Extracting Maximal Frequent Connecting Sequences for Entities

Discovering semantic relationships between entities is a crucial problem for many data analysis work. Most recent studies, however, only focus on extracting predefined semantic instances, and the current semantic relationships representations are also weak. This paper presents a new method for extracting meaningful semantic relationships from unstructured natural language sources. The method is based on the maximal frequent connecting sequences extracted from the contexts of entities. For identifying the semantic relationships of entities, connecting terms are found out and used as the seeds to discover the maximal frequent connecting sequences. Experimental results show the effectiveness of our methods.