\u003c/li\u003e\n\u003cli\u003eSaldana Ochoa, K., \u003cem\u003eCan Artificial Intelligence Mark the Next Architectural Revolution? Design Exploration in the Realm of Generative Algorithms and Search Engines\u003c/em\u003e, in \u003cem\u003eDecoding Cultural Heritage: A Critical Dissection and Taxonomy of Human Creativity through Digital Tools\u003c/em\u003e. 2024, Springer. p. 3-22.\u003c/li\u003e\n\u003cli\u003eSchoonenboom, J. and R.B. Johnson, \u003cem\u003eHow to construct a mixed methods research design.\u003c/em\u003e Kolner Zeitschrift fur Soziologie und Sozialpsychologie, 2017. \u003cstrong\u003e69\u003c/strong\u003e(Suppl 2): p. 107.\u003c/li\u003e\n\u003cli\u003eFaul, F., et al., \u003cem\u003eStatistical power analyses using G* Power 3.1: Tests for correlation and regression analyses.\u003c/em\u003e Behavior research methods, 2009. \u003cstrong\u003e41\u003c/strong\u003e(4): p. 1149-1160.\u003c/li\u003e\n\u003cli\u003ePeng, X. \u003cem\u003eStudy on the Application and Research of Visual Symbols of Traditional New Year Pictures in Modern Design\u003c/em\u003e. in \u003cem\u003eThe 6th International Conference on Arts, Design and Contemporary Education (ICADCE 2020)\u003c/em\u003e. 2021. Atlantis Press.\u003c/li\u003e\n\u003cli\u003eHo, W. and X. Ma, \u003cem\u003eThe state-of-the-art integrations and applications of the analytic hierarchy process.\u003c/em\u003e European Journal of Operational Research, 2018. \u003cstrong\u003e267\u003c/strong\u003e(2): p. 399-414.\u003c/li\u003e\n\u003cli\u003eLin, Y.H., H.L. Liu, and N. Yang. \u003cem\u003eThe Application Strategies of AIGC Technology Empowering Chinese Traditional Cultural Elements in Graphic Design\u003c/em\u003e. in \u003cem\u003e27th International Conference on Human-Computer Interaction, HCII-HCII\u003c/em\u003e. 2025. Goeteborg, SWEDEN.\u003c/li\u003e\n\u003cli\u003eHuang, S.S., et al. \u003cem\u003eVisual Representation Learning through Causal Intervention for Controllable Image Editing\u003c/em\u003e. in \u003cem\u003e2025 Conference on Computer Vision and Pattern Recognition-CVPR-Annual\u003c/em\u003e. 2025. Nashville, TN.\u003c/li\u003e\n\u003cli\u003eRen, M.L., N.Y. Chen, and H. Qiu, \u003cem\u003eHuman-machine Collaborative Decision-making: An Evolutionary Roadmap Based on Cognitive Intelligence.\u003c/em\u003e International Journal of Social Robotics, 2023. \u003cstrong\u003e15\u003c/strong\u003e(7): p. 1101-1114.\u003c/li\u003e\n\u003cli\u003eLoureiro, S.M.C., et al., \u003cem\u003eCulture, heritage looting, and tourism: A text mining review approach.\u003c/em\u003e Frontiers in Psychology, 2022. \u003cstrong\u003e13\u003c/strong\u003e.\u003c/li\u003e\n\u003cli\u003eGhiurau, D. and D.E. Popescu, \u003cem\u003eDistinguishing Reality from AI: Approaches for Detecting Synthetic Content.\u003c/em\u003e Computers, 2025. \u003cstrong\u003e14\u003c/strong\u003e(1).\u003c/li\u003e\n\u003cli\u003eFarooq, A. and C. de Vreese, \u003cem\u003eDeciphering authenticity in the age of AI: how AI-generated disinformation images and AI detection tools influence judgements of authenticity.\u003c/em\u003e Ai \u0026amp; Society, 2025.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Intangible Cultural Heritage, Diffusion Models, AI-Assisted Design, Peircean Semiotics, Symbolic Reconstruction, Door-God Woodblock Prints","lastPublishedDoi":"10.21203/rs.3.rs-8747381/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8747381/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eAs a form of China\u0026rsquo;s intangible cultural heritage, woodblock New Year prints face challenges, including insufficient innovative expression and a weakening resonance with contemporary audiences. Using the \u0026ldquo;Qin Qiong and Jing De\u0026rdquo; Door-God prints as a case study, this research constructs a Peircean triadic semiotic model and, under controlled conditions, conducts two types of image-innovation studies: traditional manual design and diffusion-model-assisted design. Quantitative indicators are established using AHP and FCE to compare visual appeal, cultural identity, and emotional resonance. The results show that AI-assisted design has advantages in visual appeal and creative diversity, while traditional design performs better in cultural identity and semantic stability. Experiments with general audiences confirm that AI outputs are readable but still fall short of manual design in terms of stability with respect to cultural conventions. This study proposes a \u0026ldquo;diffusion models \u0026times; semiotics \u0026times; quantitative evaluation\u0026rdquo; framework that provides a heritage-science-oriented analytical framework for evaluating the semantic sustainability of AI-generated cultural heritage imagery.\u003c/p\u003e","manuscriptTitle":"AI-Driven Innovation in Intangible Cultural Heritage:A Semiotic Analysis of Door-God Woodblock Prints Using Diffusion Models","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-02-10 16:10:57","doi":"10.21203/rs.3.rs-8747381/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"0bbefeb1-cf0c-48a2-9fe9-98cdd77b0ed3","owner":[],"postedDate":"February 10th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":62573531,"name":"Humanities/Cultural and media studies"},{"id":62573532,"name":"Social science/Cultural and media studies"},{"id":62573533,"name":"Physical sciences/Mathematics and computing"}],"tags":[],"updatedAt":"2026-03-11T19:24:14+00:00","versionOfRecord":[],"versionCreatedAt":"2026-02-10 16:10:57","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8747381","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8747381","identity":"rs-8747381","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}