Lifelong Learning Sentiment Classification of Continuously Increasing Data

Abstract

Social media gives a platform for users to communicate with family, friends, and colleagues. Social media contain a massive amount of data ranging from a different topic like an opinion about any product, services, or personality. Lifelong learning helps to analyze and classify the sentiment of continuously increasing movie review data. The proposed approach is a dictionary-based technique i.e. a dictionary sentiment bearing words was used to classify the text classes like the positive, negative or neutral opinion. Achieving a high level of accuracy poses a challenge due to the large volume of data sets. Proper classification of emotion and sentiment can greatly influence productivity and quality of the different product in the market. To classify, there is a need, to have enough labeled data to perform sentiment classification task. The proposed system consists of one algorithms, the Naive Bayes algorithm. The objective of this dissertation is to improve the accuracy by Naive Bayes classifier for better results. Naive Bayes Algorithm gives 82% accuracy. It is also observed that the Naive Bayes algorithm is more accurate than Support Vector Machine algorithms.