Sentiment Analysis and Prediction using Text Mining

Objectives: The main aim of the proposed system is to predict the ratings of a textual review using the concept of sentiment analysis. Prediction is an important process to know about the user sentiment. Methods/Statistical Analysis: This work has a sentiment-based rating prediction method (RPS) to upgrade the prediction accuracy in any recommender system. It basically constitutes of a factor used in predicting the rating. Initially, we calculate user’s sentiment on an item/ product based on user sentimental approach. We apply cosine similarity to find user’s sentiment similarity between the users. By taking user’s sentiment similarity into consideration we can fill the missing values and predict the rating for the products that have not been reviewed by the users. Findings: We assess the above two sentimental factors on a sample dataset collected. Eventually the results show that the sentiment distinguishes user preferences which let a helping hand to enhance the performance of the recommender system. The proposed system is executed and the results show 80% accuracy. This work is efficient in terms of the factors used. Application/Improvements: The system can be enhanced with the addition of other factors and fusing them to the recommender system. The use of matrix factorization can, however, be more efficient while using 2 or more factors.