<\/p>\n\n\n\n
Current cultural evaluations for large language models (LLMs) often reduce culture to single-turn factual recall via MCQs, failing to capture a common use case: users seeking practical help over multiple turns in culturally grounded scenarios. We introduce CultureConverse, a scalable, multilingual simulation and evaluation harness for culturally grounded assistant dialogue that covers 10 East and Southeast Asian regions, 58 subgroup identities, and 7 domains. Each simulated and evaluated episode produces a scored interaction where the assistant assists the user and infers cultural constraints from partial information. The resulting CultureConverse-DS dataset contains 14,610 benchmark (evaluation) episodes and 274,295 oracle-guided (gold-mode) dialogues. In our benchmark evaluation of 18 models, GPT-5 mini achieves the highest assistance quality. Human annotation experiments suggest that our evaluation framework is a sufficient proxy for human judgment. Performance gains from fine-tuning on 27,860 high-quality CultureConverse-DS samples improve in-domain assistance and transfer out-of-domain to cultural MCQ and safety classification benchmarks. We release the harness, both splits, and judge prompts to support interactive evaluation of cultural competency.<\/p>\n","protected":false},"excerpt":{"rendered":"
Current cultural evaluations for large language models (LLMs) often reduce culture to single-turn factual recall via MCQs, failing to capture a common use case: users seeking practical help over multiple turns in culturally grounded scenarios. We introduce CultureConverse, a scalable, multilingual simulation and evaluation harness for culturally grounded assistant dialogue that covers 10 East and […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","msr-author-ordering":[{"type":"text","value":"Bryan Chen Zhengyu Tan","user_id":0},{"type":"text","value":"Weihua Zheng","user_id":0},{"type":"text","value":"Thong T. Doan","user_id":0},{"type":"text","value":"Ngoc Bich Doan","user_id":0},{"type":"text","value":"Jia Wang Peh","user_id":0},{"type":"user_nicename","value":"Xiaoyuan Yi","user_id":"40768"},{"type":"user_nicename","value":"Jing Yao","user_id":"41925"},{"type":"user_nicename","value":"Xing Xie","user_id":"34906"},{"type":"text","value":"Nancy F. Chen","user_id":0},{"type":"text","value":"Zhengyuan Liu","user_id":0},{"type":"text","value":"Jinyeong Bak","user_id":0},{"type":"text","value":"Wafi Shamdi","user_id":0},{"type":"text","value":"Soo-Kai Chie","user_id":0},{"type":"text","value":"Liew Yu Siong","user_id":0},{"type":"text","value":"Aina Azyyati Binti Mohamad Rezal","user_id":0},{"type":"text","value":"Lew Yan Yan Vanessa","user_id":0},{"type":"text","value":"Huarui Wu","user_id":0},{"type":"text","value":"Dylan Raharja","user_id":0},{"type":"text","value":"Nadya Yuki Wangsajaya","user_id":0},{"type":"text","value":"Akane Fukushige","user_id":0},{"type":"text","value":"Kazushi Kato","user_id":0},{"type":"text","value":"K. Inoue","user_id":0},{"type":"text","value":"Tatsuya Kawahara","user_id":0},{"type":"text","value":"J. Seo","user_id":0},{"type":"text","value":"Dongjun Kim","user_id":0},{"type":"text","value":"Seungyoon Lee","user_id":0},{"type":"text","value":"Z. Pang","user_id":0},{"type":"text","value":"Rui Yang Tan","user_id":0},{"type":"text","value":"Charibeth Cheng","user_id":0},{"type":"text","value":"M. R. J. Estuar","user_id":0},{"type":"text","value":"J. Montalan","user_id":0},{"type":"text","value":"Duc Minh Pham","user_id":0},{"type":"text","value":"Roy Ka-Wei Lee","user_id":0}],"msr_publishername":"","msr_publisher_other":"","msr_booktitle":"","msr_chapter":"","msr_edition":"","msr_editors":"","msr_how_published":"arXiv","msr_isbn":"","msr_issue":"","msr_journal":"","msr_number":"","msr_organization":"","msr_pages_string":"","msr_page_range_start":"","msr_page_range_end":"","msr_series":"","msr_volume":"","msr_copyright":"","msr_conference_name":"","msr_doi":"","msr_arxiv_id":"2608.28405","msr_mag_id":"","msr_other_authors":"","msr_other_contributors":"","msr_speaker":"","msr_award":"","msr_affiliation":"","msr_institution":"","msr_host":"","msr_version":"","msr_duration":"","msr_release_tracker_id":"","msr_highlight_type":"","msr_date_display_format":"","msr_main_download_label":"","msr_external_link_label":"","msr_doi_label":"","msr_published_date":"2026-08-28","msr_startdate":"","msr_presentation_date":"","msr_highlight_text":"","msr_notes":"","msr_longbiography":"","msr_publicationurl":"","msr_external_url":"","msr_secondary_video_url":"","msr_conference_url":"","msr_journal_url":"","msr_year":2026,"msr_month":8,"msr_day":28,"msr_microsoftintellectualproperty":false,"msr_pub_id":"eeda96a691c6b596f0700a806f0285f73119f04d","msr_publication_uploader":[{"type":"url","title":"https:\/\/arxiv.org\/abs\/2608.28405","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_original_fields_of_study":[],"msr_s2_paper_id":"eeda96a691c6b596f0700a806f0285f73119f04d","msr_s2_pdf_url":"","msr_citation_count_updated":"","msr_citation_count":0,"msr_influential_citations":0,"msr_reference_count":0,"msr_s2_open_access":false,"msr_s2_author_ids":[],"msr_pub_ids":[{"provider":"s2","id":"eeda96a691c6b596f0700a806f0285f73119f04d"},{"provider":"arxiv","id":"2608.28405"},{"provider":"corpusid","id":"291471806"}],"msr_hide_image_in_river":0,"footnotes":""},"msr-research-highlight":[],"research-area":[13545,13554],"msr-publication-type":[270373],"msr-publisher":[],"msr-publication-cta":[],"msr-focus-area":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[246691,255217,260464],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-1185822","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-human-language-technologies","msr-research-area-human-computer-interaction","msr-locale-en_us","msr-field-of-study-computer-science","msr-field-of-study-human-computer-interaction-4","msr-field-of-study-natural-langage-processing"],"msr_publishername":"","msr_edition":"","msr_affiliation":"","msr_published_date":"2026-08-28","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"arXiv","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":0,"msr_main_download":"","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"url","title":"https:\/\/arxiv.org\/abs\/2608.28405","label_id":243109,"id":false,"viewUrl":false}],"msr_related_uploader":[],"msr_citation_count":0,"msr_citation_count_updated":"","msr_s2_paper_id":"eeda96a691c6b596f0700a806f0285f73119f04d","msr_influential_citations":0,"msr_reference_count":0,"msr_arxiv_id":"2608.28405","msr_s2_author_ids":[],"msr_s2_open_access":false,"msr_s2_pdf_url":null,"msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Bryan Chen Zhengyu Tan","user_id":0,"rest_url":false},{"type":"text","value":"Weihua Zheng","user_id":0,"rest_url":false},{"type":"text","value":"Thong T. Doan","user_id":0,"rest_url":false},{"type":"text","value":"Ngoc Bich Doan","user_id":0,"rest_url":false},{"type":"text","value":"Jia Wang Peh","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Xiaoyuan Yi","user_id":40768,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Xiaoyuan Yi"},{"type":"user_nicename","value":"Jing Yao","user_id":41925,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Jing Yao"},{"type":"user_nicename","value":"Xing Xie","user_id":34906,"rest_url":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Xing Xie"},{"type":"text","value":"Nancy F. Chen","user_id":0,"rest_url":false},{"type":"text","value":"Zhengyuan Liu","user_id":0,"rest_url":false},{"type":"text","value":"Jinyeong Bak","user_id":0,"rest_url":false},{"type":"text","value":"Wafi Shamdi","user_id":0,"rest_url":false},{"type":"text","value":"Soo-Kai Chie","user_id":0,"rest_url":false},{"type":"text","value":"Liew Yu Siong","user_id":0,"rest_url":false},{"type":"text","value":"Aina Azyyati Binti Mohamad Rezal","user_id":0,"rest_url":false},{"type":"text","value":"Lew Yan Yan Vanessa","user_id":0,"rest_url":false},{"type":"text","value":"Huarui Wu","user_id":0,"rest_url":false},{"type":"text","value":"Dylan Raharja","user_id":0,"rest_url":false},{"type":"text","value":"Nadya Yuki Wangsajaya","user_id":0,"rest_url":false},{"type":"text","value":"Akane Fukushige","user_id":0,"rest_url":false},{"type":"text","value":"Kazushi Kato","user_id":0,"rest_url":false},{"type":"text","value":"K. Inoue","user_id":0,"rest_url":false},{"type":"text","value":"Tatsuya Kawahara","user_id":0,"rest_url":false},{"type":"text","value":"J. Seo","user_id":0,"rest_url":false},{"type":"text","value":"Dongjun Kim","user_id":0,"rest_url":false},{"type":"text","value":"Seungyoon Lee","user_id":0,"rest_url":false},{"type":"text","value":"Z. Pang","user_id":0,"rest_url":false},{"type":"text","value":"Rui Yang Tan","user_id":0,"rest_url":false},{"type":"text","value":"Charibeth Cheng","user_id":0,"rest_url":false},{"type":"text","value":"M. R. J. Estuar","user_id":0,"rest_url":false},{"type":"text","value":"J. Montalan","user_id":0,"rest_url":false},{"type":"text","value":"Duc Minh Pham","user_id":0,"rest_url":false},{"type":"text","value":"Roy Ka-Wei Lee","user_id":0,"rest_url":false}],"msr_impact_theme":[],"msr_research_lab":[],"msr_event":[],"msr_group":[],"msr_project":[],"publication":[],"video":[],"msr-tool":[],"msr_publication_type":"misc","related_content":[],"_links":{"self":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1185822","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1185822\/revisions"}],"predecessor-version":[{"id":1188009,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/1185822\/revisions\/1188009"}],"wp:attachment":[{"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/media?parent=1185822"}],"wp:term":[{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=1185822"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=1185822"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=1185822"},{"taxonomy":"msr-publisher","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publisher?post=1185822"},{"taxonomy":"msr-publication-cta","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-cta?post=1185822"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=1185822"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=1185822"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=1185822"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=1185822"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=1185822"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=1185822"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=1185822"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/cm-edgetun.pages.dev\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=1185822"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}