Improving distant supervision using inference learning

Abstract

Distant supervision is a widely applied ap-proach to automatic training of relation extraction systems and has the advantage that it can generate large amounts of la-belled data with minimal effort. How-ever, this data may contain errors and consequently systems trained using dis-tant supervision tend not to perform as well as those based on manually labelled data. This work proposes a novel method for detecting potential false negative train-ing examples using a knowledge inference method. Results show that our approach improves the performance of relation ex-traction systems trained using distantly su-pervised data.