Design and Analysis of Emotion Classification using RNN and LSTM

Abstract

Emotion analysis is a research in Natural Language Processing which helps finding useful insights from text. Sharing thoughts, opinions, and comments on social media formulate our sentiments from text. Social Media, for occurrence, is a rich reservoir of data that is a destination for directions for which they can use to investigate people’s judgments, thoughts and emotions. Sentiment analysis gives a more informed overview of the characteristics of an author's opinion typically. In large Social Media examination, nearly all projects have focused on analyzing the expressions as positive, negative or neutral. In this research work, it is designed to characterize the terms based on adding extra columns like adverb, noun, pronouns, adjective, verb, etc. There are several techniques of dynamic textual sentiment identification, but only an insufficient number were based on deep learning. This work represents the enhancement of a novel deep learning-based scheme, Recurrent Neural Network and Long Short-Term Memory that discusses the various emotion distribution difficulties on informative data. In this modified method, it is recommended to reconstruct it into a binary distribution as well as a traditional machine learning classification problem and utilize an in-depth knowledge strategy to determine the reconstructed problem. This hybrid approach provides improved classification accuracy over classical machine learning algorithms with the comparative analysis.