Can We Rank Emotions? A Brand Love Ranking System for Emotional Terms

In this paper we examine customers´ emotional attachment to a brand name utilizing content extracted from social media. More specifically, we consider the emotions associated to brand love appearing in the form of terms in users´ Twitter posts. Building on existing work that identifies seven dimensions in brand love, we propose a probabilistic network scheme that employs a topic identification method so as to identify the aspects of the brand name. In order to address these visible key signs, we find a realistic emotional term ranking that integrates different types of information within the inference network considering the emotional dimensions of terms, their synonyms as well as the current (dynamic) aspects of the brand. Moreover, we introduce a Twitter Behavior Metric that depicts user behavior and we associate brand love to user behavior. The effectiveness of our approach is demonstrated by sampling the Twitter graph on a specific brand and examining the inference network output as well as its relationship with users´ behavior metric.