I-CARE: Intelligent Context Aware system for Recognizing Emotions from text

Abstract

Because text, either written or spoken, is the unrivaled way to communicate one's feelings and emotions, it is the best modality to harness for building efficient emotion recognition systems. However, researches on building text-based emotion recognition systems are progressing slowly compared to face or voice based systems. In fact, natural language evolution and text-related contexts are both constraints that slow down the progress of text-based emotion detection. This paper introduces I-CARE: a new Intelligent Context Aware system for Recognizing Emotions from text. This system considers three essential points to infer emotion: presence of affective keywords in the text, general context surrounding the human subject and how he/she perceives it emotionally. To do so, we use a mobile application that gathers text typed by its user whenever he/she uses an Instant Messaging (IM) application or publishes content on a social media. The application then sends it to a remote server for text processing and emotion extraction. In addition to text, the mobile application also collects contextual data, such as: location and upcoming events when available, to increase the accuracy of text-based emotion detection.