Improving Information Extraction by Acquiring External Evidence with Reinforcement Learning

TL;DR

This paper presents a reinforcement learning approach to improve information extraction by acquiring external evidence, especially in data-scarce domains, leading to significant accuracy improvements over traditional methods.

Contribution

The work introduces a novel reinforcement learning framework with a deep Q-network for selecting actions to gather external evidence, enhancing extraction accuracy in low-data scenarios.

Findings

System outperforms traditional extractors

Achieves higher accuracy with external evidence

Effective in domains with limited training data

Abstract

Most successful information extraction systems operate with access to a large collection of documents. In this work, we explore the task of acquiring and incorporating external evidence to improve extraction accuracy in domains where the amount of training data is scarce. This process entails issuing search queries, extraction from new sources and reconciliation of extracted values, which are repeated until sufficient evidence is collected. We approach the problem using a reinforcement learning framework where our model learns to select optimal actions based on contextual information. We employ a deep Q-network, trained to optimize a reward function that reflects extraction accuracy while penalizing extra effort. Our experiments on two databases -- of shooting incidents, and food adulteration cases -- demonstrate that our system significantly outperforms traditional extractors and a competitive meta-classifier baseline.