Bias modeling for distantly supervised relation extraction
Summary: Distant supervision (DS) automatically annotates free text with relation mentions from existing knowledge bases (KBs), providing a way to alleviate the problem of insufficient training data for relation extraction in natural language processing (NLP). However, the heuristic annotation process does not guarantee the correctness of the generated labels, promoting a hot research issue on how to efficiently make use of the noisy training data. In this paper, we model two types of biases to reduce noise: (1) \textit{bias-dist} to model the relative distance between points (instances) and classes (relation centers); (2) \textit{bias-reward} to model the possibility of each heuristically generated label being incorrect. Based on the biases, we propose three noise tolerant models: \textit{MIML-dist}, \textit{MIML-dist-classify}, and \textit{MIML-reward}, building on top of a state-of-the-art distantly supervised learning algorithm. Experimental evaluations compared with three landmark methods on the KBP dataset validate the effectiveness of the proposed methods.
- Relation extraction via distant supervision technology
- Extending training set in distant supervision by ontology for relation extraction
- Syntax-aware entity representations for neural relation extraction
- Encoding implicit relation requirements for relation extraction: a joint inference approach
- 10.1162/153244303322533205
This page was built for publication: Bias modeling for distantly supervised relation extraction
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1667103)
