Ontology Learning Based On Unstructured Chinese Data Sources

In the recent years, ontology is playing a more and more important role in knowledge management and the semantic web. But to build the ontology by hand is a complicated work. It has become an important subject to acquire Chinese Ontology from corpus of different fields in order to extend and implement semantic web in China.The ontology learning aims at building ontology automatically or semi-automatically by the use of artificial intelligence, natural language processing and many other area of science. The ontology learning refers to the extraction of learning content and uses this content to construct ontology. The main task of ontology learning consists of automatic or semi-automatic acquisition of every element contained in ontology, such as concept learning and relation learning.Based on the existing ontology learning theory, methods and techniques from abroad, combining the research fruits of Chinese natural language processing field, two algorithms are raised in this paper on the concept extraction and relation extraction. On the concept side, a method comes up with based on the idea "different construction of concept on different field". And sift concepts by domain relevance and domain consensus. On the relation extraction side, relations are abstracted by the way of sentence pattern. In-depth study was done in the sentence patter and divided it into single-sentence pattern,multi-sentence pattern and segment pattern. Some algorithms for semi-automatic generating pattern are proposed here. Finally, Chinese ontology learning system was realized to test the feasibility of the methods proposed.