Entity Recognition and Relations Extraction Based on the Structure of Online Encyclopedia

In order to construct the knowledge base in the field of journalism, this paper improved the knowledge representation framework of Freebase to make it more suitable for the knowledge in journalism domain. On the basis, we chose to extract entities, entity attributes and relationships between entities from Baidu Encyclopedia websites in order to analyze its structure. According to the infobox, the Character Relationship, the Relevant Characters and the Category Labels templates on the Baidu Encyclopedia webpages, we harvested entity triples (Entity1, Relation, Entity2). Then, we supplemented the characters relationship types through the Entity Relation template in Hudong Encyclopedia webpages. Through the natural language processing technology, the entity similarity algorithm, the association rules reasoning algorithm and other methods, we cleaned and supplemented the knowledge. After that, we stored the knowledge to the graph database according to the knowledge representation model. Finally, we got a better knowledge base in the field of journalism.