An automatic acquisition of domain knowledge from list-structrued text in Baidu encyclopedia

We propose a novel method which can automatically extract new concepts and semantic relations between concepts, in order to support the domain ontology evolvement. We collect the corpus from a free Chinese encyclopedia called Baidu encyclopedia, which is similar to Wikipedia. We locate lists from the Baidu encyclopedia, and extract domain knowledge from the lists. Further more, we use a knowledge assessor to ensure the validity of extracted knowledge. In the experiments, we make a practical attempt to evolve the Chinese Law Ontology (CLO V0), and show that our method can improve the completeness and coverage of CLO V0.