Research On Ontology Learning Methods For Text

Ontology, can be described on the conceptual model of information systems on semantic and knowledge levels, and has been widely used in semantic web, knowledge engineering, artificial intelligence, information retrieval, machine translation and digital libraries and other fields. The application of ontology is expanding; people of ontology needs are also growing. However, the application of ontology in most areas which are built by hand in the ontology has been implemented later. Ontology construction tools can only provide its editing, with the use of these ontology editing tools, users still need manual, which A large level of knowledge has hampered access to, which made the body can not keep updated. To determine the body can not be built manually, the user pressing needs semi-automatic or automatic way to build body. Therefore, in order to meet the needs of the user ontology learning techniques than we have, its purpose is to achieve automatic and semi-automatic ontology construction.The main task of ontology learning is the concept extraction and the relationship between accesses to justice. According to different types of data, it is necessary to adopt different ontology study technology. This paper is based on ontology points-unstructured data of learning technologies for research.This paper introduces the definition of ontology, ontology language and ontology of the type described, ontology learning methods and theoretical research, a brief introduction of the ontology learning architecture, the type of ontology learning, ontology learning tool and ontology study method.The text focuses on the ontology learning for concept extraction and concept in relation extraction. In this paper, the concept of extraction on the basis of statistical methods based on the introduction of the seeds of the concept, introduces concepts extracted steps, and makes the corresponding experimental verification, greatly improving the accuracy of the concepts extracted. For the extraction of concept relationship, this paper mainly based on statistics and pattern of two methods which are studied, for statistical methods are mainly through the concept of information degree and concept semantic similarity between conceptual relationships, to extract. Statistical method and mode method both of which are separately carried on the experiment and two methods mix rise done the experiment. Finally, from the overall consideration, it is to be tested whether the size of the data source of influence on the experimental results.