Sentiment Analysis for Social Media Big Data utilizing Advanced Hybrid Deep Learning Algorithms

Abstract

Abstract With the increase in the use of social media sites significantly and the enlarge in the users number as they provide all the basic services and daily requirements of customers for most vital activities at all levels of commercial, social, marketing, educational and others, there is an urgent need to provide a comprehensive analytical and predictive system that works on reading and analyzing the desires of users to give a comprehensive description It facilitates searches and shortens effort, time, storage and search spaces for your servers and customers alike. In this study, three data sets have been considered and tested using sentiment analysis with optimization technique convolutional neural network (CNN), hyper parameters optimized Tree Structured Parzen Estimator with LSTM, and hybrid CNN with Groo, hyper parameters optimization using Tree Structured Parzen Estimator (TPE), and Define by Run deep learning algorithms utilizing Jupiter Python programming language. The data sets represent big data for Amazon social media containing information with different classes, items, and attributes. The classification of the analyzed data will produce a recommender system that simplify the selection process for customers and users of social networking sites by providing readings and analyzes of various types of information circulated on different types of websites, so that the information is trained as a supervisory training for various tweets by collecting different customers information to predict their various interests and needs.