Bengali Hate Speech Detection from Social Media using Ensemble Machine Learning Approach.

Bengali Hate Speech Detection from Social Media using Ensemble Machine Learning Approach. | 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 Bengali Hate Speech Detection from Social Media using Ensemble Machine Learning Approach. Sadia Tarin, Farzina Akther, Pranta Paul, Tanvinur Rahman Siam This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6488486/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The increasing prevalence of hate speech in Bengali across social media is a growing concern for the government and platform providers. Timely detection andremoval of such content are essential to preventing cyber violence and real-worldconflicts. However, the informal nature of online communication, with variationsin spelling and grammar, makes identification challenging.This study proposes an ensemble-based machine learning model for detectinghate speech in Bengali. A diverse dataset was collected from various onlinesources, followed by comprehensive preprocessing and classification into threetasks: (i) binary classification (Hate Speech vs. Not Hate), (ii) multi-label classification (categorizing different types of hate speech), and (iii) target identification.We explored machine learning algorithms alongside deep learning models and theensemble approach. In our proposed approach, we applied bagging with DecisionTree classifiers to create an ensemble model. Then, we built a stacking ensemblemodel, integrating Random Forest, SVM, Logistic Regression, and the baggingensemble classifiers. It achieved 91.49% accuracy with an F1-score of 91.49% onthe imbalanced dataset, while on the balanced dataset, accuracy improved to94.37% with an F1-score of 94.37%. Hate speech detection Natural Language Processing Machine Learning Ensemble Learning Bi-LSTM CNN Full Text Additional Declarations No competing interests reported. Supplementary Files BengaliHateSpeechDetectionfromSocialMediausingEnsembleMachineLearningApproach.pdf Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6488486","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":447989734,"identity":"91e0daf8-2c0a-47d4-862b-64b22be9a621","order_by":0,"name":"Sadia Tarin","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABEUlEQVRIiWNgGAWjYHACgwMgkg1EJDDY8PCDGQWEtCTAtaTJSDaAGAb4tYCUwsBhG4ileLTItx/eeODnD7t8Punegx8etjHzGJ9fnfjhgQGDPL/YAexWnEkrONiTkGzZJnMuWSKxjY3H7MbbzRJAhxnOnJ2AXQtDjsEBngRmAzaJHDOGxDYeoJazG0BaEgxuY9ci3//G4OCfhHqYFgke4xlnN//Ap4XhRo7BYZ6EwzAtBjwG/L3b8NpicONZwWGZtONALXnJEgnnEngkbvBus0gwkMDpF/n+5M0f39hUG8jPyD348UfZf3v+/rObb/6osJHnl8bhMATggdISYJUShJQja+E/QIzqUTAKRsEoGEEAAAQtXcRz1D6HAAAAAElFTkSuQmCC","orcid":"","institution":"Port City International University","correspondingAuthor":true,"prefix":"","firstName":"Sadia","middleName":"","lastName":"Tarin","suffix":""},{"id":447989735,"identity":"b0e109dd-b2a5-49c1-883c-49548a804ead","order_by":1,"name":"Farzina Akther","email":"","orcid":"","institution":"Port City International University","correspondingAuthor":false,"prefix":"","firstName":"Farzina","middleName":"","lastName":"Akther","suffix":""},{"id":447989736,"identity":"dd7e2a82-eb30-4c91-87ec-c6f33fb01aae","order_by":2,"name":"Pranta Paul","email":"","orcid":"","institution":"Port City International University","correspondingAuthor":false,"prefix":"","firstName":"Pranta","middleName":"","lastName":"Paul","suffix":""},{"id":447989737,"identity":"3456a0a5-d29b-4110-a8bf-90174aa20ed2","order_by":3,"name":"Tanvinur Rahman Siam","email":"","orcid":"","institution":"Chittagong University of Engineering \u0026 Technology","correspondingAuthor":false,"prefix":"","firstName":"Tanvinur","middleName":"Rahman","lastName":"Siam","suffix":""}],"badges":[],"createdAt":"2025-04-20 09:23:13","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-6488486/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6488486/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":81497992,"identity":"3863564f-6e94-4e54-8518-eb025d4f4d42","added_by":"auto","created_at":"2025-04-28 03:04:35","extension":"pdf","order_by":1,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1139636,"visible":true,"origin":"","legend":"","description":"","filename":"BengaliHateSpeechDetectionfromSocialMediausingEnsembleMachineLearningApproach..pdf","url":"https://assets-eu.researchsquare.com/files/rs-6488486/v1_covered_b3c7d979-ef4a-4a76-8ff9-db28e9f28083.pdf"},{"id":81497991,"identity":"2b36c473-2bac-47dc-af2e-b66329aa14d2","added_by":"auto","created_at":"2025-04-28 03:04:29","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":1077002,"visible":true,"origin":"","legend":"","description":"","filename":"BengaliHateSpeechDetectionfromSocialMediausingEnsembleMachineLearningApproach.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6488486/v1/dd0494d2b1bec1e37743ecb0.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Bengali Hate Speech Detection from Social Media using Ensemble Machine Learning Approach.","fulltext":[],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":false,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":true,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":true,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Hate speech detection, Natural Language Processing, Machine Learning, Ensemble Learning, Bi-LSTM, CNN","lastPublishedDoi":"10.21203/rs.3.rs-6488486/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6488486/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"The increasing prevalence of hate speech in Bengali across social media is a growing concern for the government and platform providers. Timely detection andremoval of such content are essential to preventing cyber violence and real-worldconflicts. However, the informal nature of online communication, with variationsin spelling and grammar, makes identification challenging.This study proposes an ensemble-based machine learning model for detectinghate speech in Bengali. A diverse dataset was collected from various onlinesources, followed by comprehensive preprocessing and classification into threetasks: (i) binary classification (Hate Speech vs. Not Hate), (ii) multi-label classification (categorizing different types of hate speech), and (iii) target identification.We explored machine learning algorithms alongside deep learning models and theensemble approach. In our proposed approach, we applied bagging with DecisionTree classifiers to create an ensemble model. Then, we built a stacking ensemblemodel, integrating Random Forest, SVM, Logistic Regression, and the baggingensemble classifiers. It achieved 91.49% accuracy with an F1-score of 91.49% onthe imbalanced dataset, while on the balanced dataset, accuracy improved to94.37% with an F1-score of 94.37%.","manuscriptTitle":"Bengali Hate Speech Detection from Social Media using Ensemble Machine Learning Approach.","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-28 03:04:25","doi":"10.21203/rs.3.rs-6488486/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ef942749-f709-4b65-9de6-fce612abb084","owner":[],"postedDate":"April 28th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2025-04-28T03:04:25+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-28 03:04:25","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6488486","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6488486","identity":"rs-6488486","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}