Joint Learning of Entity Semantics and Relation Pattern for Relation Extraction.

Relation extraction is identifying the relationship of two given entities in the text. It is an important step in the task of knowledge extraction, which plays a vital role in automatic construction of knowledge base. When extracting entities' relations from sentences, some keywords can reflect the relation pattern, besides, the semantic properties of given entities can also help to distinguish some confusing relations. Based on the above observations, we propose a mixture convolutional neural network for the task of relation extraction, which can simultaneously learn the semantic properties of entities and the keyword information related to the relation. We conduct experiments on the SemEval-2010 Task 8 dataset. The method we propose achieves the state-of-the-art result without using any external information. Additionally, the experimental results also show that our approach can learn the semantic relationship of the given entities effectively.