Sentiment Classification of Russian Texts Using Automatically Generated Thesaurus

This paper is devoted to an approach for sentiment classification of Russian texts applying an automatic thesaurus of the subject area. This approach consists of a standard machine learning classifier and a procedure embedded into it, that uses thesaurus relationships for better sentiment analysis. The thesaurus is generated fully automatically and does not require expert’s involvement into classification process. Experiments conducted with the approach and four Russian-language text corpora, show effectiveness of thesaurus application to sentiment classification.