Streams, Contexts, and Dynamic Configuration?

                                     Thomas Eiter

                            Knowledge-Based Systems Group,
                           Institute of Logic and Computation
                             Vienna University of Technology
                       Favoritenstraße 9-11, A-1040 Vienna, Austria
                              eiter@kr.tuwien.ac.at



                                   Talk Abstract
     The increase in network connectivity and the availability of sensors and data sources
of other kind has led to a growing interest in processing data streams, and in particu-
lar reasoning over streams has received attention in the knowledge representation and
reasoning community. Various formalisms have been proposed that offer different fea-
tures in order to facilitate reasoning from and about a knowledge base over time in the
presence of streaming data, often based on temporal logic and extensions. The need for
addressing time and temporal evolution of knowledge bases has also led to respective
extensions of multi-context systems (MCS), which are a generic formalism for modeling
interlinked knowledge bases, called contexts, in an abstract way, where information
exchange between contexts is enabled by special bridge rules. Among these formalisms
are reactive, evolving, asynchronous, and streaming MCS, to name a few.
     The generic framework of MCS allows to model particular formalisms, such as
argumentation context systems which support group argumentation in the presence of
mediators. In this talk, we consider dynamic configuration as another challenging prob-
lem for possible realization by multi-context systems. Roughly speaking, components in
a system, called producers, are controlled by configurators which set parameters in order
to determine the producers’s behaviour; in turn, configurators are linked to monitors
which observe sensors and feed information to the configurations. In the DynaCon
approach, the configurations control also the monitors and may change their behaviour
dynamically; this enables more sensitive configuration, while the system behaviour may
get more complex, depending on the linkage structure, which in closed control loops
is cyclic already in plain settings. The MCS framework and its temporal extensions
provide a versatile tool for modeling dynamic configuration and the information flow
among components modeled as contexts; notably, heterogeneous components, based on
different logics or decision procedures, can be conveniently covered. We shall address
issues and problems to consider in the MCS formulation for this application, which open
new avenues of research.




?
    This work has been supported by TU Wien and the Austrian Research Promotion Agency
    project DynaCon (FFG 861263)