Sentiment analysis is like a gateway to AI based text analysis. With an explosion of text data available in digital formats, the need for automated sentiment analysis is growing rapidly. Sentiment analysis can be perfectly suited for product reviews. A product review system is very important for a E-commerce business, but the effectiveness of the product review system is not quite up to the mark as the industry requires. Thus sentiment analysis became a popular research area in opinion mining in recent years. In recent researches many Machine learning classifier based models have achieved significant accuracy in analyzing the sentiment of corpus based on the following four important tasks data preprocessing, class labeling, annotation granularity, and target identification. This paper proposed a model for classifying the sentiments based on their polarity using the Machine learning classifier Naive Bayes which is mostly used in text classification and has high success rate when compared to other algorithms which is the simplest classifier based on ‘Bayes Theorem’ because of its simplicity. This classifier works very well when the dataset is small. But as the data increased the performance of the model reduced. So, To improve the performance of the model it was integrated with Collaborative Filtering Algorithm and Multinomial Naive Bayes, the distribution that normally works well for parameter tuning. Through this approach the sentiment of the reviews of products based on trained dataset were predicted and it showed considerably a better performance .The main objective of this integrated approach is to overcome the limitations of the Naive Bayes Algorithm. The implemented results shows the effectiveness of the proposed approach for sentiment classification of the reviews through its accuracy which is calculated using the Confusion Matrix.The results thus obtained are represented graphically.
