Salvato in:
Dettagli Bibliografici
Autori principali: Chang, Yu-Cheng, Yeo, Guan-Wei, Eugene, Quah, Shih, Fan-Jie, Kuo, Yuan-Ching, Yu, Tsung-En, Hsu, Hung-Chun, Tsai, Ming-Feng, Wang, Chuan-Ju
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2509.15588
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
Sommario:
  • 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.