Multi-categorical social media sentiment analysis of corporate events

Abstract

It is indisputable that the public's sentiment toward corporations determines their brand image and enterprise value. Nonetheless, the majority of studies have merely focused on financial indexes and polarity-based sentiment analysis. The absence of an effective sentiment analysis method has limited corporations' opportunity to grasp and manage the public's sentiment systematically. In this research, we propose a practical multidimensional sentiment analysis methodology, which consists of constructing a multi-categorical lexicon process using morphological analysis, detecting corporate events with social network analysis, and classifying detected events by stakeholder-based event classification. We tested the proposed framework with data collected from Twitter. Examining the analysis, we concluded that the suggested approach provides more worthwhile information about how the public's sentiment is composed and what the causes are of each sentiment, which could not be identified from previous studies. We are also convinced about the feasibility of the proposed methodology for business purposes and the possibility of extended usage regardless of discipline.