From Bag of Sentences to Document: Distantly Supervised Relation Extraction via Machine Reading Comprehension

TL;DR

This paper introduces a document-based distant supervision approach for relation extraction, reformulating it as a machine reading comprehension task to better utilize evidence and reduce noise, achieving state-of-the-art results.

Contribution

It proposes a novel document-based DS paradigm and a new loss function, DSLoss, to improve relation extraction by leveraging comprehensive evidence and handling noisy labels.

Findings

Achieves state-of-the-art DS relation extraction performance.

Effectively leverages inter-sentence and entity-level evidence.

Reduces noisy label impact through the DSLoss function.

Abstract

Distant supervision (DS) is a promising approach for relation extraction but often suffers from the noisy label problem. Traditional DS methods usually represent an entity pair as a bag of sentences and denoise labels using multi-instance learning techniques. The bag-based paradigm, however, fails to leverage the inter-sentence-level and the entity-level evidence for relation extraction, and their denoising algorithms are often specialized and complicated. In this paper, we propose a new DS paradigm--document-based distant supervision, which models relation extraction as a document-based machine reading comprehension (MRC) task. By re-organizing all sentences about an entity as a document and extracting relations via querying the document with relation-specific questions, the document-based DS paradigm can simultaneously encode and exploit all sentence-level, inter-sentence-level, and entity-level evidence. Furthermore, we design a new loss function--DSLoss (distant supervision loss), which can effectively train MRC models using only $\langle$document, question, answer$\rangle$ tuples, therefore noisy label problem can be inherently resolved. Experiments show that our method achieves new state-of-the-art DS performance.