Reproduce “User-Centric Conversational Recommendation with Multi-Aspect User Modeling” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Shuokai Li, Ruobing Xie, Yongchun Zhu, Xiang Ao, Fuzhen Zhuang, Qing He
The User-Centric Conversational Recommendation (UCCR) model enhances conversational recommender systems by integrating comprehensive user preferences derived from current dialogue, historical sessions, and look-alike users through multi-view modeling. This approach significantly improves both recommendation accuracy and dialogue generation quality, particularly in cold-start scenarios.
Key insights
- A holistic "user understanding" is crucial for CRS, moving beyond just "dialogue understanding" to leverage a user's entire interaction history and social influences.
- Systematically incorporating historical sessions and look-alike users significantly improves CRS robustness, especially in cold-start scenarios, by providing valuable supplementary information.
- Learning intrinsic correlations between different preference views (words, entities, items) through self-supervised tasks enhances the robustness and completeness of user representations.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.