• Dr Selvapriya M

Social Media has gained remarkable attention. This is attributed to the affordability of accessing social network sites such as Twitter, Google+, Facebook and other social network sites through the internet and the web 2.0 technologies and in social media services has enormously helpful for various users to depict their feelings and opinions through news articles, blogs and tweets. Twitter is one of the important and popular social media in which anybody can leave tweets about anything that occurs. However, these emotions also include noisy instances and it is a enormous challenge to get the textual inference of brief messages. It is a difficult task to identify the identical user documents from the complete social media text message. Also, online comments are classified by attributes using a sparse feature space, which increases the complexity of the corresponding emotion classification task. Therefore, feature selection is a desirable solution for finding a solution to this problem. Hybrid Neural Network (HNN) has been proposed for Social Emotion Classification. But, reducing sparse feature space from the emotion dataset tends to be an extremely cumbersome task. Three important contributions were made in this work to address these concerns. In the first contribution of the work, Quantum Behavior Particle Swarm Optimization based Sparse Encoding (QBPSO-SEn) approach is presented to choose the Optimum features from dataset which is helpful in increasing the robustness of the CNN model. The results of the experiments show that the proposed CNN model yields an increased classification accuracy in comparison with other techniques like Hybrid Neural Network (HNN) and also Neural Network (NN) schemes. However, this scheme does not support multi-label emotion classification.The second contribution of the work, Mutation Bat Optimization based Sparse Encoding (MBO-SC) is presented for translating the sparse low-level features into tight high-level features.However, the feature selection approach is not used in this mechanism and it may influence the classification accuracy.The third contribution of the work, Integrated Feature Selection (IFS)-EWCNN is presented to boost the performance of social emotion classification done on twitter reviews.The proposed IFS-EWCNN classifier yields superior performance in comparison with the available techniques in terms of precision, recall, sensitivity, specificity, F-measure and accuracy. The experiments are performed in SemEval 2016 Task 4A for sentiment and SemEval 2018 Task 1C emotion classification using the MATLAB tool. The proposed IFS-EWCNN classifier techniques help in the accurate classification of social emotions. The results of experiments reveal that the proposed IFS-EWCNN classifier achieves 89.31% of precision, 96.40% of recall, 92.45% of f-measure, 96.23% of specificity, and 96.20% of accuracy correspondingly.

Social Media, Emotion Classification, Feature Selection, Sparse Encoding, Mutation Bat optimization, Twitter Sentiment Analysis.

"CONVOLUTIONAL NEURAL NETWORK MODELS FOR IMPROVED SOCIAL EMOTION CLASSIFICATION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.11, Issue 3, page no.a572-a578, March-2024, Available :http://www.jetir.org/papers/JETIR2403077.pdf

"CONVOLUTIONAL NEURAL NETWORK MODELS FOR IMPROVED SOCIAL EMOTION CLASSIFICATION", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.11, Issue 3, page no. ppa572-a578, March-2024, Available at : http://www.jetir.org/papers/JETIR2403077.pdf