Improving Sentiment Analysis by Handling Negation on Twitter Data Using Deep Learning Approaches

Sentiment Analysis serves as a tool for identifying and quantifying the emotion conveyed in text. Within natural language processing (NLP), effective negation handling is crucial, especially for Twitter data. Hence, a technique with Convolutional Neural Networks (CNNs) model is used for classification. The system's efficiency is evaluated using the SemEval-2017 Twitter dataset, showing improved performance, particularly with negative tweets. By comparing CNN's performance in managing negation words, the paper assesses its effectiveness on tweet data. This strategy achieves superior accuracy over traditional machine learning methods and therefore mitigates misclassification in tweets.