Enhancing the classification accuracy in sentiment analysis using hybrid and ensemble methods

Abstract

Sentiment Analysis (SA) is the current field of research in text mining field. SA is detecting opinions, sentiments, and subjectivity of text. It is the application of natural language processing techniques and text analytics to identify and extract subjective information from the frequently used sources such as web and microblogs. The main objective of sentiment analysis is to analyse reviews of products and services, and determine the scores of such sentiments. The major problem is that the reviews are mostly unstructured and thus, need classification or clustering to provide meaningful information for future use. Most of the supervised approaches in sentiment analysis fail to produce effective mining results, when trained in one domain and applied to others. Another major problem is that most of the works concentrate only on detecting the overall sentiment of any document and do not perform in-depth analysis by ignoring the latent topics and the sentiment associated with those topics. This problem is addressed in this research work. Different methods are proposed to overcome the above-mentioned problems. Hence, the objective is to improve the classification accuracy with multi-classifier systems on IMDb and multidomain datasets. Multiple Classifier or Multi-Classifier Systems (MCSs)(Ensemble Learning) fuse together multiple classification outputs for better accuracy and classification. This can result in slight differences within the results obtained by individual classifiers. When different classifiers are combined in a proper method, the combined result gives the average of best performing classifier within the ensemble of classifiers. It is used to achieve the best possible classification, by increasing the efficiency and accuracy of classification. This research work presents a survey of several machine learning techniques to enhance the classification accuracy in sentimental analysis. The accuracy of these methods is examined to assess their performance based on metrics such as precision, recall, and accu