Domain Ontology Terms And The Upper And Lower Parity Relationship Extraction Studies

With the popularity of computer and the rapid development of Internet, the form, source and quantity of also knowledge acquisition subsequently undergone a fundamental change. Extensive network contains a lot of knowledge, but also includes a lot of junk knowledge, using artificial method is to acquire knowledge is far from meeting the requirements, the study various knowledge acquisition methods to reduce artificial workload is an inevitable trend. Thus how from store vast amounts of information sources obtain users want of knowledge and how to gain vast knowledge into what way to conduct the management, sharing and reuse, the research on artificial intelligence have become an important research task.Ontology is currently as an accepted way to solve the knowledge management, sharing and reuse, so as to domain ontology construction is automatic or semi-automatic current information retrieval and construction of knowledge of the main research hotspots, its main task including domain term acquisition, relationship of acquisition, the hierarchy system, the establishment of properties and the attribute values acquisition, examples of access and their subsequent domain ontology maintenance. So domain ontology automatic or semi-automatic constructing neutron task also became the multitudinous scholar research hotspot object. In recent years, along with the natural language processing technology and information extraction technology of rapid development, we can use these technologies to complete domain ontology automatic or semi-automatic construction task.This paper focus on domain term extraction, domain term hyponymic relation extraction and syntopic relation extraction, besides studies and discusses mainly completes the following several aspects of work:(1)Domain term extraction task, for unknown words in the domain of terminology and long domain term words extraction problem,This paper proposed for automatic extracting domain terms from domain corpora based on CRFs, This paper took the field of tourism in YunNan Province as the experimental subjects, furthermore, compared with the traditional method based on Combination of Mutually information and T-Value. Artificial evaluation showed that based on CRFs had a satisfactory results. This method is very good at extracting domain terms both from unknown words and long characters of domain terms. There was a noticeable increase precision and recall of the Term Extraction. Experimental results indicate that it achieves better performance than traditional method.(2)Domain term hyponymic relation extraction and syntopic relation extraction task, currently domain term hyponymic relation extraction and syntopic relation extraction is mainly to use pattern method, the key to this model is the extraction of the pattern. In this paper, based on conditional random fields machine learning method with discrete type way, to obtain the mode, namely through feature selection, artificial marks a certain amount of training corpus constructs relationships classifier model, and this model is used to identify domain term hyponymic relation and syntopic relation. And in YunNan tourism domain experiments prove the effectiveness of this method, on the basis of verified combination feature recognition effect.(3)Design and realize two prototype system:domain term extraction prototype system and domain term hyponymic relation extraction and syntopic relation extraction model, and evaluation two systems.