Research on Ontology Instance Learning Based on Maximum Entropy Model

Ontology Instance learning is a significant part in the research of ontology evolution and is important for the applications of ontology. One of the key points of ontology instance learning is automatic instances learning from large amounts of unstructured data. In this paper, an effective method of ontology instance learning is proposed, while the experience of the methods of information extraction and the structure of ontology model--introducing maximum entropy model to the study of learning ontology instances from free texts--are also focused. The experiment is based on the Chinese Toponym Ontology. Results show that rapid and effective ontology instance learning is available.