Reproduce “Adaptive Personalized Conversational Information Retrieval” 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
Fengran Mo, Yuchen Hui, Yuxing Tian, Zhaoxuan Tan, Chuan Meng, Zhan Su, Kaiyu Huang, Jian-Yun Nie
The Adaptive Personalized Conversational Information Retrieval (APCIR) framework dynamically assesses and applies personalization in conversational search, moving beyond uniform strategies. It achieves superior retrieval effectiveness on TREC iKAT datasets by explicitly identifying personalization levels and adaptively fusing results.
Key insights
- Explicitly identifying the personalization level for each query turn, rather than applying a uniform strategy, is crucial for improving personalized conversational information retrieval performance.
- Guiding Large Language Models with Chain-of-Thought reasoning and in-context examples significantly enhances their ability to discern personalization needs and reformulate queries effectively.
- Dynamically adjusting the influence of personalized and non-personalized query results through adaptive ranking fusion effectively mitigates the negative effects of over-personalization while leveraging relevant user information.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.