Semantic Complex Event Processing for Social Media Monitoring - A Survey

Semantic Complex Event Processing (CEP) is a promising approach for analysing streams of social media data in crisis situations. Traditional CEP approaches lack the capability to semantically interpret and analyse data, which Semantic CEP attempts to address, but current approaches have a number of limitations. In this paper we survey four semantic stream processing engines, and discuss them with the specific requirements of CEP for social media monitoring in mind. Current approaches assume well-structured data, known streams and vocabularies, and mainly static event patterns and ontologies, neither of which are realistic assumptions in our scenario. Additionally, the languages commonly used for event pattern detection, i.e., SPARQL extensions, lack several important features that would facilitate more advanced statistical and textual analyses, as well as adequate support for temporal and spatial reasoning. Being able to utilize external tools for processing specific tasks would also be of great value in processing social data streams.