A Hybrid Deep Learning Approach for Sentiment Analysis using CNN and Improved SVM with Multi Objective Swarm Optimization for Domain Independent Datasets

Raviya K
2020 International Journal of Advanced Trends in Computer Science and Engineering  
In the Big Data world, there is an exponential increase in the volume of numerous data, such like text, image, audio and video, as text is the largest among them. Sentiment analysis is a trendy application in text mining, where text data concerning the feelings or attitude of the consumer is collected using different methods or techniques. Sentiment detection of online product reviews is helpful to figure out emotions and viewpoints of customers. Many researchers have just developed the Deep
more » ... rning model for obtaining tremendous performance in NLP. This paper suggested novel deep-learning hybrid architecture that is highly effective for analyzing sentiments on domain independent datasets. We blend deep Convolutional Neural Networks (CNN) and support vector machines (SVM) for a better overall classification. The reason for using CNN is not only to extract local features but also the framework for predicting sentiments and combining CNN output into SVM progress the classification. In addition, we have adopted MSPSO (Multi-swarm Particle Swarm Optimization) system to obtain optimized feature selection and to train the SVM for further improved in sentiment classification. To demonstrate that our proposed method is of a generic nature, we have evaluated on datasets of online product reviews from various domain. Evaluation exposes that the implementation of the proposed approach is reliable among all the datasets and often outshine the state-of-art systems.
doi:10.30534/ijatcse/2020/111932020 fatcat:ysklwwrcmzhehdardtr5vkk3je