Social networks event mining: A systematic literature review

Abstract

Social Networks (SNs) become a major source of reporting new events that happen in real life even before the news channels and other media sources report them nowadays. The objective of this paper is to conduct the systematic literature review (SLR) to identify the most frequently used SN for reporting and analyzing the real-time events worldwide. Furthermore, we recognize the features and techniques used for mining the real-time events from SNs. To determine the literature related to event mining (EM) and SNs, the SLR process has been used. The SLR searching phase resulted 692 total studies from different online databases that went through three phases of screening, and finally 145 papers out of 692 were chosen to include in this SLR as per inclusion criteria and RQs. Based on the data analysis of the selected 145 studies, this paper has concluded that the Twitter microblogging SN is the most used SN to repot the events in textual format. The most common features used are n-gram and TF-IDF. Results also showed that support vector machine (SVM) and naive Bayes (NB) are the most frequently used techniques for SNEM. This SLR presents the list of SNs, features, and techniques that are reporting the SN events that can be helpful for other researchers for selection of SNs and techniques for their research.