Cross-lingual detection of world events from news articles Gregor Leban, Blaˇ z Fortuna, Janez Brank, and Marko Grobelnik Joˇ zef Stefan Institute Ljubljana, Slovenia {firstname.surname}@ijs.si Abstract. In this demo we describe a system called Event Registry (http://eventregistry.org ) that can identify world events from news arti- cles. Events can be detected in different languages. For each event, the system can extract core event information and store it in a structured form that allows advanced search options. Numerous visualizations are provided for visualizing search results. Keywords: information extraction; cross-linguality; events; named entity de- tection; clustering; news 1 Introduction Each day there are thousands of small, medium and large events that are hap- pening in the world. These events range from insignificant meetings, conferences and sport events to important natural disasters and decisions made by world leaders. The way we learn about these events is from different media channels. The unstructured nature of content provided on these channels is suitable for humans, but hard to understand by computers. Identifying events described in the news and extracting main information about these events is the main goal of the system presented in this paper. Event Registry [2] is able to process news articles published in different languages world- wide. By analyzing the articles it can identify the mentioned events and extract main event information. Extracted event information is stored in a structured way that provides unique features such as searching for events by date, location, entity, topic and other event properties. Beside finding events of interest, Event Registry also provides user interface with numerous visualizations showing date, location, concept and topic aggregates of events that match the search criteria. The rest of the paper is organized as follows. We start by describing the high level architecture of the system. Next, we describe in more details the process of identification of events from individual news articles. We continue by providing information about how event information is extracted from a group of articles mentioning the event. In the end we also describe the main features of the frontend interface of the Event Registry.