An Effective Sentiment Analysis Classification for Opinion Prediction using Machine Learning Schemes

In the blooming of web technology, the rise of electronic commerce websites is rapidly increasing. Users started expressing their views similar to text reviews or comments. This research deals with the identification and extraction of the exact opinion of the customers across reviews, feedbacks, or comments. Pre-processing techniques are used to clean the dataset and the machine learning algorithm like k-nearest neighbor (KNN) is used as a solution to this problem, to have a good evaluation of sentiments (Positive, Negative, Neutral). In accumulation, the precision of the algorithm highly depends on the value of k and distance metric for classification. In our work we used k value of 6 and euclidean distance equation for classification of the KNN algorithm. The proposed KNN algorithm has been applied on publicly available standard review dataset repository, to show the exact nature and features based on customer’s opinions, and the accuracy of determining the feasibility of the k-nearest neighbor algorithm of 98% is predicted.