NORMY: Non-Uniform History Modeling for Open Retrieval Conversational Question Answering

被引:0
|
作者
Rashid, Muhammad Shihab [1 ]
Meem, Jannat Ara [1 ]
Hristidis, Vagelis [1 ]
机构
[1] Univ Calif Riverside, Comp Sci & Engn, Riverside, CA 92521 USA
关键词
question answering; history modeling; conversational; retriever; reranker; reader;
D O I
10.1109/ICSC59802.2024.00022
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Open Retrieval Conversational Question Answering (OrConvQA) answers a question given a conversation as context and a document collection. A typical OrConvQA pipeline consists of three modules: a Retriever to retrieve relevant documents from the collection, a Reranker to rerank them given the question and the context, and a Reader to extract an answer span. The conversational turns can provide valuable context to answer the final query. State-of-the-art OrConvQA systems use the same history modeling for all three modules of the pipeline. We hypothesize this as suboptimal. Specifically, we argue that a broader context is needed in the first modules of the pipeline to not miss relevant documents, while a narrower context is needed in the last modules to identify the exact answer span. We propose NORMY, the first unsupervised non-uniform history modeling pipeline which generates the best conversational history for each module. We further propose a novel Retriever for NORMY, which employs keyphrase extraction on the conversation history, and leverages passages retrieved in previous turns as additional context. We also created a new dataset for OrConvQA, by expanding the doc2dial dataset. We implemented various state-of-the-art history modeling techniques and comprehensively evaluated them separately for each module of the pipeline on three datasets: OR-QUAC, our doc2dial extension, and ConvMix. Our extensive experiments show that NORMY outperforms the state-of-the-art in the individual modules and in the end-to-end system.
引用
收藏
页码:101 / 109
页数:9
相关论文
共 50 条
  • [1] Open-Retrieval Conversational Question Answering
    Qu, Chen
    Yang, Liu
    Chen, Cen
    Qiu, Minghui
    Croft, W. Bruce
    Iyyer, Mohit
    [J]. PROCEEDINGS OF THE 43RD INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL (SIGIR '20), 2020, : 539 - 548
  • [2] Generative retrieval for conversational question answering
    Li, Yongqi
    Yang, Nan
    Wang, Liang
    Wei, Furu
    Li, Wenjie
    [J]. INFORMATION PROCESSING & MANAGEMENT, 2023, 60 (05)
  • [3] Attentive History Selection for Conversational Question Answering
    Qu, Chen
    Yang, Liu
    Qiu, Minghui
    Zhang, Yongfeng
    Chen, Cen
    Croft, W. Bruce
    Iyyer, Mohit
    [J]. PROCEEDINGS OF THE 28TH ACM INTERNATIONAL CONFERENCE ON INFORMATION & KNOWLEDGE MANAGEMENT (CIKM '19), 2019, : 1391 - 1400
  • [4] Reinforced History Backtracking for Conversational Question Answering
    Qiu, Minghui
    Huang, Xinjing
    Chen, Cen
    Ji, Feng
    Qu, Chen
    Wei, Wei
    Huang, Jun
    Zhang, Yin
    [J]. THIRTY-FIFTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THIRTY-THIRD CONFERENCE ON INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE AND THE ELEVENTH SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2021, 35 : 13718 - 13726
  • [5] Explicit History Selection for Conversational Question Answering
    Zhang Zhiyuan
    Feng Qiaoqiao
    Wang Yujie
    [J]. 2022 IEEE 34TH INTERNATIONAL CONFERENCE ON TOOLS WITH ARTIFICIAL INTELLIGENCE, ICTAI, 2022, : 1405 - 1410
  • [6] BERT with History Answer Embedding for Conversational Question Answering
    Qu, Chen
    Yang, Liu
    Qiu, Minghui
    Croft, W. Bruce
    Zhang, Yongfeng
    Iyyer, Mohit
    [J]. PROCEEDINGS OF THE 42ND INTERNATIONAL ACM SIGIR CONFERENCE ON RESEARCH AND DEVELOPMENT IN INFORMATION RETRIEVAL (SIGIR '19), 2019, : 1133 - 1136
  • [7] Open-Domain Question Answering Goes Conversational via Question Rewriting
    Anantha, Raviteja
    Vakulenko, Svitlana
    Tu, Zhucheng
    Longpre, Shayne
    Pulman, Stephen
    Chappidi, Srinivas
    [J]. 2021 CONFERENCE OF THE NORTH AMERICAN CHAPTER OF THE ASSOCIATION FOR COMPUTATIONAL LINGUISTICS: HUMAN LANGUAGE TECHNOLOGIES (NAACL-HLT 2021), 2021, : 520 - 534
  • [8] Conversational AI : Open Domain Question Answering and Commonsense Reasoning
    Basu, Kinjal
    [J]. ELECTRONIC PROCEEDINGS IN THEORETICAL COMPUTER SCIENCE, 2019, (306): : 396 - 402
  • [9] Concession-First Learning and Coarse-to-Fine Retrieval for Open-Domain Conversational Question Answering
    Li, Xibo
    Zou, Bowei
    Dong, Mengxing
    Yao, Jianmin
    Hong, Yu
    [J]. 2022 IEEE 34TH INTERNATIONAL CONFERENCE ON TOOLS WITH ARTIFICIAL INTELLIGENCE, ICTAI, 2022, : 317 - 324
  • [10] Knowledge Graph Enabled Open-Domain Conversational Question Answering
    Oduro-Afriyie, Joel
    Jamil, Hasan
    [J]. FLEXIBLE QUERY ANSWERING SYSTEMS, FQAS 2023, 2023, 14113 : 63 - 76