CFDA & CLIP at TREC iKAT 2025: Enhancing Personalized Conversational Search via Query Reformulation and Rank Fusion
This work addresses personalized conversational search for information retrieval systems, but it is incremental as it applies existing methods like RRF to new tasks.
The paper tackled the TREC iKAT 2025 tasks by using query rewriting and retrieval fusion strategies, such as Best-of-N selection and Reciprocal Rank Fusion, to enhance personalized conversational search. Results indicated that reranking and fusion improved robustness but revealed trade-offs between effectiveness and efficiency across interactive and offline tasks.
The 2025 TREC Interactive Knowledge Assistance Track (iKAT) featured both interactive and offline submission tasks. The former requires systems to operate under real-time constraints, making robustness and efficiency as important as accuracy, while the latter enables controlled evaluation of passage ranking and response generation with pre-defined datasets. To address this, we explored query rewriting and retrieval fusion as core strategies. We built our pipelines around Best-of-$N$ selection and Reciprocal Rank Fusion (RRF) strategies to handle different submission tasks. Results show that reranking and fusion improve robustness while revealing trade-offs between effectiveness and efficiency across both tasks.