Revealing the Structure and the Cause-Effect Relations in Partly Observable Systems

Nowadays the growing demands are the dominant concepts in most parts of life. To satisfy these demands, planning of more complex and flexible problem-solving systems is required. In the last decades many new technologies and techniques were developed to handle these growing demands. Apparently the object-oriented modelling methodology was the most efficient among them. However, nowadays the growing demands of the market gradually outgrow the abilities of the pure object-oriented concepts. One of the main reasons of this is that the decomposition techniques can not handle efficiently the numerous sub-systems with varying objective functions and constraints. The common problems appear in the untreatable complexity and the missed deadlines. The use of artificial intelligence means new concepts in the field the development processes. The agent-based programming gives the possibility to describe the functionality of the required system not only by using action-reaction pairs but by defining the goals and constraints in the system. Machine-learning helps to determine the connections, relations and logical behaviour in the dynamism of the modelled system and helps to reveal the effects of the none-modelled systems in the modelled system. This paper shows a method for revealing and handling the effects of a none-modelled system according to the observed behaviour of the modelled system