{"paper":{"title":"Task-Aware Automated User Profile Generation for Recommendation Simulation Using Large Language Models","license":"http://creativecommons.org/licenses/by/4.0/","headline":"APG4RecSim automatically generates realistic user profiles for LLM-based recommender simulations with minimal supervision.","cross_cats":[],"primary_cat":"cs.IR","authors_text":"Chenglong Ma, Danula Hettiachchi, Jeffrey Chan, Xinye Wanyan, Ziqi Xu","submitted_at":"2026-05-13T13:20:39Z","abstract_excerpt":"Large Language Model (LLM)-based agent simulation has emerged as a promising approach to meet the increasing demand for real-time and rigorous evaluation in modern recommender systems. A typical LLM-driven simulation framework comprises three essential components: the profile module, memory module, and action module. However, existing studies have primarily concentrated on enhancing the memory and action modules, with limited attention to profile generation, which plays a pivotal role in ensuring realistic agent behaviours and aligning simulated interactions with real user dynamics. Moreover, "},"claims":{"count":4,"items":[{"kind":"strongest_claim","text":"APG4RecSim achieves the best overall performance on discrimination, ranking, and rating tasks, improving ranking quality by up to 7% in nDCG@10 and reducing rating distribution divergence by 8% in JSD compared to existing profile-generation baselines.","source":"verdict.strongest_claim","status":"machine_extracted","claim_id":"C1","attestation":"unclaimed"},{"kind":"weakest_assumption","text":"That profiles generated by the LLM with minimal supervision accurately capture real user dynamics and produce simulated interactions that align with actual user behavior across datasets.","source":"verdict.weakest_assumption","status":"machine_extracted","claim_id":"C2","attestation":"unclaimed"},{"kind":"one_line_summary","text":"APG4RecSim automatically generates realistic user profiles for LLM-based recommendation simulations, outperforming manual baselines by up to 7% in nDCG@10 and 8% in JSD on three benchmark datasets.","source":"verdict.one_line_summary","status":"machine_extracted","claim_id":"C3","attestation":"unclaimed"},{"kind":"headline","text":"APG4RecSim automatically generates realistic user profiles for LLM-based recommender simulations with minimal supervision.","source":"verdict.pith_extraction.headline","status":"machine_extracted","claim_id":"C4","attestation":"unclaimed"}],"snapshot_sha256":"d25f85a076a745c43c289788c4d48c76b2a74f0f727e17cf5ed8498b3c75fa9e"},"source":{"id":"2605.13497","kind":"arxiv","version":1},"verdict":{"id":"a832b348-b149-46d4-8665-8e71a7f96785","model_set":{"reader":"grok-4.3"},"created_at":"2026-05-14T18:18:02.132887Z","strongest_claim":"APG4RecSim achieves the best overall performance on discrimination, ranking, and rating tasks, improving ranking quality by up to 7% in nDCG@10 and reducing rating distribution divergence by 8% in JSD compared to existing profile-generation baselines.","one_line_summary":"APG4RecSim automatically generates realistic user profiles for LLM-based recommendation simulations, outperforming manual baselines by up to 7% in nDCG@10 and 8% in JSD on three benchmark datasets.","pipeline_version":"pith-pipeline@v0.9.0","weakest_assumption":"That profiles generated by the LLM with minimal supervision accurately capture real user dynamics and produce simulated interactions that align with actual user behavior across datasets.","pith_extraction_headline":"APG4RecSim automatically generates realistic user profiles for LLM-based recommender simulations with minimal supervision."},"references":{"count":51,"sample":[{"doi":"10.1145/3450613","year":2021,"title":"Himan Abdollahpouri, Masoud Mansoury, Robin Burke, Bamshad Mobasher, and Edward Malthouse. 2021. User-centered Evaluation of Popularity Bias in Recom- mender Systems. InProceedings of the 29th ACM Con","work_id":"486bb563-eb99-467e-91da-2ca1040c4284","ref_index":1,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1016/j.knosys.2016.03.006","year":2016,"title":"Mohammad Yahya H. Al-Shamri. 2016. User profiling approaches for demo- graphic recommender systems.Know.-Based Syst.100, C (May 2016), 175–187. doi:10.1016/j.knosys.2016.03.006","work_id":"ead7636a-b726-471a-ab37-7eaa13d48705","ref_index":2,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.1145/3613904.3642081","year":2024,"title":"InProceedings of the CHI Conference on Human Factors in Computing Systems (CHI ’24)","work_id":"de206c3b-7e99-4e52-8b63-f058b0773327","ref_index":3,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"10.18653/v1/2025.acl-","year":2025,"title":"In: Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)","work_id":"1ef694be-4483-4d1c-b319-30be63d024de","ref_index":4,"cited_arxiv_id":"","is_internal_anchor":false},{"doi":"","year":2023,"title":"Nicholas Carlini, Daphne Ippolito, Matthew Jagielski, Katherine Lee, Florian Tramèr, and Chiyuan Zhang. 2023. Quantifying Memorization Across Neu- ral Language Models. InThe Eleventh International Con","work_id":"f616300e-1733-4ee6-8062-63a9356edbc3","ref_index":6,"cited_arxiv_id":"","is_internal_anchor":false}],"resolved_work":51,"snapshot_sha256":"0c857abd8e00f2ba03f631566ca1c566987407b5a1840cfabb1071a69cc678d3","internal_anchors":1},"formal_canon":{"evidence_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"author_claims":{"count":0,"strong_count":0,"snapshot_sha256":"258153158e38e3291e3d48162225fcdb2d5a3ed65a07baac614ab91432fd4f57"},"builder_version":"pith-number-builder-2026-05-17-v1"}