The research activities of the CER group are summarised here.
Today's organisations collect data in various structured and unstructured digital formats. In particular, the abundance and continuous growth of networks of different types of sensor has led to corresponding data streams, which are hard to handle and utilize in the decision-making process of organizations. The analysis and interpretation of the collected data needs to be automated, in order for large data volumes to be transformed into operational knowledge. Events are particularly important pieces of knowledge, as they represent activities of special significance within an organisation. Therefore, the "recognition of events" is of utmost importance. Consider, for example, the recognition of non-ecological and unsafe driving in public transportation given data from the vehicle sensors, the recognition of violent behaviour in public spaces given video content, and the recognition of various types of cardiac arrhythmia given electrocardiographs.
The Complex Event Recognition (CER) group of NCSR "Demokritos" works towards advanced and efficient methods for the recognition of complex events in a multitude of large, heterogeneous and interdependent data streams. In particular, we are developing novel CER methods, which take as input streams of low-level events, e.g. sensor-based events, such as a change in temperature, and combine them to infer complex high-level events of interest, such as the start of a fire incident or a fault in the cooling system of a vehicle. Our approach is primarily based on the Event Calculus, a simple logic-based formalism that supports effective reasoning about complex events. In addition to its formal semantics, this approach has allowed us to develop efficient real-time recognition methods, as well as to combine logic-based with statistical reasoning, in order to handle the inherent uncertainty of most event recognition applications. Furthermore, it lends itself directly to the development of machine learning methods for acquiring logic-based event descriptions from data.
The CER group is part of the Software & Knowledge Engineering Lab of the Institute of Informatics and Telecommunications in the National Center for Scientific Research "Demokritos".