Sentiment Analysis using logistic regression and effective word score heuristic

Abstract

Sentiment Analysis is a method for judging somebody's sentiment or feeling with respect to a specific thing. It is utilized to recognize and arrange the sentiments communicated in writings. The web-based social networking sites like twitter draws in a huge number of clients that are online for imparting their insights in the form of tweets or comments. The tweets can be then classified into positive, negative, or neutral. In the proposed work, logistic regression classification is used as a classifier and unigram as a feature vector. For accuracy, k fold cross validation data mining technique is used. For choosing precise training sample, tweet subjectivity is utilized. The idea of Effective Word Score heuristic is likewise presented to find the polarity score of words that are frequently used. This additional heuristic can speed up the classification process of sentiments with standard machine learning approaches.