Published Paper Details:

Sentiment analysis used machine learning to remove the significant data for subjective data analysis by positive, negative or neutral feelings. Different sentiment classification techniques are discussed for performing the opinion mining. However, they failed to enhance the accuracy level and time complexity. To overcome this a novel method called as Lasso Regressive Percentage Similarity based Extreme Learning Network Method (LRPS-ELM) is introduced to perform main objective of efficient sentiment classification with better accuracy and minimum time consumption. Extreme Learning Classification comprises different layers for categorizing the given reviews through performing preprocessing, feature extraction and classification. The number of reviews is initially collected from input dataset as an input. Then, the input reviews are sent to hidden layer 1. In that layer, the review preprocessing is performed through stop words removal and stem words elimination. Then, the pre-processed review is sent to the second hidden layer. In that layer, Lasso Regression is carried out for performing efficient feature extraction from preprocessed reviews. Later, the extracted features are sent to the third hidden layer for classification. In that layer, Percentage Similarity Function is carried out by identifying the user opinion. Finally, accurate sentiment classification is carried out with higher accuracy. Simulation setting is provided with various metrics such as, prediction accuracy, error rate and prediction time with respect to number of reviews. In the proposed LRPS-ELM method results are verified to achieve better accuracy and lesser time complexity when compared to conventional methods.