Research on an ontology-based language model

Ontology is widely applied to various areas of natural language processing, which regarded as a modeling tool of conceptual model of describing information in the level of semantic and knowledge. In this paper, we put forward an ontology-based language model with the ontological character, which is language-independent and can be used in cross-language information processing. Construct an ontology representation of feature information using a similarity framework; get the usage of feature information in using semantics, pragmatics, syntactic, etc. by specifically machine learning method and express as abstract ontology knowledge of feature information. Processing actual document, the usage of feature information will be got in actual document and the similarity can be evaluated with ontology knowledge. We did three experiments in Chinese, i.e. IR, text similarity computing, specific information identification. The preliminary experiment testifies that recall and precision is advanced.