Rewriting Conversational Utterances with Instructed Large Language Models

被引:1
作者
Galimzhanova, Elnara [1 ]
Muntean, Cristina Ioana [2 ]
Nardini, Franco Maria [2 ]
Perego, Raffaele [2 ]
Rocchietti, Guido [2 ]
机构
[1] Univ Pisa, Pisa, Italy
[2] ISTI CNR, Pisa, Italy
来源
2023 IEEE INTERNATIONAL CONFERENCE ON WEB INTELLIGENCE AND INTELLIGENT AGENT TECHNOLOGY, WI-IAT | 2023年
关键词
conversational systems; query rewriting; LLMs; ChatGPT; information retrieval;
D O I
10.1109/WI-IAT59888.2023.00014
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Many recent studies have shown the ability of large language models (LLMs) to achieve state-of-the-art performance on many NLP tasks, such as question answering, text summarization, coding, and translation. In some cases, the results provided by LLMs are on par with those of human experts. These models' most disruptive innovation is their ability to perform tasks via zero-shot or few-shot prompting. This capability has been successfully exploited to train instructed LLMs, where reinforcement learning with human feedback is used to guide the model to follow the user's requests directly. In this paper, we investigate the ability of instructed LLMs to improve conversational search effectiveness by rewriting user questions in a conversational setting. We study which prompts provide the most informative rewritten utterances that lead to the best retrieval performance. Reproducible experiments are conducted on publicly-available TREC CAST datasets. The results show that rewriting conversational utterances with instructed LLMs achieves significant improvements of up to 25.2% in MRR, 31.7% in Precision@1, 27% in NDCG@3, and 11.5% in Recall@500 over state-of-the-art techniques.
引用
收藏
页码:56 / 63
页数:8
相关论文
共 32 条
  • [1] Harnessing Evolution of Multi-Turn Conversations for Effective Answer Retrieval
    Aliannejadi, Mohammad
    Chakraborty, Manajit
    Rissola, Esteban Andres
    Crestani, Fabio
    [J]. CHIIR'20: PROCEEDINGS OF THE 2020 CONFERENCE ON HUMAN INFORMATION INTERACTION AND RETRIEVAL, 2020, : 33 - 42
  • [2] Probabilistic models of information retrieval based on measuring the divergence from randomness
    Amati, G
    Van Rijsbergen, CJ
    [J]. ACM TRANSACTIONS ON INFORMATION SYSTEMS, 2002, 20 (04) : 357 - 389
  • [3] Brown TB, 2020, ARXIV, DOI DOI 10.48550/ARXIV.2005.14165
  • [4] Dalton J., 2020, TREC '20
  • [5] Dalton J., 2021, TREC 21
  • [6] CAsT-19: A Dataset for Conversational Information Seeking
    Dalton, Jeffrey
    Xiong, Chenyan
    Kumar, Vaibhav
    Callan, Jamie
    [J]. PROCEEDINGS OF THE 43RD INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL (SIGIR '20), 2020, : 1985 - 1988
  • [7] Gao JF, 2022, Arxiv, DOI [arXiv:2201.05176, 10.48550/ARXIV.2201.05176, DOI 10.48550/ARXIV.2201.05176]
  • [8] Hao J., 2022, EMNLP 2022
  • [9] Learning to Rewrite Queries
    He, Yunlong
    Tang, Jiliang
    Ouyang, Hua
    Kang, Changsung
    Yin, Dawei
    Chang, Yi
    [J]. CIKM'16: PROCEEDINGS OF THE 2016 ACM CONFERENCE ON INFORMATION AND KNOWLEDGE MANAGEMENT, 2016, : 1443 - 1452
  • [10] Billion-Scale Similarity Search with GPUs
    Johnson, Jeff
    Douze, Matthijs
    Jegou, Herve
    [J]. IEEE TRANSACTIONS ON BIG DATA, 2021, 7 (03) : 535 - 547