Temporal Sentiment Tracking and Analysis on Large-scale Social Events

Online Social Networks (OSNs) are emergent resources for large-scale multi-purpose data analytics. Sentiment analysis (SA) is a trending research area on OSNs. SA approaches for studying and analyzing events are still missing several shortcomings. Unlike other approaches that analyzed micro-scaled events such as "marriage", "graduation", we analyzed the sentiment of large-scale social events such as "festivals". In this paper, we address the problem of finding the sentiment of large-scale social events and introduce a novel method for this goal. To address this problem, we utilize a lexical approach. The features used in our method are universal and composed of auxiliary and essential features from OSNs. Auxiliary features are non-textual features used to emphasize the sentiment polarity. Moreover, we track the temporal interchanges of audience sentiment on OSNs. We finally empirically validate that our method can outperform with high precision and recall values.