Alexandra La Fleur, Adrian Paschke and Kia Teymourian

Complex Event Extraction from Real-Time News Streams

Information overload on news data is a known problem these days. People and organizations have an increasing demand for extraction of relevant information from massive amounts of news data arriving in real-time as news streams. In this paper, we present a novel approach for real-time extraction of news, based on user specifications and by using background knowledge from specific news domains. We create a powerful filtering service which limits the news data to the concrete and essential preferences of a user. In our approach, enrichment of real-time news with background knowledge is a preprocessing step. We use a Complex Event Processor to detect complex events from the enriched articles and match them to the user specified query. Each time a news article is matched, its result is notified to the user immediately. Our experimental evaluation shows that our approach is feasible for detecting news in real-time with high precision and recall.