A Survey of Textual Event Extraction from Social Networks

In the last decade, mining textual content on social networks to extract relevant data and useful knowledge is becoming an omnipresent task. One common application of text mining is Event Extraction, which is considered as a complex task divided into multiple sub-tasks of varying di culty. In this paper, we present a survey of the main existing text mining techniques which are used for many di erent event extraction aims. First, we present the main data-driven approaches which are based on statistics models to convert data to knowledge. Second, we expose the knowledgedriven approaches which are based on expert knowledge to extract knowledge usually by means of pattern-based approaches. Then we present the main existing hybrid approaches that combines data-driven and data-knowledge approaches. We end this paper with a comparative study that recapitulates the main features of each presented method. Key-words: Event Extraction, Text Mining, Information Extraction, So-