Research on interactive behavior analyzing in new-type distributed software system

In new-type distributed software environment, the interactive behaviors produced by loosely-coupled software entities behave of emerging, dynamic, accidental, correlative and repeating properties. The paper tries to extract them into definable and relatively stable behavior analysis model between behavior process and corresponding effects. Software behaviors are investigated in concrete context according to historically experiential knowledge and currently practical evidences, Multi-Entity Bayesian Network (MEBN) is adopted to represent and analyze complicated software behaviors. Multi-grain knowledge model reuse, quick network model construction and flexible reasoning technology of MEBN satisfy the demands of behavior representation and analysis in new-type distributed software environment. Simulation experiment confirms that the method presented in this paper has unique advantage over others.