An event processing approach to text stream analysis
Andreas Bauer, Christian G. Wolff · 2014
Information filtering is a crucial task in a world where data is generated steadily and at a high rate, helping users in distinguishing relevant from irrelevant content. This requires efficient processing of continuous streams of textual data. Event processing allows for real time processing of data streams. But up to now event processing has mainly been investigated in the context of business transaction-oriented domains like logistics or finance, but not explicitly in terms of text stream processing and information filtering. The growth of applications that analyze social media streams lets such an approach appear reasonable. Therefore we propose a common vocabulary represented by a text domain event model as well as a reference architecture for text stream processing and information filtering, in order to facilitate the implementation and the assessment of event processing applications for text streams. In addition we describe results from actual use cases that employ this architecture.