Research Of Text Mining About Semantic Relation Recognition

Text Mining also known as text data mining, or knowledge discovery from textual databases, refers to obtain interesting or useful patterns for users from the unstructured text information. The generally recognized text mining is defined as follows:text mining is the process of extracting the unknown, understandable and available knowledge ultimately from a large number of text data, meanwhile uses this knowledge to organize information better for future reference.Text mining about semantic relation discovery is a research focus; the main idea is to discovery terms and the semantic relationships between the terms with scanning the text of natural language and automated processing. The various semantic relations between concepts and language units is important form of knowledge, these semantic relations mainly are hypernymy/hyponymy, part-whole, causality, synonymy, antonymy, inference and so on. From the theorectical point, the semantic relations discovery in text mining research will make natural language processing from the lexical analysis, syntactic analysis deep into the senamtic analysis; from the application level, the senamtic relations discovery in text mining will provide the theory and method for automatic or semi-automatic ontology construnction.Based on the military aircraft corpus, study the military aircraft domain concepts hierarchy semantic relations discovery, use text mining processing and basic process, combined natural language processing, information extraction, ontology automatic construction theory and related research method, based on semantic relations text mining conduct the research, explore and experiments. The main work and research include the following:(1) Theory and research of semantic relations discovery in text mining. This article reviews the the natural language processing, text mining, automatic construction of ontology research aspects have been summarized.(2) The construction of the military aircraft text processing corpus. Retrive the articles related to military aircraft articles in the Wikipedia and CNKI database. The military aircraft, including 1951 terms corpus,304 articles, in which the phrase extracted 3324. The corpus is the basic of the experiment, and it also provides materials and support for the ontology automatic construction, or other relevant work.(3) The research and experiment of template matching for semantic discovery. According to the characteristics of military aircraft, propose specific semantic relationships. Training some corpus, thus obtained corresponds with the semantic relationship between the types of template. Use the edit distance has been summarized the relationship between the template matching, and the corpus for testing to verify the effectiveness of the method.(4) The reaserch and experiment based on complex network. Use the natural language with the network characteristics to discover the semantic relationship. Use terms and words which have relation with terms as the node of the net to construct complex network. In this complex network, every community represents a relation. Next, use terms as node, use relation has been found as edge, construct the complex network of military aircraft domain concepts hierarchy, and analysis it.