Emotion-Based System for Social Media Content Processing and Event Monitoring

摘要

In this paper, we propose an open source web-based emotion retrieval service for textual social media contents, called MediaTagger, which can extract emotional value from online content coming from diversified Internet-based services. MediaTagger mashes up state of the art emotion services and allocates the right emotion retrieval service using several emotion visualization metaphors based on the content of each service. MediaTagger also incorporates a flexible and adaptable emotion authoring service based on Na?ve Bayes machine learning theorem. The authoring service within the system also helps creating new domains of emotion extraction services. We share some quantitative and qualitative results that show the viability of our system as well as the experience and satisfaction of regular end-users. We will describe a case study that leverages the features provided by the MediaTagger. This study shows the utilization of emotion-based textual posts for monitoring and decision-support systems during major events such as Hajj.