Theme
Due to its capability to support continual monitoring, real-time data processing has become a very important mechanism in many application areas: traffic management, logistics, eHealth, smart grids, to name but a few. However, current solutions are limited by their “classical” approaches based on database or distributed computing technologies focusing mainly on the correlation between data (events) in real-time in order to discover interesting situations (millions of events can be correlated in a sec).
However, (1) real-time data processing is a highly challenging engineering problem: due to a very dynamic environment patterns of interest are continuously changing and the methods for their automatic discovery and efficient management (incl. evolution) are needed, (2) due to various ambiguities in data (noisy data, missing data, …) efficient real-time event correlation requires reasoning about events and their context and (3) finally, since there might be hundred of discovered situations, the resulting information must be presented in a compact (several levels of abstraction) and contextualized way in order to support an efficient decision making process.
In this tutorial we discuss how semantic technologies can help in resolving these challenges and presents some practical experience in developing and using such methods. We outline also the potential killer applications for semantic-based real-time data processing.