Data-driven Social Mood Analysis through the Conceptualization of Emotional Fingerprints

Abstract A body of knowledge shows the emerging of evidence according to a better account for the emotional spectrum is achievable by employing a complete selection of emotion keywords. Basic emotions, such as Ekman’s ones, cannot be considered universal, but are related to with implicit thematic affairs within the corpus under analysis. The paper tracks some preliminary experiments obtained by employing a data-driven methodology that captures emotions, relying on domain data that you want to model. The experimentation consists of investigating the corresponding conceptual space based on a set of terms (i.e., keywords) that are representative of the domain and the determination. Furthermore, the conceptual space is exploited as a bridge between the textual content and its sub-symbolic mapping as an “emotional fingerprint” into a six-dimensional hyperspace.