Sentiment Analysis using Long Short-Term Memory

Abstract

Sentiment analysis is becoming increasingly beneficial to monitor social media, allowing us to obtain insight into public opinion surrounding certain topics. Long Short-Term Memory (LSTM) has been widely applied in sentiment analysis tasks due to its ability to model the sequential nature of text data and capture contextual information. By considering the ordering of words and their dependencies, LSTM can better understand the sentiment expressed in a sentence or document. The proposed work focusing on identifying the opinion of the text utilizing Machine Learning procedures like LSTM. We propose LSTM based system to characterize Food Review data. The audits dataset is taken from the Amazon fine food reviews dataset. An audit here will be characterized given the memory of an LSTM cell state in the neural network. In this study, we are doing the effect of dropout layer, Comparison of batch, activation function, optimization algorithm, Single and Multi LSTM layer.