Abstract

Dynamic Markov Logic Networks as a knowledge base for cognitive technical systems

The talk discusses some key aspects of knowledge representation for cognitive technical systems with respect to their application within the SFB TRR 62. Markov Logic Networks are introduced as an adequate means for expressing dynamic, symbolic knowledge; they provide us with a probabilistic, relational calculus with a resemblance to first-order logics. As an example, a simple model for supporting output-channel selection in a multi-device environment is presented. In addition, our approach for online-capable, approximate inference in dynamic MLNs is outlined.