{"paper":{"title":"Continual Learning with Multilingual Foundation Model","license":"http://creativecommons.org/licenses/by/4.0/","headline":"Language-specific decision thresholds improve F1 scores by 2-5 percent in multilingual reclaimed slur detection without retraining the model.","cross_cats":["cs.AI","cs.LG"],"primary_cat":"cs.CL","authors_text":"Barathi Ganesh HB, Juuso Eronen, Michal Ptaszynski, Rene Melendez","submitted_at":"2026-05-13T12:10:47Z","abstract_excerpt":"This paper presents a multi-stage framework for detecting reclaimed slurs in multilingual social media discourse. It addresses the challenge of identifying reclamatory versus non-reclamatory usage of LGBTQ+-related slurs across English, Spanish, and Italian tweets. The framework handles three intertwined methodological challenges like data scarcity, class imbalance, and cross-linguistic variation in sentiment expression. It integrates data-driven model selection via cross-validation, semantic-preserving augmentation through back-translation, inductive transfer learning with dynamic epoch-level"},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"The threshold-based optimization yields 2-5% absolute F1 improvement without requiring model retraining.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That the back-translation via GPT-4o-mini accurately preserves the semantic content and class distribution ratios for the slur reclamation task.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"Framework using XLM-RoBERTa, back-translation augmentation, and language-specific thresholds detects reclaimed slurs with 2-5% F1 score gains.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"Language-specific decision thresholds improve F1 scores by 2-5 percent in multilingual reclaimed slur detection without retraining the model.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"bf47014e8b1d4939a992d828d48f762e5593f95dde596133fe79499e03515136"},"source":{"id":"2605.13415","kind":"arxiv","version":1},"verdict":{"id":"ca9da202-5e73-4691-a71e-d56d5307137c","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T19:36:34.915424Z","strongest_claim":"The threshold-based optimization yields 2-5% absolute F1 improvement without requiring model retraining.","one_line_summary":"Framework using XLM-RoBERTa, back-translation augmentation, and language-specific thresholds detects reclaimed slurs with 2-5% F1 score gains.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That the back-translation via GPT-4o-mini accurately preserves the semantic content and class distribution ratios for the slur reclamation task.","pith_extraction_headline":"Language-specific decision thresholds improve F1 scores by 2-5 percent in multilingual reclaimed slur detection without retraining the model."},"references":{"count":19,"sample":[{"doi":"","year":2024,"title":"E. Zsisku, A. Zubiaga, H. Dubossarsky, Hate speech detection and reclaimed language: Mitigating false positives and compounded discrimination, in: Proceedings of the 16th ACM Web Science Conference, 2","work_id":"6f3474e8-64bd-4b9b-8ec5-148ed10cd55c","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2022,"title":"B. R. Chakravarthi, R. Priyadharshini, T. Durairaj, J. P. McCrae, P. Buitelaar, P. Kumaresan, R. Pon- nusamy, Overview of the shared task on homophobia and transphobia detection in social media commen","work_id":"63aacc00-72e5-4b21-8eae-f9c0a326c649","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2020,"title":"Popa-Wyatt, Reclamation: Taking back control of words, Grazer Philosophische Studien 97 (2020) 159–176","work_id":"2eb95cfb-e7f9-423a-955a-5fdfc85b5810","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2026,"title":"C. Ferrando, L. Draetta, M. Madeddu, M. Sosto, V. Patti, P. Rosso, C. Bosco, J. Mata, E. Gualda, Multipride at evalita 2026: Overview of the multilingual automatic detection of slur reclamation in the","work_id":"e7497aeb-5d3e-47f8-8e71-56b200cdb748","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":1987,"title":"R. J. Tallarida, R. B. Murray, Chi-square test, in: Manual of pharmacologic calculations: with computer programs, Springer, 1987, pp. 140–142","work_id":"fb48e6c1-2088-494d-b4e6-a6cb8f490d97","ref_index":5,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":19,"snapshot_sha256":"c04bb4b4d8e15b7d0c5e73db8cca8989658a200c051dfb8efc8fde48807ab57c","internal_anchors":2},"formal_canon":{"evidence_count":2,"snapshot_sha256":"3e247ec02a4d77fce6ad5f732d7d920c89b1240c5a2e58a48bf3212dfa0f9e90"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}