Identifying Conflict Hotspots in Papua through Integrated Graph Theory and Clustering Analysis

Identifying Conflict Hotspots in Papua through Integrated Graph Theory and Clustering Analysis | 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 Research Article Identifying Conflict Hotspots in Papua through Integrated Graph Theory and Clustering Analysis Alvian Sroyer, Henderina Morin, Ishak Beno This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8360541/v1 This work is licensed under a CC BY 4.0 License Status: Under Review Version 1 posted 4 You are reading this latest preprint version Abstract Conflict in Papua has long attracted academic and policy attention, yet most studies remain descriptive and qualitative, leaving gaps in predictive capacity and quantitative validation. This study addresses these limitations by applying an integrated computational framework that combines graph-based modeling and clustering algorithms to conflict data recorded between 2018 and 2024. The aim is to identify central nodes in the conflict network, statistically validate spatial hotspots, and bridge social sciences with computational analysis. The research design is exploratory and quantitative. Conflict incidents were compiled from Human Rights Monitor (HRM) and Komnas HAM RI reports, then standardized into adjacency matrices and spatial coordinates. Using graph theory, centrality measures (degree and betweenness) and modularity were computed to determine structural hubs and community partitions within the conflict network. In parallel, clustering methods K-Means and DBSCAN were applied to classify districts into high, medium, and low conflict clusters. Robustness was tested through Silhouette Scores, ensuring statistical validation of hotspot identification. Results indicate a clear escalation of conflict from 2018 to 2024, with violent clashes rising from 5 to 20 incidents and displacement cases increasing twentyfold. Yahukimo, Intan Jaya, and Nduga consistently emerged as central hubs. Graph analysis confirmed Yahukimo’s dual role as hub and broker (degree centrality = 0.29, betweenness = 0.38), while modularity (Q ≈ 0.42) revealed distinct community partitions, especially between highland and western districts. Clustering validation yielded Silhouette Scores above 0.6 for high-intensity clusters, demonstrating strong cohesion and separation. The novelty of this study lies in its integration of graph-theoretic metrics and unsupervised clustering algorithms to build a hybrid quantitative model for conflict hotspot prediction in Papua. Previous works have typically relied on narrative or GIS-based spatial mapping without structural validation; this research introduces a dual-layer analytical approach that not only detects but also explains inter-district linkages and community modularity within the conflict network. The framework thus transforms descriptive conflict data into an interpretable mathematical topology, capable of generating early-warning indicators and policy-relevant insights.This research contributes in three ways. First, it provides quantitative indicators of conflict centrality, advancing beyond descriptive accounts. Second, it statistically validates conflict hotspots, distinguishing sustained clusters from sporadic outliers such as Fakfak and Tambrauw. Third, it demonstrates the value of integrating graph theory with clustering analysis, offering a predictive and holistic framework for conflict studies. The methodological novelty reinforces the potential of computational social science in Indonesia’s conflict research landscape, marking a shift from post-factum documentation toward proactive, data-driven conflict modeling. Conflict analysis graph theory clustering Papua computing Figures Figure 1 Figure 2 Introduction Conflict in Papua represents one of Indonesia’s most enduring and complex socio-political problems. For decades, the region has faced cycles of violence, human rights violations, and forced displacement, reflecting structural inequality between the state and indigenous Papuans (Komnas HAM 2023 ). The escalation of armed clashes between the West Papua National Liberation Army (TPNPB-OPM) and Indonesian security forces has triggered humanitarian crises that include the destruction of villages, civilian casualties, and internal displacement numbering in the thousands (Humanrightsmonitor 2024 ). These realities reveal that the Papuan conflict is not merely a political disagreement but a multidimensional system rooted in geography, identity, and governance (Fortunato and Hric 2016; Humanrightsmonitor 2024 ; Javed et al. 2018 ; Komnas HAM 2023 ). Most existing studies on the Papua conflict focus on qualitative narratives—historical marginalization, identity politics, or human-rights advocacy. While indispensable for contextual understanding, such approaches often lack the empirical depth and predictive capacity needed for evidence-based policy. The data that do exist—compiled by Komnas HAM RI, Human Rights Monitor, and the International Crisis Group—are typically underutilized in quantitative modeling. As a result, conflict research remains reactive: documentation follows events rather than forecasting their likelihood or diffusion. In general, research on conflict in Papua can be divided into three parts(Agustus et al. 2025 ; Zahidi and Othman 2024). First, descriptive and qualitative studies that highlight historical events, marginalization, and violence, but have limitations in predictive capacity. On the other hand, GIS-based mapping approaches that visualize conflict locations and hot spots are often not supported by statistical validation. So that computational research continues to be developed in the study of international conflicts(Wani 2024 ). Network science knowledge allows the identification of hubs and actors acting as intermediaries, while classification algorithms such as K-Means and DBSCAN can identify hotspots that correlate with sporadic research. However, in the Indonesian context, especially the conflict in Papua, the use of this computational method is still somewhat limited, although the availability of data is increasing through the International Crisis Group. (Humanrightsmonitor 2024 ; Komnas HAM 2023 ) With this in mind, there are several research gaps(Economou, Norman, and Gentleman 2025). First, no research in Papua has systematically used graph theory to identify conflicting communications. Although clustering algorithms are useful in epidemiology and criminology, they are hardly used for Papua conflict data. So each lacks a computational integration framework that allows graph analysis and classification as a reliable prediction or early warning tool(Khemani et al. 2024 ). Therefore, this research contributes (novelty) by using computational tools based on graph theory and clustering algorithms to analyze the conflict in Papua from 2018 to 2024 (Economou, Norman, and Gentleman 2025). This study investigates the existing literature in Indonesia on computational-based conflict analysis(Barabási et al. 2023 ). Practically, this study provides a data-driven dashboard to build a more efficient warning policy system and intervention strategy(Feyisope 2023 ). In graph theory, a conflict network can be represented as a graph G (V, E), where V denotes districts and E represents relational ties shared events, actors, or temporal co-occurrences. Structural indicators such as degree centrality reveal the extent to which a district is directly involved in incidents, while betweenness centrality measures its brokerage role in linking otherwise disconnected sub-regions (Barabási et al., 2023 ). Modularity (Q) quantifies the extent of community separation within the network—high Q values indicate cohesive clusters of conflict that are relatively isolated from others (Tokala et al., 2025 ). Clustering algorithms complement these structural metrics. K-Means partitions data into k groups by minimizing intra-cluster variance, thereby classifying regions by conflict intensity. DBSCAN, a density-based approach, identifies both dense conflict zones and sparse outliers—useful for detecting sporadic violence in peripheral districts (Daszykowski & Walczak, 2020 ). Together, these methods enable a dual-layer interpretation: graphs uncover relational structure, while clustering validates spatial concentration. The purpose of this study is to find conflict hot spots in conflict networks in Papua using graph theory, then statistically validate conflict hot spots using classification algorithms, and combine both approaches to evaluate academic contributions while providing suggestions for conflict resolution in Papua(Economou, Norman, and Gentleman 2025). Experimental Section Materials This research utilizes updated secondary conflict data compiled from Human Rights Monitor (HRM) and Komnas HAM RI covering the period 2018–2024(Agustus et al. 2025 ; Zahidi and Othman 2024). The final dataset includes 40 districts across Papua and Papua Barat, consisting of 173 documented incidents comprising armed clashes and extrajudicial killings. Each district is represented as a spatial node with coordinates taken from the official geospatial registry. The dataset was then standardized into numerical attributes (frequency of incidents per district) and categorical attributes (type of incidents) (McInnes, Healy, and Astels 2017; Rautenstrauch and Ohler 2025). Unlike the previous version of the study, the updated dataset reflects the full administrative landscape of Papua and aligns with the newly generated two-cluster spatial map, separating high-intensity conflict areas (Cluster 1) from low-intensity areas (Cluster 2). Methods This study uses a quantitative export approach that integrates graph theory and classification algorithms to identify conflict hotspots and simulation centers(Economou, Norman, and Gentleman 2025). The research process was carried out in several stages: Data preparation, with normalization of Location names, creation of proximity matrices from internal datasets, and spatial coordination for each district. (McInnes, Healy, and Astels 2017) Graph analysis, by constructing a network of conflicts where the distribution is a node, as well as identifying centers, intermediaries, and community members using centrality degrees, centrality between, and modularity(Economou, Norman, and Gentleman 2025; Tokala et al. 2025 )(Javed et al. 2018 ). Clustering Analysis, using the K-Means (K-3) algorithm to classify conflict intensity and DBSCAN to identify internal and outlier actors(Moncrieff, Kilibarda, and Gaggioli 2024). Cluster Validation, using Silhouette Score to assess the strength and coherence of each cluster Visualize, by looking at time trends, conflict networks, and spatial distributions using MATLAB applications(Moncrieff, Kilibarda, and Gaggioli 2024). Results and Discussion Results The spatial clustering analysis reveals a highly uneven and structured distribution of conflict across Papua during the 2018–2024 period(Agustus et al. 2025 ; Zahidi and Othman 2024). Using a two-cluster classification based on the total number of conflict incidents per district, the results demonstrate a clear separation between high-intensity and low-intensity conflict zones. Cluster 1 consists of seven districts with very high conflict intensity (≥ 20 recorded incidents), namely Intan Jaya, Yahukimo, Puncak, Nduga, Mimika, Puncak Jaya, and Pegunungan Bintang. These districts form a geographically contiguous hotspot corridor concentrated in the Central Highlands, indicating persistent and structurally embedded patterns of violence rather than isolated events. In contrast, Cluster 2 includes the remaining 35 districts, each recording fewer than 20 incidents over the observation period. These districts are predominantly located in coastal regions, lowland areas, and administrative urban centers, where conflict occurrences are relatively sporadic, fragmented, and weakly interconnected. The sharp contrast between the two clusters confirms that conflict in Papua is not spatially random but instead concentrated within specific geographic and socio-political contexts. The resulting spatial configuration reinforces the interpretation of the Central Highlands as the epicenter of conflict escalation in Papua, while surrounding regions function as peripheral or buffer zones with substantially lower exposure to sustained violence. By embedding spatial visualization within a statistically validated clustering framework, this analysis moves beyond descriptive mapping and provides quantitative evidence of localized conflict systems that can inform targeted intervention and early-warning strategies. Conflicts at the district level seem to be concentrated rather than spread evenly throughout Papua(Agustus et al. 2025 ; Zahidi and Othman 2024). Data shows that Intan Jaya, Yahukimo, and Nduga account for more than half of the total conflicts, with Intan Jaya alone accounting for 25% of all incidents. Incidentally, suburban districts such as Fakfak and Tambrauw each accounted for less than 5%, confirming their marginal role in the conflict network. A summary of the distribution at the district level, which includes the total amount and Contribution level, can be seen in Table 1 . Table 1 Distribution of Conflict Incidents by District (2018–2024) Jumlah Insiden Kabupaten/Kota Persentase (%) 45 Intan Jaya 25.14% 33 Yahukimo 18.44% 29 Puncak 16.20% 27 Nduga 15.08% 24 Mimika 13.41% 22 Puncak Jaya 12.29% 20 Pegunungan Bintang 11.17% 16 Maybrat 8.94% 13 Paniai 7.26% 11 Jayawijaya 6.15% 9 Deiyai, Dogiyai, Lanny Jaya 5.03% 8 Tolikara 4.47% 7 Jayapura (Kab.), Tambrauw 3.91% 6 Asmat, Nabire 3.35% 5 Mappi, Yalimo 2.79% 4 Boven Digoel, Fakfak, Manokwari, Merauke 2.23% 3 Kaimana, Kota Sorong, Mamberamo Tengah, Sorong Selatan, Teluk Bintuni 1.68% 2 Keerom, Kota Jayapura 1.12% 1 Biak Numfor, Kepulauan Yapen, Mamberamo Raya, Manokwari Selatan, Pegunungan Arfak, Sarmi, Waropen, Teluk Wondama 0.56% Table 1 shows the grouping of 40 districts/cities in the Papua region based on the same number of conflict incidents during the 2018–2024 period. This grouping provides a clearer picture of the level of conflict intensity, while facilitating the identification of hotspot areas and areas with low disturbances. In general, the data pattern shows that the distribution of conflict is uneven, but rather narrows down to certain areas, especially in the Central Highlands region. This map presents the results of a two-cluster spatial classification based on the total number of conflict incidents recorded in each district. Cluster 1 (orange squares) represents high-intensity conflict districts (≥ 20 incidents), primarily concentrated in the Central Highlands corridor, including Intan Jaya, Yahukimo, Nduga, Puncak, Mimika, Puncak Jaya, and Pegunungan Bintang. Cluster 2 (blue circles) consists of districts with low-to-moderate conflict intensity (< 20 incidents) distributed across coastal and lowland regions. The spatial pattern demonstrates a strong localization of conflict hotspots rather than a uniform regional spread. Mathematical Formulation of Metrics To reinforce the results, several indicators are calculated mathematically. The degree centrality for node vi is defined as follows: $$\:{C}_{D}\left({v}_{i}\right)=\frac{\text{d}\text{e}\text{g}\left({v}_{i}\right)}{n-1}$$ It measures the direct connectivity of nodes in the network. Betweenness centrality is calculated by: $$\:{C}_{B}\left({v}_{i}\right)=\sum\:_{s\ne\:{v}_{i}\ne\:t}\frac{{\sigma\:}_{st}\left({v}_{i}\right)}{{\sigma\:}_{st}}$$ This assesses the role of the node as the shortest trajectory link. To detect the community, modularity is used with the formula: $$\:Q=\frac{1}{2m}\sum\:_{i,j}\left[{A}_{ij}-\frac{{k}_{i}{k}_{j}}{2m}\right]\delta\:({c}_{i},{c}_{j})$$ Meanwhile, clustering is carried out with K-Means, which minimizes: $$\:J=\sum\:_{j=1}^{k}\sum\:_{i=1}^{n}{‖{x}_{i}-{\mu\:}_{j}‖}^{2}$$ And DBSCAN with neighbors- \(\:\epsilon\:\) : $$\:{N}_{\epsilon\:}\left(p\right)=\left\{q\in\:D⌋dist\:(p,q)\le\:\epsilon\:\right\}$$ Validation of the quality of the clister is carried out by the Silhouette Score : $$\:s\left(i\right)=\frac{b\left(i\right)-a\left(i\right)}{\text{max}\left\{a\left(i\right),b\left(i\right)\right\}}$$ A value above 0.6 indicates good cluster quality. Using modularity optimization, the detection immunity results in a modularity Q score of 0.42, which indicates that the network is not random but naturally divided into clear clusters(Economou, Norman, and Gentleman 2025; Tokala et al. 2025 ). Three different groups of people emerge. The first is Intan Jaya, Yahukimo, and Nduga, the second is Maybrat, Paniai, Pegunungan Bintang, and the third is Fakfak and Tambrauw, which are low-intensity peripheral groups. This division adds to the quantitative and qualitative claims previously made about the concentration of violence in the central highlands. Grouping analysis with K-Means and DBSCAN complements the graph-based approach(Daszykowski and Walczak 2020 ); the grouping of K-Means (K = 3) results in the same grouping as modularity analysis, while DBSCAN finds Fakfak and Tambrauw as noise points rather than synonymous clusters. The Silhouette score further demonstrates the strength of the clustering, as Yahukimo (0.66), Intan Jaya (0.68), and Nduga (0.64) all exceeded the 0.6 threshold, which indicates strong cohesion and separation in high-intensity clusters. Areas with high conflict can be clearly seen from areas with moderate and low conflict, as shown in Fig. 3. Table 2 summarizes the key metrics from graph theory and grouping analysis, showing the convergence between structural and spatial methods for combining important findings Table 2 Summary of Graph and Clustering Metrics Kabupaten Degree Centrality Betweenness Centrality Cluster Silhouette Score Intan Jaya 0.29 0.35 High (C1) 0.68 Yahukimo 0.29 0.38 High (C1) 0.66 Puncak 0.27 0.31 High (C1) 0.64 Nduga 0.25 0.32 High (C1) 0.63 Mimika 0.23 0.28 High(C1) 0.59 Puncak Jaya 0.22 0.26 High(C1) 0.57 Pegunungan Bintang 0.21 0.18 High(C1) 0.55 (Kelompok) 35 Kabupaten Klaster 2 0.12 (avg.) 0.09 (avg.) Low (C2) 0.41 (avg.) Table 2 presents a summary of graph metrics and clustering results for all observed regions, with an emphasis on seven districts in Cluster 1 that have very high conflict intensity (≥ 20 incidents), as well as one aggregate row for Cluster 2 consisting of 35 districts with lower levels of conflict. From the table, it can be seen that the districts in Cluster 1 have a much higher degree of centrality and centrality than the average of districts in Cluster 2. This indicates that these regions not only experience greater frequency of conflict, but also play a more important structural role in the conflict network—being the nodes most frequently connected or traversed in various chain of incidents. Intan Jaya and Yahukimo occupy the most dominant positions, with a degree centrality value of 0.29 and an intervalness of 0.35 and 0.38, respectively. This value shows that the two regions are the epicenters of the conflict network, where various types of violent events often recur and affect the dynamics of the surrounding region. Puncak, Nduga, Mimika, Puncak Jaya, and Pegunungan Bintang also have high centrality values, which reinforces the pattern that the Central Mountains belt is the region with the highest conflict density and the most complex event connectivity. Meanwhile, the group of districts that entered Cluster 2 was characterized by a much lower centrality value (average of 0.12 for degree and 0.09 for betweenness), as well as a smaller silhouette score value (0.41). This lower silhouette score shows that the districts in Cluster 2 have a more sporadic pattern of conflict, are isolated, and do not form a strong network structure. In other words, these areas are not involved in a systematic configuration of conflict and are more likely to experience incidents of a local or incidental nature. Discussion As mentioned earlier, research on the conflict in Papua is generally based on a qualitative approach that includes descriptive analysis, socio-political perspectives, and ethnographic studies(Economou, Norman, and Gentleman 2025). This contribution is essential for contextual understanding, but it also lacks predictive power and often provides quantitative indicators that can be used for intervention strategies or early warning systems. Three main characteristics were identified from the literature review. First of all, there is no systematic research that uses graph theory to reduce the number of actors or locations in conflict networks in Papua; this approach has been widely used in international conflict studies. Second, the analysis of hotspot conflicts is mostly descriptive without using additional algorithms such as K-Means and DBSCAN to determine whether the hotspot in question is statistically significant. Third, there is currently no plan to integrate a structural network-based approach with a cluster-based spatial perspective, which results in a holistic picture of the dynamics of conflict in Papua. The methodology of this study involves the application of computational tools that combine graph analysis and classification algorithms on conflict data in Papua from 2018 to 2024.(Brandt et al. 2022 ; Hegre et al. 2019 )(Chauvel 2019 ) A major contribution of graph analysis is its ability to identify hostpots in conflict networks(Economou, Norman, and Gentleman 2025; Tokala et al. 2025 ). The findings of the study showed that Yahukimo, Intan Jaya, and Nduga were the locations of the community groups with the highest level of centrality, where Yahukimo had the highest dual role as a center and intermediary (0.29 and 0.38). It provides qualitative information that was previously only described in descriptive form in literature. Furthermore, modularity analysis using a Q score of ≈ 0.42 showed that the conflict network in Papua was not very strong, which indicated that communities could interact with each other in a friendly manner. The three groups that emerged included the high-intensity group, namely Yahukimo, Intan Jaya, and Nduga; the medium-intensity group, namely Maybrat, the Bintang Mountains, and Paniai; and the low-intensity groups on the west coast of Papua, namely Fakfak and Tambrauw. The study also provides quantitative validation of previous qualitative claims that violence is more consistent in the Central Highlands region (D’Orazio and Lin 2022) Clustering analysis also contributes by improving the results of graph analysis(Agustus et al. 2025 ; Zahidi and Othman 2024). Through the K-Means algorithm, the conflict distribution is divided into three main groups, namely high, medium, and low. The results showed that Tambrauw and Fakfak were included in the low kloster, Maybrat, Paniai, and Pegunungan Bintang were included in the medium kloster, and Yahukimo, Intan Jaya, and Nduga were the highest klosters and therefore were significant hostpots. The DBSCAN algorithm then provides insights by detecting outliers or noise, where Tambrauw and Fakfak are identified as regions with sporadic points that do not affect the continuous cluster pattern. Further validation using the Silhouette Score yielded an average of 0.6 for high clatter intensity, indicating high internal cohesion and clear inter-cluster segregation. Thus, this study can provide quantitative information about the hotspot of conflict in Papua that was previously unavailable. Integration of graph and clustering findings The results of the graph analysis were consistent with the results of spatial clustering. Formally, for each node vi, an integrative function is defined: $$\:f\left({v}_{i}\right)=\alpha\:{C}_{D}\left({v}_{i}\right)+\beta\:{C}_{B}\left({v}_{i}\right)+\gamma\:Q\left({c}_{i}\right)$$ With \(\:\alpha\:,\beta\:,\gamma\:\) as weights, the value \(\:f\left({v}_{i}\right)\) High indicates that the node has a dominant position in the graph, as well as being the nucleus of the cluster. In this study, Yahukimo, Intan Jaya, and Nduga obtained scores \(\:f\left({v}_{i}\right)\) In accordance with its role as a conflict hotspot. The K-Means optimization principle explains this consistency, in line with \(\:{C}_{D}\) and \(\:{C}_{B}\) tinggi akan cenderung mendekati centroid \(\:{\mu\:}_{j}\) . So it is naturally included in the high-intensity cluster. Thus, the integration between graph theory and clustering provides a solid mathematical framework for predicting new conflict hotspots in the future. The findings in this study are not only limited to the application of graph theory and clustering clearly and concisely, but also include the integration of the two theories(Economou, Norman, and Gentleman 2025). The results of this integration provide some important insights. First, there is a convergence of graph and clustering results that simultaneously identify Yahukimo, Intan Jaya, and Nduga as conflicting parties, thereby increasing the reliability of the findings. Second, this study offers a very comprehensive perspective, and graph analysis explains how the area is spatially regulated based on internal factors. Third, this integration framework creates predictive advantages, as evidenced by centrality values and Silhouette Scores over time, which can serve as hot spot indicators. Because of this, this study secretly presents some methodologies that were previously used in the study of conflict in Papua. This research makes three important contributions from an academic perspective. First, methodological changes in the study of the Papuan conflict using network science and cluster validation methods. It also expands the literature on computational conflict analysis in Southeast Asia. Second, this study confirms previous qualitative findings; for example, numerical measures such as Silhouette score, modularity, and centrality indicate that the central mountains are at the center of the conflict. Third, this study shows how social sciences and computational methods are interconnected, utilizing analytical tools such as MATLAB, thereby fostering interdisciplinary collaboration. From a practical perspective, the findings of the study are directly correlated with public policy. First, graph indicators and groupings can serve as key components in conflict early warning systems, for example, through the development of data-driven monitoring dashboards. Second, the results of the study make it possible to prioritize interventions in districts with high centrality, such as Yahukimo, Intan Jaya, and Nduga, so that resources can be allocated more effectively. Third, early identification of districts with low but potentially increasing incidences, such as the Bintang Mountains, can help prevent conflict escalation. Because of this practical relevance, research focused on peace and the protection of civilians is of great importance to Indonesian governments, local governments, and civil society organizations. (Franciscans and Email 2022) Therefore, this study succeeded in filling the methodological gap in the literature related to the Papuan conflict. This study uses centrality indicators to measure quantitatively, unlike previous research, which was only descriptive. This study statistically validated the grouping of hotspots, which were previously only visually mapped. This research shows that conflict is a networked and grouped system, not just an event. These results not only confirm long-standing claims about the amount of violence in the central mountains but also enhance the status of the analysis as an actionable framework and predictor. Nevertheless, some limitations need to be considered. First, reports that rely on secondary data from Komnas HAM and HRM may be inaccurate. These findings would be more reliable if the dataset were stained with ACLED. Second, because the research data is still annual, it is not able to capture more detailed temporal dynamics. Analyzing monthly or weekly data can help with predictions. Third, although the study focused on districts as units of analysis, actor-based modeling—such as security forces or armed groups—can provide a deeper understanding. To overcome these limitations, future research could combine broader datasets, dynamic network models, and machine learning techniques such as graph neural networks to improve predictive potential. Policy Implications and Strategic Relevance The findings of this study carry significant implications for conflict governance and public policy in Papua. The convergence of graph-theoretic metrics and clustering results demonstrates that conflict is not only concentrated geographically but also structurally embedded within a limited number of districts that function as central nodes in the broader conflict network. Districts such as Yahukimo, Intan Jaya, and Nduga exhibit both high conflict intensity and high network centrality, indicating that violence in these areas is recurrent, interconnected, and capable of influencing surrounding regions. From a policy perspective, this structure implies that conflict mitigation strategies should prioritize interventions in high-centrality districts rather than dispersing resources evenly across the region. Targeted peacebuilding, security sector reform, and humanitarian protection efforts in these core hotspots are likely to produce spillover benefits by weakening the structural backbone of the conflict network. In contrast, districts within the low-intensity cluster require monitoring-oriented strategies focused on prevention and resilience-building, as sporadic incidents in these areas may escalate if structural pressures from the core zones are left unaddressed. The integration of clustering validation further strengthens the policy relevance of the findings. High Silhouette Scores for the core conflict cluster confirm that these hotspots represent statistically coherent and persistent patterns of violence rather than short-term fluctuations. This characteristic makes the proposed framework particularly suitable for incorporation into early-warning systems, where rising centrality or cluster cohesion can serve as quantitative indicators of escalation risk. More broadly, the study demonstrates the value of computational social science for conflict analysis in Indonesia. By transforming descriptive conflict records into a networked and clustered system, the framework provides actionable insights that bridge academic analysis and policy application. Such an approach enables decision-makers to move from reactive documentation toward proactive, data-driven conflict prevention strategies tailored to the specific structural dynamics of Papua. Conclusion This study examined the dynamics of conflict in Papua during the 2018–2024 period by integrating graph-theoretic analysis and clustering algorithms within a unified computational framework. The results demonstrate that conflict in Papua is neither random nor evenly distributed across space. Instead, it forms a structured and localized system characterized by distinct hotspots, strong inter-district linkages, and persistent patterns of violence concentrated in specific geographic corridors. Temporal analysis indicates a clear escalation of conflict intensity over time, with both violent incidents and forced displacement increasing substantially across the observation period. Spatially, the two-cluster classification reveals that a small group of districts—Intan Jaya, Yahukimo, Puncak, Nduga, Mimika, Puncak Jaya, and Pegunungan Bintang—constitutes the core conflict hotspots. These districts, primarily located in the Central Highlands, account for a disproportionate share of total incidents and form a contiguous zone of high vulnerability. In contrast, the majority of districts fall into a low-intensity cluster characterized by sporadic and weakly connected incidents. Graph-based network analysis further strengthens these findings by showing that hotspot districts are not only locations of frequent violence but also occupy structurally central positions within the conflict network. High values of degree centrality and betweenness centrality—particularly in Yahukimo and Intan Jaya—indicate that these districts function as hubs and brokers through which conflict dynamics propagate across regions. The modularity structure of the network confirms the existence of well-defined conflict communities, reinforcing the interpretation of Papua’s conflict as an interconnected system rather than a collection of isolated events. The integration of clustering validation with network metrics represents a key methodological contribution of this study. High Silhouette Scores for the core conflict cluster provide statistical confirmation that these hotspots are cohesive and persistent, not temporary anomalies. By embedding spatial visualization within a validated graph–clustering framework, this research advances conflict analysis beyond descriptive GIS mapping and qualitative narratives toward a reproducible, data-driven model capable of supporting early-warning and risk assessment. From a practical perspective, the findings suggest that conflict mitigation and peacebuilding efforts in Papua should prioritize structurally central districts within the high-intensity cluster. Interventions targeted at these core nodes are likely to yield broader stabilizing effects across the network, whereas uniform or geographically diffuse strategies may dilute impact. At the same time, districts in the low-intensity cluster require preventive monitoring to reduce the risk of spillover and escalation. Despite its contributions, this study has limitations. The analysis relies on secondary conflict reports with annual temporal resolution, which may obscure finer-grained dynamics. Future research could incorporate higher-frequency data, actor-based network modeling, and advanced machine learning approaches—such as dynamic networks or graph neural networks—to enhance predictive capability. Nevertheless, this study provides robust empirical evidence that conflict in Papua operates as a structured, clustered, and networked system, offering both methodological innovation and actionable insights for conflict prevention and sustainable peacebuilding. Declarations Author Contribution All authors made significant contributions to this study. The concept and design of the research were carried out collaboratively. Data collection and analysis were carried out by the authors. Interpretation of results, script writing, and critical revision are carried out together. All authors have read and approved the final manuscript. Acknowledgement This research is funded by a research grant from the Ministry of Higher Education, Research, and Technology through the Fundamental Research program in 2025. We really appreciate the financial support provided for the implementation of this research. Also, to LPPM Cenderawasih University for administrative management, so that the implementation of research can run well. References Agustus F, Sjafirial D, Andoko M, Muniasari CM, Savitri PS, Nurisnaeny, and Heny Batara Maya (2025) Strengthening the Papua Steering Committee Strategy: Reducing Instability for Accelerated Development in Papua, Indonesia. Social Sci Humanit Open 11. 10.1016/j.ssaho.2025.101413 Barabási DánielL, Bianconi G, Bullmore E, Burgess M, Chung SY, Eliassi-Rad T, Dileep George, et al (2023) Neuroscience Needs Network Science. J Neurosci 43(34):5989–5995. 10.1523/JNEUROSCI.1014-23.2023 Brandt PT, Vito D’Orazio, Khan L, Li YF, Osorio J, and Marcus Sianan (2022) Conflict Forecasting with Event Data and Spatio-Temporal Graph Convolutional Networks. Int Interact 48(4):800–822. 10.1080/03050629.2022.2036987 Chauvel R (2019) Governance and the Cycle of Violence in Papua: The Nduga Crisis. Asia-Pacific Journal: Japan Focus 17(2). 10.1017/s1557466019015018 D’Orazio V, and Yu Lin (2022) Forecasting Conflict in Africa with Automated Machine Learning Systems. Int Interact 48(4):714–738. 10.1080/03050629.2022.2017290 Daszykowski M, Walczak B (2020) 2.26 - Density-Based Clustering Methods. Comprehensive Chemometrics: Chemical and Biochemical Data Analysis . Second Edition: Four Volume Set 2:565–580. 10.1016/B978-0-444-64165-6.03005-6 Economou KN, Cassie R, Norman, Gentleman WC (2025) Identifying Robust Features of Community Structure in Complex Networks. Phys Rev E 111(4):44303. 10.1103/PhysRevE.111.044303 Feyisope C (2023) Early Warning System of Conflict Prevention Strategy in Nigeria. International J Res Sci Innovation X(V 163–185. 10.51244/IJRSI Fortunato S, and Darko Hric (2016) Community Detection in Networks: A User Guide. Phys Rep 659:1–44. 10.1016/j.physrep.2016.09.002 Franciscans B, Skp-kc, and Papua Email. CONTROL AND PARALYZATION, OF PAPUA ’, S POLITICAL RESISTANCE ’ THE QUESTION OF HUMAN RIGHTS WILL BE DEALT WITH LATER ’ (2022) Paralyzing Papuan Political Resistance, Human Rights Regulated from Behind. 48(1) Hegre Håvard, Allansson M, Basedau M, Colaresi M, Croicu M, Fjelde H, Hoyles F et al (2019) ViEWS: A Political Violence Early-Warning System. J Peace Res 56(2):155–174. 10.1177/0022343319823860 Humanrightsmonitor (2024) Annual Report 2023: Human Rights and Conflict in West Papua. Humanrightsmonitor.Org (December 2018). https://humanrightsmonitor.org/reports/hrm-annual-report-human-rights-and-conflict-in-west-papua-2023/ Javed M, Aqib MS, Younis S, Latif J, Qadir, and Adeel Baig (2018) Community Detection in Networks: A Multidisciplinary Review. J Netw Comput Appl 108:87–111. 10.1016/j.jnca.2018.02.011 Khemani B, Patil S, Ketan Kotecha, and, Tanwar S (2024) A Review of Graph Neural Networks: Concepts, Architectures, Techniques, Challenges, Datasets, Applications, and Future Directions. J Big Data 11(1). 10.1186/s40537-023-00876-4 Komnas HAM (2023) Laporan Komnas HAM RI Tahun 2023.: 1–167 McInnes L, John Healy, and, Astels S (2017) Hdbscan: Hierarchical Density Based Clustering. J Open Source Softw 2(11):205. 10.21105/joss.00205 Moncrieff M, Kilibarda P, and Gloria Gaggioli (2024) Social Network Analysis and Counterterrorism: A Double-Edged Sword for International Humanitarian Law. J Confl Secur Law 29(1):165–183. 10.1093/jcsl/krae002 Rautenstrauch P, and Uwe Ohler (2025) Shortcomings of Silhouette in Single-Cell Integration Benchmarking. Nature Biotechnology (Mdc). 10.1038/s41587-025-02743-4 Tokala S, Enduri MK, Jaya Lakshmi T, and Koduru Hajarathaiah (2025) Evaluating Community Detection Algorithms: A Focus on Effectiveness and Efficiency. J Scientometr Res 14(1):62–74. 10.5530/jscires.20250839 Wani AA (2024) Comprehensive Analysis of Clustering Algorithms: Exploring Limitations and Innovative Solutions. PeerJ Comput Sci 10:1–45. 10.7717/PEERJ-CS.2286 Zahidi M, Syaprin, and Muhammad Fuad Bin Othman (2024) Human Rights Issue in Papua: A Systematic Literature Review. Revista UNISCI 2024(65):107–126. 10.31439/UNISCI-203 Additional Declarations No competing interests reported. Cite Share Download PDF Status: Under Review Version 1 posted Reviewers invited by journal 05 May, 2026 Editor assigned by journal 13 Jan, 2026 Submission checks completed at journal 16 Dec, 2025 First submitted to journal 14 Dec, 2025 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-8360541","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":636785381,"identity":"f4950b65-78b8-4287-8046-1c201d797a38","order_by":0,"name":"Alvian Sroyer","email":"data:image/png;base64,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","orcid":"","institution":"Cenderawasih University","correspondingAuthor":true,"prefix":"","firstName":"Alvian","middleName":"","lastName":"Sroyer","suffix":""},{"id":636785383,"identity":"8aa07da2-f7a6-432c-af5a-0ab193f7933a","order_by":1,"name":"Henderina Morin","email":"","orcid":"","institution":"Cenderawasih University","correspondingAuthor":false,"prefix":"","firstName":"Henderina","middleName":"","lastName":"Morin","suffix":""},{"id":636785386,"identity":"15038492-4e50-4458-bd28-045934c38987","order_by":2,"name":"Ishak Beno","email":"","orcid":"","institution":"Cenderawasih University","correspondingAuthor":false,"prefix":"","firstName":"Ishak","middleName":"","lastName":"Beno","suffix":""}],"badges":[],"createdAt":"2025-12-14 23:53:04","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8360541/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8360541/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":109264546,"identity":"573ac23d-6a65-445f-8c5f-51c9228137a0","added_by":"auto","created_at":"2026-05-14 12:10:21","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":105377,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTrend of Conflict Incidents in Papua (2018–2024)\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-8360541/v1/e72787865a72a3d9104ebe5a.jpeg"},{"id":109264299,"identity":"9d6f64c1-4483-4f4e-9730-f8f8842b8886","added_by":"auto","created_at":"2026-05-14 12:09:35","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":159003,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFigure 1. \u003c/strong\u003eSpatial Distribution of Conflict Clusters in Papua (2018–2024)\u003c/p\u003e","description":"","filename":"PetaKonflik2Klater.jpg","url":"https://assets-eu.researchsquare.com/files/rs-8360541/v1/c812851795d60545dc55acaa.jpg"},{"id":109264726,"identity":"74e2c3f1-5c58-4956-9f42-8bbc0a4310c1","added_by":"auto","created_at":"2026-05-14 12:10:55","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":515160,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8360541/v1/28afd683-92c1-418d-9dcc-5bc6873ca6a4.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Identifying Conflict Hotspots in Papua through Integrated Graph Theory and Clustering Analysis","fulltext":[{"header":"Introduction","content":"\u003cp\u003eConflict in Papua represents one of Indonesia\u0026rsquo;s most enduring and complex socio-political problems. For decades, the region has faced cycles of violence, human rights violations, and forced displacement, reflecting structural inequality between the state and indigenous Papuans (Komnas HAM \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). The escalation of armed clashes between the West Papua National Liberation Army (TPNPB-OPM) and Indonesian security forces has triggered humanitarian crises that include the destruction of villages, civilian casualties, and internal displacement numbering in the thousands (Humanrightsmonitor \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). These realities reveal that the Papuan conflict is not merely a political disagreement but a multidimensional system rooted in geography, identity, and governance (Fortunato and Hric 2016; Humanrightsmonitor \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Javed et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e; Komnas HAM \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003c/p\u003e \u003cp\u003eMost existing studies on the Papua conflict focus on qualitative narratives\u0026mdash;historical marginalization, identity politics, or human-rights advocacy. While indispensable for contextual understanding, such approaches often lack the empirical depth and predictive capacity needed for evidence-based policy. The data that do exist\u0026mdash;compiled by Komnas HAM RI, Human Rights Monitor, and the International Crisis Group\u0026mdash;are typically underutilized in quantitative modeling. As a result, conflict research remains reactive: documentation follows events rather than forecasting their likelihood or diffusion.\u003c/p\u003e \u003cp\u003eIn general, research on conflict in Papua can be divided into three parts(Agustus et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zahidi and Othman 2024). First, descriptive and qualitative studies that highlight historical events, marginalization, and violence, but have limitations in predictive capacity. On the other hand, GIS-based mapping approaches that visualize conflict locations and hot spots are often not supported by statistical validation. So that computational research continues to be developed in the study of international conflicts(Wani \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2024\u003c/span\u003e). Network science knowledge allows the identification of hubs and actors acting as intermediaries, while classification algorithms such as K-Means and DBSCAN can identify hotspots that correlate with sporadic research. However, in the Indonesian context, especially the conflict in Papua, the use of this computational method is still somewhat limited, although the availability of data is increasing through the International Crisis Group. (Humanrightsmonitor \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Komnas HAM \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2023\u003c/span\u003e)\u003c/p\u003e \u003cp\u003eWith this in mind, there are several research gaps(Economou, Norman, and Gentleman 2025). First, no research in Papua has systematically used graph theory to identify conflicting communications. Although clustering algorithms are useful in epidemiology and criminology, they are hardly used for Papua conflict data. So each lacks a computational integration framework that allows graph analysis and classification as a reliable prediction or early warning tool(Khemani et al. \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2024\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTherefore, this research contributes (novelty) by using computational tools based on graph theory and clustering algorithms to analyze the conflict in Papua from 2018 to 2024 (Economou, Norman, and Gentleman 2025). This study investigates the existing literature in Indonesia on computational-based conflict analysis(Barab\u0026aacute;si et al. \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Practically, this study provides a data-driven dashboard to build a more efficient warning policy system and intervention strategy(Feyisope \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2023\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn graph theory, a conflict network can be represented as a graph G (V, E), where V denotes districts and E represents relational ties shared events, actors, or temporal co-occurrences. Structural indicators such as degree centrality reveal the extent to which a district is directly involved in incidents, while betweenness centrality measures its brokerage role in linking otherwise disconnected sub-regions (Barab\u0026aacute;si et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Modularity (Q) quantifies the extent of community separation within the network\u0026mdash;high Q values indicate cohesive clusters of conflict that are relatively isolated from others (Tokala et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eClustering algorithms complement these structural metrics. K-Means partitions data into k groups by minimizing intra-cluster variance, thereby classifying regions by conflict intensity. DBSCAN, a density-based approach, identifies both dense conflict zones and sparse outliers\u0026mdash;useful for detecting sporadic violence in peripheral districts (Daszykowski \u0026amp; Walczak, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e). Together, these methods enable a dual-layer interpretation: graphs uncover relational structure, while clustering validates spatial concentration.\u003c/p\u003e \u003cp\u003eThe purpose of this study is to find conflict hot spots in conflict networks in Papua using graph theory, then statistically validate conflict hot spots using classification algorithms, and combine both approaches to evaluate academic contributions while providing suggestions for conflict resolution in Papua(Economou, Norman, and Gentleman 2025).\u003c/p\u003e"},{"header":"Experimental Section","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eMaterials\u003c/h2\u003e \u003cp\u003eThis research utilizes updated secondary conflict data compiled from Human Rights Monitor (HRM) and Komnas HAM RI covering the period 2018\u0026ndash;2024(Agustus et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zahidi and Othman 2024). The final dataset includes 40 districts across Papua and Papua Barat, consisting of 173 documented incidents comprising armed clashes and extrajudicial killings. Each district is represented as a spatial node with coordinates taken from the official geospatial registry. The dataset was then standardized into numerical attributes (frequency of incidents per district) and categorical attributes (type of incidents) (McInnes, Healy, and Astels 2017; Rautenstrauch and Ohler 2025). Unlike the previous version of the study, the updated dataset reflects the full administrative landscape of Papua and aligns with the newly generated two-cluster spatial map, separating high-intensity conflict areas (Cluster 1) from low-intensity areas (Cluster 2).\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eMethods\u003c/h3\u003e\n\u003cp\u003eThis study uses a quantitative export approach that integrates graph theory and classification algorithms to identify conflict hotspots and simulation centers(Economou, Norman, and Gentleman 2025). The research process was carried out in several stages:\u003c/p\u003e \u003cp\u003e \u003col\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eData preparation, with normalization of Location names, creation of proximity matrices from internal datasets, and spatial coordination for each district. (McInnes, Healy, and Astels 2017)\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eGraph analysis, by constructing a network of conflicts where the distribution is a node, as well as identifying centers, intermediaries, and community members using centrality degrees, centrality between, and modularity(Economou, Norman, and Gentleman 2025; Tokala et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e)(Javed et al. \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2018\u003c/span\u003e).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eClustering Analysis, using the K-Means (K-3) algorithm to classify conflict intensity and DBSCAN to identify internal and outlier actors(Moncrieff, Kilibarda, and Gaggioli 2024).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eCluster Validation, using Silhouette Score to assess the strength and coherence of each cluster\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003cspan\u003e \u003cli\u003e \u003cp\u003eVisualize, by looking at time trends, conflict networks, and spatial distributions using MATLAB applications(Moncrieff, Kilibarda, and Gaggioli 2024).\u003c/p\u003e \u003c/li\u003e \u003c/span\u003e \u003c/ol\u003e \u003c/p\u003e"},{"header":"Results and Discussion","content":"\n\u003ch3\u003eResults\u003c/h3\u003e\n\u003cp\u003eThe spatial clustering analysis reveals a highly uneven and structured distribution of conflict across Papua during the 2018\u0026ndash;2024 period(Agustus et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zahidi and Othman 2024). Using a two-cluster classification based on the total number of conflict incidents per district, the results demonstrate a clear separation between high-intensity and low-intensity conflict zones. Cluster 1 consists of seven districts with very high conflict intensity (\u0026ge;\u0026thinsp;20 recorded incidents), namely Intan Jaya, Yahukimo, Puncak, Nduga, Mimika, Puncak Jaya, and Pegunungan Bintang. These districts form a geographically contiguous hotspot corridor concentrated in the Central Highlands, indicating persistent and structurally embedded patterns of violence rather than isolated events.\u003c/p\u003e \u003cp\u003eIn contrast, Cluster 2 includes the remaining 35 districts, each recording fewer than 20 incidents over the observation period. These districts are predominantly located in coastal regions, lowland areas, and administrative urban centers, where conflict occurrences are relatively sporadic, fragmented, and weakly interconnected. The sharp contrast between the two clusters confirms that conflict in Papua is not spatially random but instead concentrated within specific geographic and socio-political contexts.\u003c/p\u003e \u003cp\u003eThe resulting spatial configuration reinforces the interpretation of the Central Highlands as the epicenter of conflict escalation in Papua, while surrounding regions function as peripheral or buffer zones with substantially lower exposure to sustained violence. By embedding spatial visualization within a statistically validated clustering framework, this analysis moves beyond descriptive mapping and provides quantitative evidence of localized conflict systems that can inform targeted intervention and early-warning strategies.\u003c/p\u003e \u003cp\u003eConflicts at the district level seem to be concentrated rather than spread evenly throughout Papua(Agustus et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zahidi and Othman 2024). Data shows that Intan Jaya, Yahukimo, and Nduga account for more than half of the total conflicts, with Intan Jaya alone accounting for 25% of all incidents. Incidentally, suburban districts such as Fakfak and Tambrauw each accounted for less than 5%, confirming their marginal role in the conflict network. A summary of the distribution at the district level, which includes the total amount and Contribution level, can be seen in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e.\u003c/p\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\u003eDistribution of Conflict Incidents by District (2018\u0026ndash;2024)\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=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eJumlah Insiden\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKabupaten/Kota\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePersentase (%)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eIntan Jaya\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e25.14%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eYahukimo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e18.44%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePuncak\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e16.20%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eNduga\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e15.08%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e24\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMimika\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e13.41%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePuncak Jaya\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e12.29%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePegunungan Bintang\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e11.17%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e16\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMaybrat\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e8.94%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e13\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePaniai\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e7.26%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eJayawijaya\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e6.15%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e9\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDeiyai, Dogiyai, Lanny Jaya\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e5.03%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e8\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eTolikara\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e4.47%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eJayapura (Kab.), Tambrauw\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3.91%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e6\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAsmat, Nabire\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3.35%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMappi, Yalimo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.79%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBoven Digoel, Fakfak, Manokwari, Merauke\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e2.23%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKaimana, Kota Sorong, Mamberamo Tengah, Sorong Selatan, Teluk Bintuni\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.68%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eKeerom, Kota Jayapura\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e1.12%\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eBiak Numfor, Kepulauan Yapen, Mamberamo Raya, Manokwari Selatan, Pegunungan Arfak, Sarmi, Waropen, Teluk Wondama\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.56%\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\u003eTable\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e shows the grouping of 40 districts/cities in the Papua region based on the same number of conflict incidents during the 2018\u0026ndash;2024 period. This grouping provides a clearer picture of the level of conflict intensity, while facilitating the identification of hotspot areas and areas with low disturbances. In general, the data pattern shows that the distribution of conflict is uneven, but rather narrows down to certain areas, especially in the Central Highlands region.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThis map presents the results of a two-cluster spatial classification based on the total number of conflict incidents recorded in each district. Cluster 1 (orange squares) represents high-intensity conflict districts (\u0026ge;\u0026thinsp;20 incidents), primarily concentrated in the Central Highlands corridor, including Intan Jaya, Yahukimo, Nduga, Puncak, Mimika, Puncak Jaya, and Pegunungan Bintang. Cluster 2 (blue circles) consists of districts with low-to-moderate conflict intensity (\u0026lt;\u0026thinsp;20 incidents) distributed across coastal and lowland regions. The spatial pattern demonstrates a strong localization of conflict hotspots rather than a uniform regional spread.\u003c/p\u003e\n\u003ch3\u003eMathematical Formulation of Metrics\u003c/h3\u003e\n\u003cp\u003eTo reinforce the results, several indicators are calculated mathematically. The degree centrality for node vi is defined as follows:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:{C}_{D}\\left({v}_{i}\\right)=\\frac{\\text{d}\\text{e}\\text{g}\\left({v}_{i}\\right)}{n-1}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eIt measures the direct connectivity of nodes in the network.\u003c/p\u003e \u003cp\u003eBetweenness centrality is calculated by:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:{C}_{B}\\left({v}_{i}\\right)=\\sum\\:_{s\\ne\\:{v}_{i}\\ne\\:t}\\frac{{\\sigma\\:}_{st}\\left({v}_{i}\\right)}{{\\sigma\\:}_{st}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThis assesses the role of the node as the shortest trajectory link.\u003c/p\u003e \u003cp\u003eTo detect the community, modularity is used with the formula:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:Q=\\frac{1}{2m}\\sum\\:_{i,j}\\left[{A}_{ij}-\\frac{{k}_{i}{k}_{j}}{2m}\\right]\\delta\\:({c}_{i},{c}_{j})$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eMeanwhile, clustering is carried out with K-Means, which minimizes:\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:J=\\sum\\:_{j=1}^{k}\\sum\\:_{i=1}^{n}{‖{x}_{i}-{\\mu\\:}_{j}‖}^{2}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eAnd DBSCAN with neighbors-\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\epsilon\\:\\)\u003c/span\u003e\u003c/span\u003e :\u003cdiv id=\"Eque\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Eque\" name=\"EquationSource\"\u003e\n$$\\:{N}_{\\epsilon\\:}\\left(p\\right)=\\left\\{q\\in\\:D\u0026rfloor;dist\\:(p,q)\\le\\:\\epsilon\\:\\right\\}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eValidation of the quality of the clister is carried out by the Silhouette Score :\u003cdiv id=\"Equf\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equf\" name=\"EquationSource\"\u003e\n$$\\:s\\left(i\\right)=\\frac{b\\left(i\\right)-a\\left(i\\right)}{\\text{max}\\left\\{a\\left(i\\right),b\\left(i\\right)\\right\\}}$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eA value above 0.6 indicates good cluster quality.\u003c/p\u003e \u003cp\u003eUsing modularity optimization, the detection immunity results in a modularity Q score of 0.42, which indicates that the network is not random but naturally divided into clear clusters(Economou, Norman, and Gentleman 2025; Tokala et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). Three different groups of people emerge. The first is Intan Jaya, Yahukimo, and Nduga, the second is Maybrat, Paniai, Pegunungan Bintang, and the third is Fakfak and Tambrauw, which are low-intensity peripheral groups. This division adds to the quantitative and qualitative claims previously made about the concentration of violence in the central highlands.\u003c/p\u003e \u003cp\u003eGrouping analysis with K-Means and DBSCAN complements the graph-based approach(Daszykowski and Walczak \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2020\u003c/span\u003e); the grouping of K-Means (K\u0026thinsp;=\u0026thinsp;3) results in the same grouping as modularity analysis, while DBSCAN finds Fakfak and Tambrauw as noise points rather than synonymous clusters. The Silhouette score further demonstrates the strength of the clustering, as Yahukimo (0.66), Intan Jaya (0.68), and Nduga (0.64) all exceeded the 0.6 threshold, which indicates strong cohesion and separation in high-intensity clusters. Areas with high conflict can be clearly seen from areas with moderate and low conflict, as shown in Fig.\u0026nbsp;3.\u003c/p\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e summarizes the key metrics from graph theory and grouping analysis, showing the convergence between structural and spatial methods for combining important findings\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\u003eSummary of Graph and Clustering Metrics\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=\"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 \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eKabupaten\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDegree Centrality\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eBetweenness Centrality\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eCluster\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSilhouette Score\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eIntan Jaya\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh (C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eYahukimo\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.38\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh (C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.66\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePuncak\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh (C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.64\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNduga\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.32\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh (C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.63\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMimika\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.23\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.28\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh(C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.59\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePuncak Jaya\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh(C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePegunungan Bintang\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eHigh(C1)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.55\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e(Kelompok) 35 Kabupaten Klaster 2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e0.12 (avg.)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e0.09 (avg.)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003eLow (C2)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.41 (avg.)\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\u003eTable\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e presents a summary of graph metrics and clustering results for all observed regions, with an emphasis on seven districts in Cluster 1 that have very high conflict intensity (\u0026ge;\u0026thinsp;20 incidents), as well as one aggregate row for Cluster 2 consisting of 35 districts with lower levels of conflict. From the table, it can be seen that the districts in Cluster 1 have a much higher \u003cem\u003edegree of centrality and centrality\u003c/em\u003e than the average of districts in Cluster 2. This indicates that these regions not only experience greater frequency of conflict, but also play a more important structural role in the conflict network\u0026mdash;being the nodes most frequently connected or traversed in various chain of incidents.\u003c/p\u003e \u003cp\u003eIntan Jaya and Yahukimo occupy the most dominant positions, with a degree centrality value of 0.29 and an intervalness of 0.35 and 0.38, respectively. This value shows that the two regions are the epicenters of the conflict network, where various types of violent events often recur and affect the dynamics of the surrounding region. Puncak, Nduga, Mimika, Puncak Jaya, and Pegunungan Bintang also have high centrality values, which reinforces the pattern that the Central Mountains belt is the region with the highest conflict density and the most complex event connectivity.\u003c/p\u003e \u003cp\u003eMeanwhile, the group of districts that entered Cluster 2 was characterized by a much lower centrality value (average of 0.12 for degree and 0.09 for betweenness), as well as a smaller silhouette score value (0.41). This lower silhouette score shows that the districts in Cluster 2 have a more sporadic pattern of conflict, are isolated, and do not form a strong network structure. In other words, these areas are not involved in a systematic configuration of conflict and are more likely to experience incidents of a local or incidental nature.\u003c/p\u003e\n\u003ch3\u003eDiscussion\u003c/h3\u003e\n\u003cp\u003eAs mentioned earlier, research on the conflict in Papua is generally based on a qualitative approach that includes descriptive analysis, socio-political perspectives, and ethnographic studies(Economou, Norman, and Gentleman 2025). This contribution is essential for contextual understanding, but it also lacks predictive power and often provides quantitative indicators that can be used for intervention strategies or early warning systems. Three main characteristics were identified from the literature review. First of all, there is no systematic research that uses graph theory to reduce the number of actors or locations in conflict networks in Papua; this approach has been widely used in international conflict studies. Second, the analysis of hotspot conflicts is mostly descriptive without using additional algorithms such as K-Means and DBSCAN to determine whether the hotspot in question is statistically significant. Third, there is currently no plan to integrate a structural network-based approach with a cluster-based spatial perspective, which results in a holistic picture of the dynamics of conflict in Papua. The methodology of this study involves the application of computational tools that combine graph analysis and classification algorithms on conflict data in Papua from 2018 to 2024.(Brandt et al. \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2022\u003c/span\u003e; Hegre et al. \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2019\u003c/span\u003e)(Chauvel \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2019\u003c/span\u003e)\u003c/p\u003e \u003cp\u003eA major contribution of graph analysis is its ability to identify hostpots in conflict networks(Economou, Norman, and Gentleman 2025; Tokala et al. \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2025\u003c/span\u003e). The findings of the study showed that Yahukimo, Intan Jaya, and Nduga were the locations of the community groups with the highest level of centrality, where Yahukimo had the highest dual role as a center and intermediary (0.29 and 0.38). It provides qualitative information that was previously only described in descriptive form in literature. Furthermore, modularity analysis using a Q score of \u0026asymp;\u0026thinsp;0.42 showed that the conflict network in Papua was not very strong, which indicated that communities could interact with each other in a friendly manner. The three groups that emerged included the high-intensity group, namely Yahukimo, Intan Jaya, and Nduga; the medium-intensity group, namely Maybrat, the Bintang Mountains, and Paniai; and the low-intensity groups on the west coast of Papua, namely Fakfak and Tambrauw. The study also provides quantitative validation of previous qualitative claims that violence is more consistent in the Central Highlands region (D\u0026rsquo;Orazio and Lin 2022)\u003c/p\u003e \u003cp\u003eClustering analysis also contributes by improving the results of graph analysis(Agustus et al. \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Zahidi and Othman 2024). Through the K-Means algorithm, the conflict distribution is divided into three main groups, namely high, medium, and low. The results showed that Tambrauw and Fakfak were included in the low kloster, Maybrat, Paniai, and Pegunungan Bintang were included in the medium kloster, and Yahukimo, Intan Jaya, and Nduga were the highest klosters and therefore were significant hostpots. The DBSCAN algorithm then provides insights by detecting outliers or noise, where Tambrauw and Fakfak are identified as regions with sporadic points that do not affect the continuous cluster pattern. Further validation using the Silhouette Score yielded an average of 0.6 for high clatter intensity, indicating high internal cohesion and clear inter-cluster segregation. Thus, this study can provide quantitative information about the hotspot of conflict in Papua that was previously unavailable.\u003c/p\u003e\n\u003ch3\u003eIntegration of graph and clustering findings\u003c/h3\u003e\n\u003cp\u003eThe results of the graph analysis were consistent with the results of spatial clustering. Formally, for each node vi, an integrative function is defined:\u003cdiv id=\"Equg\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equg\" name=\"EquationSource\"\u003e\n$$\\:f\\left({v}_{i}\\right)=\\alpha\\:{C}_{D}\\left({v}_{i}\\right)+\\beta\\:{C}_{B}\\left({v}_{i}\\right)+\\gamma\\:Q\\left({c}_{i}\\right)$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eWith \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:,\\beta\\:,\\gamma\\:\\)\u003c/span\u003e\u003c/span\u003e as weights, the value \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:f\\left({v}_{i}\\right)\\)\u003c/span\u003e\u003c/span\u003e High indicates that the node has a dominant position in the graph, as well as being the nucleus of the cluster. In this study, Yahukimo, Intan Jaya, and Nduga obtained scores \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:f\\left({v}_{i}\\right)\\)\u003c/span\u003e\u003c/span\u003e In accordance with its role as a conflict hotspot.\u003c/p\u003e \u003cp\u003eThe K-Means optimization principle explains this consistency, in line with \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{C}_{D}\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{C}_{B}\\)\u003c/span\u003e\u003c/span\u003e tinggi akan cenderung mendekati centroid \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\mu\\:}_{j}\\)\u003c/span\u003e\u003c/span\u003e. So it is naturally included in the high-intensity cluster. Thus, the integration between graph theory and clustering provides a solid mathematical framework for predicting new conflict hotspots in the future.\u003c/p\u003e \u003cp\u003eThe findings in this study are not only limited to the application of graph theory and clustering clearly and concisely, but also include the integration of the two theories(Economou, Norman, and Gentleman 2025). The results of this integration provide some important insights. First, there is a convergence of graph and clustering results that simultaneously identify Yahukimo, Intan Jaya, and Nduga as conflicting parties, thereby increasing the reliability of the findings. Second, this study offers a very comprehensive perspective, and graph analysis explains how the area is spatially regulated based on internal factors. Third, this integration framework creates predictive advantages, as evidenced by centrality values and Silhouette Scores over time, which can serve as hot spot indicators. Because of this, this study secretly presents some methodologies that were previously used in the study of conflict in Papua.\u003c/p\u003e \u003cp\u003eThis research makes three important contributions from an academic perspective. First, methodological changes in the study of the Papuan conflict using network science and cluster validation methods. It also expands the literature on computational conflict analysis in Southeast Asia. Second, this study confirms previous qualitative findings; for example, numerical measures such as Silhouette score, modularity, and centrality indicate that the central mountains are at the center of the conflict. Third, this study shows how social sciences and computational methods are interconnected, utilizing analytical tools such as MATLAB, thereby fostering interdisciplinary collaboration.\u003c/p\u003e \u003cp\u003eFrom a practical perspective, the findings of the study are directly correlated with public policy. First, graph indicators and groupings can serve as key components in conflict early warning systems, for example, through the development of data-driven monitoring dashboards. Second, the results of the study make it possible to prioritize interventions in districts with high centrality, such as Yahukimo, Intan Jaya, and Nduga, so that resources can be allocated more effectively. Third, early identification of districts with low but potentially increasing incidences, such as the Bintang Mountains, can help prevent conflict escalation. Because of this practical relevance, research focused on peace and the protection of civilians is of great importance to Indonesian governments, local governments, and civil society organizations. (Franciscans and Email 2022)\u003c/p\u003e \u003cp\u003eTherefore, this study succeeded in filling the methodological gap in the literature related to the Papuan conflict. This study uses centrality indicators to measure quantitatively, unlike previous research, which was only descriptive. This study statistically validated the grouping of hotspots, which were previously only visually mapped. This research shows that conflict is a networked and grouped system, not just an event. These results not only confirm long-standing claims about the amount of violence in the central mountains but also enhance the status of the analysis as an actionable framework and predictor.\u003c/p\u003e \u003cp\u003eNevertheless, some limitations need to be considered. First, reports that rely on secondary data from Komnas HAM and HRM may be inaccurate. These findings would be more reliable if the dataset were stained with ACLED. Second, because the research data is still annual, it is not able to capture more detailed temporal dynamics. Analyzing monthly or weekly data can help with predictions. Third, although the study focused on districts as units of analysis, actor-based modeling\u0026mdash;such as security forces or armed groups\u0026mdash;can provide a deeper understanding. To overcome these limitations, future research could combine broader datasets, dynamic network models, and machine learning techniques such as graph neural networks to improve predictive potential.\u003c/p\u003e\n\u003ch3\u003ePolicy Implications and Strategic Relevance\u003c/h3\u003e\n\u003cp\u003eThe findings of this study carry significant implications for conflict governance and public policy in Papua. The convergence of graph-theoretic metrics and clustering results demonstrates that conflict is not only concentrated geographically but also structurally embedded within a limited number of districts that function as central nodes in the broader conflict network. Districts such as Yahukimo, Intan Jaya, and Nduga exhibit both high conflict intensity and high network centrality, indicating that violence in these areas is recurrent, interconnected, and capable of influencing surrounding regions.\u003c/p\u003e \u003cp\u003eFrom a policy perspective, this structure implies that conflict mitigation strategies should prioritize interventions in high-centrality districts rather than dispersing resources evenly across the region. Targeted peacebuilding, security sector reform, and humanitarian protection efforts in these core hotspots are likely to produce spillover benefits by weakening the structural backbone of the conflict network. In contrast, districts within the low-intensity cluster require monitoring-oriented strategies focused on prevention and resilience-building, as sporadic incidents in these areas may escalate if structural pressures from the core zones are left unaddressed.\u003c/p\u003e \u003cp\u003eThe integration of clustering validation further strengthens the policy relevance of the findings. High Silhouette Scores for the core conflict cluster confirm that these hotspots represent statistically coherent and persistent patterns of violence rather than short-term fluctuations. This characteristic makes the proposed framework particularly suitable for incorporation into early-warning systems, where rising centrality or cluster cohesion can serve as quantitative indicators of escalation risk.\u003c/p\u003e \u003cp\u003eMore broadly, the study demonstrates the value of computational social science for conflict analysis in Indonesia. By transforming descriptive conflict records into a networked and clustered system, the framework provides actionable insights that bridge academic analysis and policy application. Such an approach enables decision-makers to move from reactive documentation toward proactive, data-driven conflict prevention strategies tailored to the specific structural dynamics of Papua.\u003c/p\u003e"},{"header":"Conclusion","content":" \u003cp\u003eThis study examined the dynamics of conflict in Papua during the 2018\u0026ndash;2024 period by integrating graph-theoretic analysis and clustering algorithms within a unified computational framework. The results demonstrate that conflict in Papua is neither random nor evenly distributed across space. Instead, it forms a structured and localized system characterized by distinct hotspots, strong inter-district linkages, and persistent patterns of violence concentrated in specific geographic corridors.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eTemporal analysis indicates a clear escalation of conflict intensity over time, with both violent incidents and forced displacement increasing substantially across the observation period. Spatially, the two-cluster classification reveals that a small group of districts\u0026mdash;Intan Jaya, Yahukimo, Puncak, Nduga, Mimika, Puncak Jaya, and Pegunungan Bintang\u0026mdash;constitutes the core conflict hotspots. These districts, primarily located in the Central Highlands, account for a disproportionate share of total incidents and form a contiguous zone of high vulnerability. In contrast, the majority of districts fall into a low-intensity cluster characterized by sporadic and weakly connected incidents.\u003c/p\u003e \u003cp\u003eGraph-based network analysis further strengthens these findings by showing that hotspot districts are not only locations of frequent violence but also occupy structurally central positions within the conflict network. High values of degree centrality and betweenness centrality\u0026mdash;particularly in Yahukimo and Intan Jaya\u0026mdash;indicate that these districts function as hubs and brokers through which conflict dynamics propagate across regions. The modularity structure of the network confirms the existence of well-defined conflict communities, reinforcing the interpretation of Papua\u0026rsquo;s conflict as an interconnected system rather than a collection of isolated events.\u003c/p\u003e \u003cp\u003eThe integration of clustering validation with network metrics represents a key methodological contribution of this study. High Silhouette Scores for the core conflict cluster provide statistical confirmation that these hotspots are cohesive and persistent, not temporary anomalies. By embedding spatial visualization within a validated graph\u0026ndash;clustering framework, this research advances conflict analysis beyond descriptive GIS mapping and qualitative narratives toward a reproducible, data-driven model capable of supporting early-warning and risk assessment.\u003c/p\u003e \u003cp\u003eFrom a practical perspective, the findings suggest that conflict mitigation and peacebuilding efforts in Papua should prioritize structurally central districts within the high-intensity cluster. Interventions targeted at these core nodes are likely to yield broader stabilizing effects across the network, whereas uniform or geographically diffuse strategies may dilute impact. At the same time, districts in the low-intensity cluster require preventive monitoring to reduce the risk of spillover and escalation.\u003c/p\u003e \u003cp\u003eDespite its contributions, this study has limitations. The analysis relies on secondary conflict reports with annual temporal resolution, which may obscure finer-grained dynamics. Future research could incorporate higher-frequency data, actor-based network modeling, and advanced machine learning approaches\u0026mdash;such as dynamic networks or graph neural networks\u0026mdash;to enhance predictive capability. Nevertheless, this study provides robust empirical evidence that conflict in Papua operates as a structured, clustered, and networked system, offering both methodological innovation and actionable insights for conflict prevention and sustainable peacebuilding.\u003c/p\u003e "},{"header":"Declarations","content":"\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eAll authors made significant contributions to this study. The concept and design of the research were carried out collaboratively. Data collection and analysis were carried out by the authors. Interpretation of results, script writing, and critical revision are carried out together. All authors have read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e \u003cp\u003eThis research is funded by a research grant from the Ministry of Higher Education, Research, and Technology through the Fundamental Research program in 2025. We really appreciate the financial support provided for the implementation of this research. Also, to LPPM Cenderawasih University for administrative management, so that the implementation of research can run well.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAgustus F, Sjafirial D, Andoko M, Muniasari CM, Savitri PS, Nurisnaeny, and Heny Batara Maya (2025) Strengthening the Papua Steering Committee Strategy: Reducing Instability for Accelerated Development in Papua, Indonesia. Social Sci Humanit Open 11. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.ssaho.2025.101413\u003c/span\u003e\u003cspan address=\"10.1016/j.ssaho.2025.101413\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBarab\u0026aacute;si D\u0026aacute;nielL, Bianconi G, Bullmore E, Burgess M, Chung SY, Eliassi-Rad T, Dileep George, et al (2023) Neuroscience Needs Network Science. J Neurosci 43(34):5989\u0026ndash;5995. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1523/JNEUROSCI.1014-23.2023\u003c/span\u003e\u003cspan address=\"10.1523/JNEUROSCI.1014-23.2023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBrandt PT, Vito D\u0026rsquo;Orazio, Khan L, Li YF, Osorio J, and Marcus Sianan (2022) Conflict Forecasting with Event Data and Spatio-Temporal Graph Convolutional Networks. Int Interact 48(4):800\u0026ndash;822. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/03050629.2022.2036987\u003c/span\u003e\u003cspan address=\"10.1080/03050629.2022.2036987\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChauvel R (2019) Governance and the Cycle of Violence in Papua: The Nduga Crisis. Asia-Pacific Journal: Japan Focus 17(2). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1017/s1557466019015018\u003c/span\u003e\u003cspan address=\"10.1017/s1557466019015018\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eD\u0026rsquo;Orazio V, and Yu Lin (2022) Forecasting Conflict in Africa with Automated Machine Learning Systems. Int Interact 48(4):714\u0026ndash;738. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1080/03050629.2022.2017290\u003c/span\u003e\u003cspan address=\"10.1080/03050629.2022.2017290\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDaszykowski M, Walczak B (2020) 2.26 - Density-Based Clustering Methods. \u003cem\u003eComprehensive Chemometrics: Chemical and Biochemical Data Analysis\u003c/em\u003e. Second Edition: Four Volume Set 2:565\u0026ndash;580. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/B978-0-444-64165-6.03005-6\u003c/span\u003e\u003cspan address=\"10.1016/B978-0-444-64165-6.03005-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEconomou KN, Cassie R, Norman, Gentleman WC (2025) Identifying Robust Features of Community Structure in Complex Networks. Phys Rev E 111(4):44303. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1103/PhysRevE.111.044303\u003c/span\u003e\u003cspan address=\"10.1103/PhysRevE.111.044303\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeyisope C (2023) Early Warning System of Conflict Prevention Strategy in Nigeria. International J Res Sci Innovation X(V 163\u0026ndash;185. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.51244/IJRSI\u003c/span\u003e\u003cspan address=\"10.51244/IJRSI\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFortunato S, and Darko Hric (2016) Community Detection in Networks: A User Guide. Phys Rep 659:1\u0026ndash;44. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.physrep.2016.09.002\u003c/span\u003e\u003cspan address=\"10.1016/j.physrep.2016.09.002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFranciscans B, Skp-kc, and Papua Email. CONTROL AND PARALYZATION, OF PAPUA \u0026rsquo;, S POLITICAL RESISTANCE \u0026rsquo; THE QUESTION OF HUMAN RIGHTS WILL BE DEALT WITH LATER \u0026rsquo; (2022) Paralyzing Papuan Political Resistance, Human Rights Regulated from Behind. 48(1)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHegre H\u0026aring;vard, Allansson M, Basedau M, Colaresi M, Croicu M, Fjelde H, Hoyles F et al (2019) ViEWS: A Political Violence Early-Warning System. J Peace Res 56(2):155\u0026ndash;174. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1177/0022343319823860\u003c/span\u003e\u003cspan address=\"10.1177/0022343319823860\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHumanrightsmonitor (2024) Annual Report 2023: Human Rights and Conflict in West Papua. \u003cem\u003eHumanrightsmonitor.Org\u003c/em\u003e (December 2018). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://humanrightsmonitor.org/reports/hrm-annual-report-human-rights-and-conflict-in-west-papua-2023/\u003c/span\u003e\u003cspan address=\"https://humanrightsmonitor.org/reports/hrm-annual-report-human-rights-and-conflict-in-west-papua-2023/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJaved M, Aqib MS, Younis S, Latif J, Qadir, and Adeel Baig (2018) Community Detection in Networks: A Multidisciplinary Review. J Netw Comput Appl 108:87\u0026ndash;111. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1016/j.jnca.2018.02.011\u003c/span\u003e\u003cspan address=\"10.1016/j.jnca.2018.02.011\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhemani B, Patil S, Ketan Kotecha, and, Tanwar S (2024) A Review of Graph Neural Networks: Concepts, Architectures, Techniques, Challenges, Datasets, Applications, and Future Directions. J Big Data 11(1). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s40537-023-00876-4\u003c/span\u003e\u003cspan address=\"10.1186/s40537-023-00876-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKomnas HAM (2023) Laporan Komnas HAM RI Tahun 2023.: 1\u0026ndash;167\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMcInnes L, John Healy, and, Astels S (2017) Hdbscan: Hierarchical Density Based Clustering. J Open Source Softw 2(11):205. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21105/joss.00205\u003c/span\u003e\u003cspan address=\"10.21105/joss.00205\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoncrieff M, Kilibarda P, and Gloria Gaggioli (2024) Social Network Analysis and Counterterrorism: A Double-Edged Sword for International Humanitarian Law. J Confl Secur Law 29(1):165\u0026ndash;183. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1093/jcsl/krae002\u003c/span\u003e\u003cspan address=\"10.1093/jcsl/krae002\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRautenstrauch P, and Uwe Ohler (2025) Shortcomings of Silhouette in Single-Cell Integration Benchmarking. Nature Biotechnology (Mdc). \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1038/s41587-025-02743-4\u003c/span\u003e\u003cspan address=\"10.1038/s41587-025-02743-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTokala S, Enduri MK, Jaya Lakshmi T, and Koduru Hajarathaiah (2025) Evaluating Community Detection Algorithms: A Focus on Effectiveness and Efficiency. J Scientometr Res 14(1):62\u0026ndash;74. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.5530/jscires.20250839\u003c/span\u003e\u003cspan address=\"10.5530/jscires.20250839\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWani AA (2024) Comprehensive Analysis of Clustering Algorithms: Exploring Limitations and Innovative Solutions. PeerJ Comput Sci 10:1\u0026ndash;45. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.7717/PEERJ-CS.2286\u003c/span\u003e\u003cspan address=\"10.7717/PEERJ-CS.2286\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZahidi M, Syaprin, and Muhammad Fuad Bin Othman (2024) Human Rights Issue in Papua: A Systematic Literature Review. Revista UNISCI 2024(65):107\u0026ndash;126. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.31439/UNISCI-203\u003c/span\u003e\u003cspan address=\"10.31439/UNISCI-203\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":false,"email":"","identity":"sn-social-sciences","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"SN Social Sciences","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"VoR Journals","inReviewEnabled":false,"inReviewRevisionsEnabled":false},"keywords":"Conflict analysis, graph theory, clustering, Papua, computing","lastPublishedDoi":"10.21203/rs.3.rs-8360541/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8360541/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eConflict in Papua has long attracted academic and policy attention, yet most studies remain descriptive and qualitative, leaving gaps in predictive capacity and quantitative validation. This study addresses these limitations by applying an integrated computational framework that combines graph-based modeling and clustering algorithms to conflict data recorded between 2018 and 2024. The aim is to identify central nodes in the conflict network, statistically validate spatial hotspots, and bridge social sciences with computational analysis. The research design is exploratory and quantitative. Conflict incidents were compiled from Human Rights Monitor (HRM) and Komnas HAM RI reports, then standardized into adjacency matrices and spatial coordinates. Using graph theory, centrality measures (degree and betweenness) and modularity were computed to determine structural hubs and community partitions within the conflict network. In parallel, clustering methods K-Means and DBSCAN were applied to classify districts into high, medium, and low conflict clusters. Robustness was tested through Silhouette Scores, ensuring statistical validation of hotspot identification. Results indicate a clear escalation of conflict from 2018 to 2024, with violent clashes rising from 5 to 20 incidents and displacement cases increasing twentyfold. Yahukimo, Intan Jaya, and Nduga consistently emerged as central hubs. Graph analysis confirmed Yahukimo\u0026rsquo;s dual role as hub and broker (degree centrality\u0026thinsp;=\u0026thinsp;0.29, betweenness\u0026thinsp;=\u0026thinsp;0.38), while modularity (Q\u0026thinsp;\u0026asymp;\u0026thinsp;0.42) revealed distinct community partitions, especially between highland and western districts. Clustering validation yielded Silhouette Scores above 0.6 for high-intensity clusters, demonstrating strong cohesion and separation. The novelty of this study lies in its integration of graph-theoretic metrics and unsupervised clustering algorithms to build a hybrid quantitative model for conflict hotspot prediction in Papua. Previous works have typically relied on narrative or GIS-based spatial mapping without structural validation; this research introduces a dual-layer analytical approach that not only detects but also explains inter-district linkages and community modularity within the conflict network. The framework thus transforms descriptive conflict data into an interpretable mathematical topology, capable of generating early-warning indicators and policy-relevant insights.This research contributes in three ways. First, it provides quantitative indicators of conflict centrality, advancing beyond descriptive accounts. Second, it statistically validates conflict hotspots, distinguishing sustained clusters from sporadic outliers such as Fakfak and Tambrauw. Third, it demonstrates the value of integrating graph theory with clustering analysis, offering a predictive and holistic framework for conflict studies. The methodological novelty reinforces the potential of computational social science in Indonesia\u0026rsquo;s conflict research landscape, marking a shift from post-factum documentation toward proactive, data-driven conflict modeling.\u003c/p\u003e","manuscriptTitle":"Identifying Conflict Hotspots in Papua through Integrated Graph Theory and Clustering Analysis","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-14 12:08:43","doi":"10.21203/rs.3.rs-8360541/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"reviewersInvited","content":"","date":"2026-05-06T00:59:03+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-01-14T04:51:03+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2025-12-17T04:02:51+00:00","index":"","fulltext":""},{"type":"submitted","content":"SN Social Sciences","date":"2025-12-14T23:35:21+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":false,"email":"","identity":"sn-social-sciences","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"","title":"SN Social Sciences","twitterHandle":"","acdcEnabled":false,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"VoR Journals","inReviewEnabled":false,"inReviewRevisionsEnabled":false}}],"origin":"","ownerIdentity":"032e1041-02a2-4aa7-a145-fb57ff9002be","owner":[],"postedDate":"May 14th, 2026","published":true,"recentEditorialEvents":[{"type":"reviewersInvited","content":"17","date":"2026-05-06T00:59:03+00:00","index":"","fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"under-review","subjectAreas":[],"tags":[],"updatedAt":"2026-05-14T12:08:47+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-14 12:08:43","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8360541","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8360541","identity":"rs-8360541","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}