A Study the Adaptive Big Data Model for Sentiment Analysis

An adaptable approach for sentiment analysis of social media data using deep learning and topic modeling

Abstract

This study proposed an adaptable model for topic-level SA of social media data using deeplearning. A topic modeling-based strategy, live latent semantic indexing with regularization constraint, wasdeveloped for topic extraction from streaming data. This study uses a long short-term memory networkwith a topic-level attention mechanism for SA. The proposed model's meat and potatoes are its ability toextract topics from streaming sentences from social media platforms in an online manner via adaptiveupdates to the model for new incoming streaming data, followed by topic-level SA on those streamingsentences to determine the polarity of the detected topics. The online latent semantic indexing topiccoherencyscore produced demonstrates the effectiveness of the suggested method for topic detectionon real-time data. The SemEval-2017 dataset is used to train the model for SA.