Relation Inference and Type Identification Based on Brain Knowledge Graph

Large-scale brain knowledge bases, such as Linked Brain Data, integrate and synthesize domain knowledge on the brain from various data sources. Although it is designed to provide comprehensive understanding of the brain from multiple perspectives and multi-scale, the correctness and specificity of the extracted knowledge is very important. In this paper, we propose a framework of relation inference and relation type identification to solve the upper problem. Firstly, we propose a quadrilateral closure method based on the network topology to verify and infer the binary relations. Secondly, we learn a model based on artificial neural network to predict the potential relations. Finally, we propose a model free method to identify the specific type of relations based on dependency parsing. We test our verified relations on the annotated data, and the result demonstrates a promising performance.