Paper: End-to-End Relation Extraction Using Distant Supervision from External Semantic Repositories

ACL ID P11-2048
Title End-to-End Relation Extraction Using Distant Supervision from External Semantic Repositories
Venue Annual Meeting of the Association of Computational Linguistics
Session Main Conference
Year 2011
Authors

In this paper, we extend distant supervision (DS) based on Wikipedia for Relation Extrac- tion (RE) by considering (i) relations defined in external repositories, e.g. YAGO, and (ii) any subset of Wikipedia documents. We show that training data constituted by sentences containing pairs of named entities in target re- lations is enough to produce reliable supervi- sion. Our experiments with state-of-the-art re- lation extraction models, trained on the above data, show a meaningful F1 of 74.29% on a manually annotated test set: this highly im- proves the state-of-art in RE using DS. Addi- tionally, our end-to-end experiments demon- strated that our extractors can be applied to any general text document.