Survey on Feature Selection for Data Mining and its Application in Opinion Mining
S Sumanth, S Siddarama · Mapana Journal of Sciences · 2023
Sentiment Analysis (SA) and opinion mining is used for the systems of business intelligence in analyzing public opinion towards various brands and implementing market strategies. Machine learning aims at developing the algorithm in such a way that the performance of a system is optimized with past data or experience. In the case of SA, the feature selection method is used for the identification of different goals such as reduction of cost of computation, avoidance of over-fitting, and enhancing the accuracy of classification of the model. The methods of feature selection can reduce their original feature sets by the removal of irrelevant features for the classification of text sentiment and their accuracy. The paper surveys various feature selection techniques available in the literature. The study shows that feature selection significantly improves the classification of the sentiments, but it depends on the technique adopted and the number of features selected.