Few-Shot Conversational Dense Retrieval

Abstract

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 long-tail 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. Our code is publicly available at https://github.com/thunlp/ConvDR.

arXiv commentAccepted by SIGIR 2021

References in corpus (3)

• Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer
• Query Resolution for Conversational Search with Limited Supervision
• Open-Retrieval Conversational Question Answering

Cited by in corpus (14)

• Information Retrieval: Recent Advances and Beyond
• Retrieving Supporting Evidence for Generative Question Answering
• Query Performance Prediction: From Ad-hoc to Conversational Search
• Pre-train a Discriminative Text Encoder for Dense Retrieval via Contrastive Span Prediction
• Exploiting Simulated User Feedback for Conversational Search: Ranking, Rewriting, and Beyond
• Query Performance Prediction using Relevance Judgments Generated by Large Language Models
• Question rewriting? Assessing its importance for conversational question answering
• Bridging the Gap: From Ad-hoc to Proactive Search in Conversations
• ProCIS: A Benchmark for Proactive Retrieval in Conversations
• Beyond Single Items: Exploring User Preferences in Item Sets with the Conversational Playlist Curation Dataset
• ZeQR: Zero-shot Query Reformulation for Conversational Search
• Rewriting Conversational Utterances with Instructed Large Language Models
• Improving Conversational Passage Re-ranking with View Ensemble
• Towards Self-Contained Answers: Entity-Based Answer Rewriting in Conversational Search