Discovery of Dependency Tree Patterns for Relation Extraction
Hongzhi Xu, Changjian Hu, Guoyang Shen · 2009
Abstract. Relation extraction is to identify the relations between pairs of named entities. In this paper, we try to solve the problem of relation extraction by discovering dependency tree patterns (a pattern is an embedded sub dependency tree indicating a relation instance). Our approach is to find an optimal rule (pattern) set automatically based on the proposed dependency tree pattern mining algorithm. The experimental results show that the extracted patterns can achieve a high precision and a reasonable recall rate when used as rules to extract relation instances. Furthermore, an additional experiment shows that other machine learning based relation extraction methods can also benefit from the extracted patterns by using them as features. 1