Monitoring, Security and Rescue Techniques in Multiagent Systems

In [4] we presented a new approach to data analysis based on flow distribution study in a flow network, called here a flow graph. The introduced flow graph differs from that proposed by Ford and Fulkerson [1], Branches of the flow graph are interpreted as decision rules, whereas the flow graph is supposed to describe a decision algorithm. Hence, we propose to model decision processes as flow graphs and analyze decisions in terms of flow spreading in the graph. With every decision rule three coefficients are associated, called strength, certainty and the coverage factors. These coefficients were widely used in data mining and rough set theory, but in fact they were first introduced by à Lukasiewicz [1] in connection with his study of logic and probability. These coefficients have a probabilistic flavor, but they can be also interpreted in a deterministic way, describing flow distribution in the flow graph. Moreover, these coefficients satisfy Bayes’ rule. Thus, in the presented approach Bayes’ rule has entirely deterministic interpretation, without reference to its probabilistic nature, inherently associated with classical Bayesian philosophy. In this paper we introduce hierarchical structure of flow networks by allowing to substitute of single branch in the flow graph by another flow graph. This ”zooming” operation can be repeated recursively, which allows us to look at data with different accuracy and move from general picture to details. A simple tutorial example will be used to illustrate the introduced ideas.