Estimating the sentiment strength of a word using word similarity

摘要

>Identifying the sentiment orientation and strength of a sentence or phrase is one of the crucial issues in opinion mining. A dictionary approach to this problem is to use the sentiment orientations and strengths for the words included in a given document unit. This paper proposes a novel method for estimating the sentiment strength of a word using similarities between a target sentiment word and sentiment words in dictionary. To calculate the word similarity, we introduce the notions of relative similarity and adjusted similarity. The experimental results from the analysis of Facebook text data demonstrate the effectiveness of the proposed method in terms of the root-mean-square-error.