Dense Hierarchical Retrieval for Open-domain Question Answering
Плотный иерархический поиск для ответов на вопросы в открытом домене
2021-01-01
SCID: 54.1/jv2wmhbf
Discuss with AI
dense hierarchical retrievaldense passage retrievaldocument-level retrievernegative samplingopen-domain question answering
Figures from the paper
Abstract (AI)
Dense neural text retrieval has achieved promising results on open-domain Question Answering (QA), where latent representations of questions and passages are exploited for maximum inner product search in the retrieval process.However, current dense retrievers require splitting documents into short passages that usually contain local, partial and sometimes biased context, and highly depend on the splitting process.As a consequence, it may yield inaccurate and misleading hidden representations, thus deteriorating the final retrieval result.In this work, we propose Dense Hierarchical Retrieval (DHR), a hierarchical framework which can generate accurate dense representations of passages by utilizing both macroscopic semantics in the document and microscopic semantics specific to each passage.Specifically, a document-level retriever first identifies relevant documents, among which relevant passages are then retrieved by a passage-level retriever.The ranking of the retrieved passages will be further calibrated by examining the document-level relevance.In addition, hierarchical title structure and two negative sampling strategies (i.e., In-Doc and In-Sec negatives) are investigated.We apply DHR to large-scale open-domain QA datasets.DHR significantly outperforms the original dense passage retriever, and helps an end-to-end QA system outperform the strong baselines on multiple open-domain QA benchmarks.
Key Findings
1
DHR uses a two-stage hierarchy: document retrieval identifies relevant documents, followed by passage retrieval and document-level relevance calibration.
2
Dense Hierarchical Retrieval (DHR) improves passage representations by combining document-level semantics with passage-specific local semantics.
3
Integrating DHR into an end-to-end QA system surpasses strong baselines across multiple open-domain QA benchmarks.
4
On large-scale open-domain QA datasets, DHR significantly outperforms the original dense passage retriever.
5
The method investigates hierarchical title structures and two negative-sampling strategies, In-Doc and In-Sec negatives.
Research Object
open-domain question answering retrieval, specifically documents and passages used for answering questions
Research Subject
hierarchical dense representation and relevance ranking of passages using document-level and passage-level semantics
Publication Details
Publication Date
2021-01-01
Journal
Publisher
ISSN
Cited by
23
Open access PDF
Access Type
Author Information
Download PDF
Subscribe to digest
References available in scid.ai7
Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformer2019
Dense Passage Retrieval for Open-Domain Question Answering2020
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks2020
Leveraging Passage Retrieval with Generative Models for Open Domain Question Answering2021
REALM: Retrieval-Augmented Language Model Pre-Training2020
Answering Complex Open-Domain Questions with Multi-Hop Dense Retrieval2020
Qualitative Spatial Reasoning over Questions (Short Paper)2022