Research Of Credibility Of The Interactive Entity's Behavior In Distributed System

With the distributed software widely used in the telecommunications,finance, and trading platforms, etc, the nature of its credibility such asavailability, reliability and security has received higher expectations anddemands by people. To solve the increasingly prominent issue of thecredibility of distributed software, we need to implement effectivemonitoring to its physical interactions and analysis its intention.Thedistributed software shows its characteristics of a large, loose aggregate andcomplex behavior. In order to analyze the interactive behavior of itsintention effectively, the thesis carries out its research progressively fromthose aspects such as monitoring technology, sequence generation,sequence decreasing, sequence model developing, sequence comparing andintention analysis in a systematical way.Adopting AOP technology and reflective technology monitoring togenerate interactive events, In this thesis, the event is taken as a startingpoint and the sequence of events which reflect the physical interactions asthe analysis object. Cyclic reduction algorithm with the sequence of actionshave been used to extract the sequence of abatement streamlined but alsoreflects the physical behavior of sub-sequences. Template-based graphmining algorithm is used to plan the training sample data and enrich theknowledge base. The use of Ant Colony Algorithm to the intention analysisof the interactive behavior entity. In this paper, physical interactionbehavior analysis method can be used to find the bottlenecks whichconstrains the system intuitively and analyze the physical interaction offraud, cheating and other abnormal behavior, so that analysis of behavioralintentions can truly become the guiding dynamic evolution of softwarereliability and the"helm"of its continuous optimization.The major tasks of this thesis include:(1) The physical interaction behavior lead to many loop subsequencewhich causes great difficulties to the excavation and analysis of behavior,so the thesis proposes a template mining classification strategy, through theoperation of loop to remove duplication part automatically so that thebehavior template stored in the Knowledge Base becomes not only simplebut also has a broad representation so that it is convenient for the follow-up analysis of online behavior intentions.(2) With the continuous running of the system, the number of thetemplates increases rapidly. If the matching accuracy and efficiency is nothigh, it is difficult to satisfy the effectiveness of the system but increasedthe burden on the system. The ant colony algorithm with time and space inthe thesis has the characteristics of low complexity and conducive toovercome the efficiency problem in the matching algorithm and can moreaccurately analyze the intention of physical behavior.(3) Finally, an application case and a simulated experiment are giventhrough performance testing and analysis to validate and evaluate theunique advantage of the analysis method of interactive behavior suggestedin this thesis.