Independent component support vector regressive deep learning for sentiment classification

Abstract

Sentiment analysis is the process used to extract the keywords from the user review and classify the text into positive, negative and neutral. Sentiment analysis and opinion mining is the field used to analyzes people opinion, sentiment, evaluation, attitude and emotion. Many researchers introduced opinion mining methods to improve the detection accuracy. But, the time consumption was not reduced and accuracy level was not improved. In order to address these problems, an Independent Component Support Vector Regressive Deep Learnt Sentiment Classification (ICSD) Method is designed. ICSD Method comprises five layers to perform sentiment classification. ICSD Method performs pre-processing, feature extraction and classification to enhance the classification accuracy with extracted words from user review comments. Initially in ICSD Method, the user review comments are collected as an input at the input layer. The user review comments are sent to the hidden layer 1. In that layer, ICSD Method pre-processes the use review comments for removing the stop words to reduce the file size. The pre-processed user review comments are sent to the hidden layer 2. In hidden layer 2, Independent Component Feature Extraction is carried out in ICSD Technique to extract the opinion word from review comments. The associated opinion words are arranged for the semantic similarity of sentiment based on the extracted word to minimize the time consumption. After that, the extracted words are transferred to the hidden layer 3. In that layer, Support Vector Regressive Sentiment Classification is performed in ICSD Method to examine the semantic opinion words for determining the sentiment class label. The sentiment class are categorized into positive, neutral and negative sentiments. Finally, the results are sent to the output layer. This in turn helps to minimize the time consumption and to progress the accuracy of opinion mining. The performance evaluation of ICSD Method is carried out with consumer product and services reviews extracted. The parameters used are evaluated with number of customer review words, accuracy, time complexity and false positive rate. Experimental evaluation of ICSD Method minimizes the time consumption and false positive rate during opinion extraction from reviewers.