Automatic Extraction Method of Mongolian Synonymous Compounds

This paper proposes an automatic extraction method of Mongolian synonymous compounds based on multiple resources such as dictionary resources, lexical semantic knowledge base resources, and corpus resources, and conducts experimental research on the basis of them. The experimental results of single methods show that the method based on dictionary resources has the highest accuracy, reaching 95.2%, and the method based on lexical semantic knowledge base resources has the highest recall rate and F-measure, reaching 66.7% and 78.2%. The fusion of the three methods overcomes the deficiency of using only one resource method, effectively improves the performance of the method, and realizes the multi-angle and complementary synonym extraction. Compared with the method of using one resource, the experimental results of the fusion method have been improved to varying degrees, and its accuracy, recall rate and F-measure have reached 86.5%, 98.2% and 92% respectively.