Leveraging Sentiment Analysis and Negation Phrase Identification for Mitigating Information Overload in E-commerce

In the era of e-commerce, online reviews have become crucial for both consumers and merchants. However, the overwhelming volume of reviews has resulted in Information Overload (10). To address this, we employ Sentiment Analysis (SA) on user comments, utilizing machine learning to classify sentiments. Our study focuses on the challenge of accurately predicting sentiment orientation, particularly noting that adjectives and adverbs with negative prefixes can convey opposite sentiments. We introduce a negation phrase identification algorithm to address this issue. By combining sentiment analysis and negation phrase identification, our research aims to alleviate information overload, providing valuable insights for e-commerce stakeholders. Our model, utilizing the Bayes classifier, achieves an 89% accuracy, showcasing the effectiveness of machine learning in understanding customer sentiment. The research findings indicate that employing machine learning algorithms, particularly the Bayes algorithm, on textual data will allow businesses to gain valuable insights into customer sentiment, surpassing the limitations of manual efforts. The combined approach positions the sentiment analysis methodology as a forefront contender in natural language understanding, promising more reliable and contextually accurate assessments of sentiment from user-generated text data.