RoRED: Bootstrapping labeling rule discovery for robust relation extraction

Wenjun Hou, Liang Hong, Haoshuai Xu, Wei Yin

Abstract: Highlights•A bootstrapping labeling rule discovery framework is proposed for robust relation extraction.•PN-rules are defined to represent positive and negative rules that jointly filter incorrectly matched sentences.•Semantic-matching mechanism determines missing sentences based on semantic associations between rules, words, and sentences.•Co-training-based rule verification approach refines matched sentences to improve the overall quality of rule discovery.