Uncertain Complex Event Processing Based on Markov Logical Networks

The main task of complex event processing is to identify high-level events of interest from the original events. However, the traditional complex event processing technology can only solve deterministic events, but it can't do anything about uncertain events. In the real world, uncertainty can't be avoided, such as loss in event collection, imperfect rule definition, and even in some cases, the incident is accompanied by a certain degree of confidence, and we only pay attention to the last situation, it is soft evidence. For this situation, we need to update our beliefs in a timely manner. In this work, we propose a uncertainty complex event processing method based on Markov logic networks, and adopt a Markov Monte Carlo method, called MC-SAT-PC, which can easily combine the probability attribute of event into inference. In the process, without excessive computational overhead. Our experiments in the field of activity identification demonstrate the feasibility of the method.