Relevance-guided Supervision for OpenQA with ColBERT

TL;DR

This paper introduces ColBERT-QA, a fine-grained neural retrieval model for OpenQA that uses relevance-guided supervision to improve retrieval accuracy and achieve state-of-the-art results on multiple datasets.

Contribution

It adapts the ColBERT model for OpenQA and proposes an iterative weak supervision strategy to enhance retrieval performance.

Findings

Achieves state-of-the-art results on Natural Questions, SQuAD, and TriviaQA.

Improves retrieval accuracy over coarse-grained models.

Demonstrates effectiveness of relevance-guided supervision in OpenQA.

Abstract

Systems for Open-Domain Question Answering (OpenQA) generally depend on a retriever for finding candidate passages in a large corpus and a reader for extracting answers from those passages. In much recent work, the retriever is a learned component that uses coarse-grained vector representations of questions and passages. We argue that this modeling choice is insufficiently expressive for dealing with the complexity of natural language questions. To address this, we define ColBERT-QA, which adapts the scalable neural retrieval model ColBERT to OpenQA. ColBERT creates fine-grained interactions between questions and passages. We propose an efficient weak supervision strategy that iteratively uses ColBERT to create its own training data. This greatly improves OpenQA retrieval on Natural Questions, SQuAD, and TriviaQA, and the resulting system attains state-of-the-art extractive OpenQA performance on all three datasets.