Automatic Acquiring of Chinese Semantic Relations Based on Knowledge Corpus

摘要

Semantic relations play an important role in natural semantic understanding. The sentiment Semantic Dictionary is the basis of sentiment analysis research. This paper uses Modern Chinese Dictionary and baidu interpretation from the Internet as the knowledge corpus. By analyzing the knowledge structure of the corpus, we define different rules to automatically extract the semantic relationship of emotional words. The concept of fine - grained division of emotional intensity and semantic similarity is introduced. The emotional similarity is divided into four levels based on different rules. The dictionary of negative words and degree adverbs are constructed and the degree of adverbs is divided into different levels. The degree of adverbs and negative words in front of the emotional words are used to determine the strength of the emotional words. The method of random selection and manual verification is used for evaluation. The accuracy of synonymous relationship is over 90%, and the availability is high. There is almost no error in the emotional strength relationship, but due to rule restrictions, there are few words that extract intensity levels.