2026-06-26T21:48:17Z https://meral.edu.mm/oai

oai:meral.edu.mm:recid/6270 2021-12-13T05:26:09Z 1582963342780:1605779935331 user-uit

Domain-specific Sentiment Dictionary Construction for Sentiment Classification Aye Aye Mar Nyein Thwet Thwet Aung Su Su Htay Sentiment dictionaries are commonly used to solve the problem of sentiment classification for customer reviews. The number of sentiment words in the generalized dictionaries such as SentiWordNet is limited and lack of many sentiment words especially domain-specific sentiment words. Different domains have different sentiment words and the sentiment of a word depends on the domain in which it is used. In this paper, an approach based on Point-wise Mutual Information (PMI) is proposed to construct a domain-specific sentiment dictionary effectively and automatically. The proposed system is evaluated on three diverse datasets from different domains by using 10-fold cross validation. Accordingly to the experimental results, the goodness of the extracted dictionary is relatively high and significantly improves the performance of sentiment classification. The experimental results show that the extracted domain-specific dictionary outperforms the generalized dictionary, SentiWordNet. The proposed method learns the domain-specific sentiment words efficiently and it is domain adaptable. 2017-11-02 http://hdl.handle.net/20.500.12678/0000006270 https://meral.edu.mm/records/6270