Research on Sentiment Dictionary Based on Sentiment Analysis in News Domain

The known Chinese sentiment dictionary has universal applicability, lacks the expansion of proprietary domains, and lacks support for network words. This paper proposes a new approach and applies it to the Internet news sentiment analysis text provided by the CCF Big Data & Computing Intelligence Contest. First, the corpus is segmented in Chinese to obtain candidate sentiment words that have not appeared in the sentiment dictionary. The candidate sentiment words are calculated in different sentimental texts. The proportion of appearance is obtained, and the emotional tendency of candidate emotional words is obtained, and then it is merged with the emotional vocabulary ontology database to obtain the expanded emotional dictionary. Experiments show that the expanded sentiment dictionary is better than the known Chinese sentiment vocabulary ontology database in the above corpus.