Context-Aware Sentiment Analysis Using Long Short-Term Memory

Abstract

The manual analysis of extensive textual datasets poses substantial challenges due to its time-consuming nature. Sentiment analysis through machine learning is an automated technique that efficiently detects positive and negative sentiments in text, providing understandings from social media comments, survey responses, and product reviews to support data-driven decisions. This research implements sentiment analysis using word embedding and LSTM to process unstructured text, optimizing understanding with minimum manual effort, it excels at managing large datasets, this approach is applied to text reviews and achieved the accuracy of 86.30%. Built on the foundations Convolutional Neural Networks (CNN) and LSTM model showcases superior performance in sentiment classification. Future research will explore diverse embedding models and broader datasets to further enhance the model's versatility and precision.