Few-Shot Conversational Dense Retrieval

被引:54
作者
Yu, Shi [1 ]
Liu, Zhenghao [1 ]
Xiong, Chenyan [2 ]
Feng, Tao [1 ]
Liu, Zhiyuan [1 ]
机构
[1] Tsinghua Univ, Beijing, Peoples R China
[2] Microsoft Res, Beijing, Peoples R China
来源
SIGIR '21 - PROCEEDINGS OF THE 44TH INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL | 2021年
关键词
Conversational Retrieval; Dense Retrieval; Knowledge Distillation;
D O I
10.1145/3404835.3462856
中图分类号
TP [自动化技术、计算机技术];
学科分类号
0812 ;
摘要
Dense retrieval (DR) has the potential to resolve the query understanding challenge in conversational search by matching in the learned embedding space. However, this adaptation is challenging due to DR models' extra needs for supervision signals and the longtail nature of conversational search. In this paper, we present a Conversational Dense Retrieval system, ConvDR, that learns contextualized embeddings for multi-turn conversational queries and retrieves documents solely using embedding dot products. In addition, we grant ConvDR few-shot ability using a teacher-student framework, where we employ an ad hoc dense retriever as the teacher, inherit its document encodings, and learn a student query encoder to mimic the teacher embeddings on oracle reformulated queries. Our experiments on TREC CAsT and OR-QuAC demonstrate ConvDR's effectiveness in both few-shot and fully-supervised settings. It outperforms previous systems that operate in the sparse word space, matches the retrieval accuracy of oracle query reformulations, and is also more efficient thanks to its simplicity. Our analyses reveal that the advantages of ConvDR come from its ability to capture informative context while ignoring the unrelated context in previous conversation rounds. This makes ConvDR more effective as conversations evolve while previous systems may get confused by the increased noise from previous turns.
引用
收藏
页码:829 / 838
页数:10
相关论文
共 48 条
  • [41] Xiong Chenyan, 2020, ARXIV PREPRINT ARXIV
  • [42] Xiong Chenyan, 2020, P SIGIR, P2421
  • [43] Xiong Lee, P ICLR
  • [44] Yang J.-H., 2019, TREC
  • [45] Anserini: Enabling the Use of Lucene for Information Retrieval Research
    Yang, Peilin
    Fang, Hui
    Lin, Jimmy
    [J]. SIGIR'17: PROCEEDINGS OF THE 40TH INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL, 2017, : 1253 - 1256
  • [46] Yu Shi, 2020, ARXIV PREPRINT ARXIV, P1933
  • [47] Generating Clarifying Questions for Information Retrieval
    Zamani, Hamed
    Dumais, Susan
    Craswell, Nick
    Bennett, Paul
    Lueck, Gord
    [J]. WEB CONFERENCE 2020: PROCEEDINGS OF THE WORLD WIDE WEB CONFERENCE (WWW 2020), 2020, : 418 - 428
  • [48] Zhan Jingtao, 2020, CoRR