An Investigation of Effectiveness and Possibility to Automatic Construction of Estimation Patterns Using Customer Reviews
2016 5th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI) (2016)
July 10, 2016 to July 14, 2016
Recently, the customer reviews became many important and useful information about evaluation and/or needs of customers. Many research about sentiment analysis for the customer reviews were proposed aggressively. In this paper, we focus on evaluation sentence patterns and the words appearing in them to estimate sentiment values of texts, and compared with a Bag-of-Words-based SVM classification. We investigated how many feature words obtained by SVM are included in the words extracted from the sentences of evaluation sentence patterns, and roughly 75% are included, which suggests that to just focus on a limited number sentences of evaluation sentence patterns can produce a comparable result even in its early stage.
Sentiment analysis, Compounds, Dictionaries, Support vector machines, Feature extraction, Estimation, Syntactics
M. Okada, K. Takeuchi and K. Hashimoto, "An Investigation of Effectiveness and Possibility to Automatic Construction of Estimation Patterns Using Customer Reviews," 2016 5th IIAI International Congress on Advanced Applied Informatics (IIAI-AAI), Kumamoto, Japan, 2016, pp. 267-272.