Overview of Distant Supervised Relation Extraction

Relation extraction is a fundamental task in natural language processing, aiming at extracting relational triples from plain text. However, there are fewer instances in the manually constructed dataset to meet the learning needs of relation extraction models. Distant supervision approach has attracted the interest of numerous researchers due to its ability to construct large datasets at a low cost. Nevertheless, there are certain problems with distant supervision approach due to overly strong assumptions. In this paper, we introduce three main problems in distant supervised relation extraction: the noise labeling problem, the triple-overlapping problem, and the long-tail data distribution problem. Furthermore, we summarize current solutions and evaluation strategies and then summarize future research challenges.