Knowledge-Empowered Representation Learning for Chinese Medical Reading Comprehension: Task, Model and Resources

TL;DR

This paper introduces a new multi-target Chinese Medical MRC task, a high-quality dataset, and a specialized BERT model that fuses medical knowledge, significantly improving medical reading comprehension accuracy.

Contribution

It presents a novel multi-target MRC task in the medical domain, a high-quality dataset, and a knowledge-fused BERT model tailored for Chinese medical reading comprehension.

Findings

CMedBERT outperforms baseline models in medical MRC tasks.

The dataset enables effective training and evaluation of medical MRC models.

Knowledge fusion improves comprehension accuracy.

Abstract

Machine Reading Comprehension (MRC) aims to extract answers to questions given a passage. It has been widely studied recently, especially in open domains. However, few efforts have been made on closed-domain MRC, mainly due to the lack of large-scale training data. In this paper, we introduce a multi-target MRC task for the medical domain, whose goal is to predict answers to medical questions and the corresponding support sentences from medical information sources simultaneously, in order to ensure the high reliability of medical knowledge serving. A high-quality dataset is manually constructed for the purpose, named Multi-task Chinese Medical MRC dataset (CMedMRC), with detailed analysis conducted. We further propose the Chinese medical BERT model for the task (CMedBERT), which fuses medical knowledge into pre-trained language models by the dynamic fusion mechanism of heterogeneous features and the multi-task learning strategy. Experiments show that CMedBERT consistently outperforms strong baselines by fusing context-aware and knowledge-aware token representations.