문재완 박사과정, 박성민 석박통합과정 CIKM 2025 국제 학술대회 논문 채택

5 Aug 2025

DIAL 연구실 소속 인공지능학과 문재완(박사과정, 공동 제1저자) 학생, 박성민(석박통합과정, 공동 제1저자), 이종욱(교신저자) 교수가 참여한 논문 "LLM-Enhanced Linear Autoencoders for Recommendation"이 데이터마이닝 분야 최우수 국제 학술대회인 ACM International Conference on Information and Knowledge Management (CIKM 2025, short paper)에 최종 게재가 승인되었으며 오는 11월에 발표될 예정입니다. 

Abstract

Large language models (LLMs) have been widely adopted to enrich the semantic representation of textual item information in recommender systems. However, existing linear autoencoders (LAEs) that incorporate textual information often rely on sparse word co-occurrence patterns, limiting their capacity to capture rich textual semantics. To address this, we propose L³AE, the first integration of LLMs into the LAE framework. L³AE effectively integrates the heterogeneous knowledge of both textual semantics and user-item interactions through a two-phase optimization strategy. (i) L³AE first constructs a semantic-level item correlation matrix from LLM-derived item representations. (ii) It then learns an item-to-item weight matrix from collaborative signals while distilling semantic item correlations as regularization. Notably, each phase of L³AE is optimized through closed-form solutions, which ensures global optimality and computational efficiency. Extensive experiments demonstrate that L³AE consistently outperforms state-of-the-art LLM-enhanced models on three benchmark datasets, achieving gains of 27.6% in Recall@20 and 39.3% in NDCG@20.