Discovering Semantic Relationships for Knowledgebase

Abstract

Discovering the semantic relationships in knowledgebase is critical in information processing and knowledge management. Previous studies of discovering semantic relationships are mainly based on the information extraction using manually annotated training set and predefined semantic relationship patterns. In this paper, we propose a new method to automatically discover the semantic relationships between two concepts in knowledgebase through text classifying and information filtering. The documents related to the two concepts in knowledgebase are at first classified into different taxonomies and the connecting terms capturing the semantic relationships between the two concepts are extracted. The experimental results show that our method has provided an efficient and effective way for the automatic discovering of semantic relationships for information management.