SENTIMENT ANALYSIS USING REPRESENTATIVE TERMS - A GROUPING APPROACH FOR BINARY CLASSIFICATION OF DOCUMENTS
Sanju Joseph · 2012
Automatic classification of customer reviews as eit her positive or negative has been of great interest among the academic and business community in the recent t imes. In this paper, an attempt has been made to represent the text documents using just eight repre sentative terms (RT) viz. good, very good, excellen t, recommended, bad, very bad, disgusting, and never r ecommended. Thus a new way of representing text documents as a structured data matrix has been crea ted. A consistent classification accuracy of near 8 0% and above was achieved for datasets of various size s ranging from 403 to 25000. The precision (P), recall(R) and F-Measure were also very consistent a nd comparable to the previously reported results. A comparative analysis of classification performance has been carried out using machine learning algorit hms like Naive Bayes (NB), Bayesian logistic regression (BLR), multi layer perceptron (MLP) etc., revealed that the proposed way of representing the text docu ments results in consistently superior performance.