MULTI-ASPECT SENTIMENT ANALYSIS OF SOCIAL MEDIA TEXT USING LEXICON BASED FASTEXT

Abstract

Social media platforms serve as vast repositories of diverse user-generated content expressing opinions, sentiments, and experiences across multiple facets of life. Traditional Sentiment Analysis (SA) methods often fall short in capturing the multifaceted nature of opinions present in social media text, as they tend to oversimplify the analysis by focusing solely on overall sentiment polarity. This paper introduces a novel approach to overcome this constraint by the usage of Multi-Aspect SA utilizing a Lexicon-Based FasText framework. By leveraging FasText, an efficient word embedding technique, enriched with lexicon-based SA, this research aims to discern sentiments across various aspects or dimensions within social media texts. Our methodology involves the creation and utilization of domain-specific sentiment lexicons to capture sentiments related to different aspects such as product features, service quality, user experience, and more. The FasText framework is employed to generate context-aware word representations, enhancing SA accuracy across diverse aspects within social media texts. The experimental evaluation demonstrates how effective the recommended approach is in extracting nuanced sentiments, providing insights into multidimensional user opinions present in social media content. This research contributes to advancing SA methodologies by enabling a more granular and comprehensive understanding of sentiment expressions within the dynamic landscape of social media communication.