Research On The Homonyms Disambiguation Algorithm Based On Mongolian Nouns Semantic Network

With the implement of national western development strategy, the development of minority areas of high-tech enterprise has been referred to the agenda of the state departments concerned, the Mongolian language information processing is no exception, in comparison with the study of English and Chinese semantics, information processing oriented Mongolian semantic research starts late, many aspects are in the early exploration stage. At present most of references and research in terms of ambiguity have been focused on Chinese and English, while in the aspect of the Mongolian, although gradual research results have been achieved, But compared to the research achievements in the aspects of English and Chinese, it still has some gaps. But with the constant exploration and innovation of the Mongolian semantic knowledge in recent years, The emerging scientific research has accelerated the pace of Mongolian information processing, Mongolian words automatically eliminate problems as the difficulties of Mongolian information processing, good scientific research achievements have been obtained in the continuous exploration and innovation under the scholars, And the research content of this paper makes full use of the predecessors of research results and research experience, and then puts forward an algorithm of eliminating the homograph ambiguity of semantic web based on Mongolian nouns, and the experiment was carried out to verify the algorithm effectively.Thirty years of Mongolian information processing practice study shows that ambiguity resolution problem has involves the vocabulary grammar and semantic properties. Mongolian homograph ambiguity problem is the fundamental key research topics in the field of Mongolian information processing, Mongolian homograph disambiguation links as an intermediate task is the key to solve the problem of machine translation, text classification, speech recognition and information retrieval etc. According to incomplete statistics, Homograph occupy 5.1% of the proportion of the dictionary, while the proportion accounted for in real corpus up to 11.6%, so the correctly identify and label of the Homograph in the dictionary and corpus will be directly related to the Mongolian syntax, semantics analysis of the results.An important application of the semantic web is vocabulary ambiguity resolution, in both English and Chinese lexical semantic web are related to this research. The main purpose of the construction of Mongolian semantic knowledge base is to provide service for various applications, Mongolian noun semantic network has been completed is no exception. In order to make full use of the semantic knowledge of the semantic dictionary, this study will be committed to the use of Mongolian noun semantic network combined with existing Mongolian lexical resources to complete the homograph disambiguation. With reference to the existing homograph disambiguation algorithm and then proposing a new algorithm of eliminating ambiguity based on the lower semantic tree of Mongolian noun semantic web, and through the manual and automatic annotation does corpus tagging experiments for the proposed algorithm. The whole experiment was carried out in a sentence as a unit, first of all, the statistical method is adopted in the mathematical model using mutual information to extract the collocation directives which exists in the same sentence with Mongolian homograph, and then through Mongolian noun semantic similarity calculation method to calculate semantic similarity between collocation directives and collocation directives in homograph collocation Thesaurus, selecting the average similarity value corresponding to the maximum of the collocation of homograph meanings as the correct meaning, and tagging in corpus. The experimental results show that this method is feasible, and achieved 76% accuracy.