Retraction Notice: Classification Accuracy in Sentiment Analysis using Hybrid and Ensemble Methods
Prachi Juyal · 2022 IEEE World Conference on Applied Intelligence and Computing (AIC) · 2022
Sentiment Analysis (SA) is a relatively young subject of research in the science of text mining. SA is seeking for opinions, feelings, and subjectivity in the content that it reads. The main purpose of sentiment analysis is to look at product and service reviews to see how many ratings they get. The ensemble model (EM) is proposed in this paper to improve machine learning (ML) results by combining many ML models into a single optimal predictive model. To improve the accuracy of the data mining model, two or more analytical models that are related but not identical are run, and the results are aggregated into a single score. Ensemble methods such as bagging, boosting, stacking, and voting are commonly utilized. Voting is best for classification problems, while regression is usually done by averaging. As a result, the ensemble model using the majority voting technique is used to improve sentiment analysis classification accuracy. For classification, the proposed ensemble classifier incorporates the results of prior work as well as AdaBoosting in SVM. These algorithms’ classification results are fed into the majority voting procedure as inputs. Each classifier’s forecast for test phrases is taken, and the final output prediction with more than half of the votes is deemed the winner. The suggested method outperforms both individual machine learning algorithms and previously proposed hybrid methods in terms of classification accuracy.