Analyzing And Predicting The Run-time Behavior Of Software In Electronic Trading

In the open network environment, the new distributed software has shown is different from the centralized software and traditional distributed software features, which are open dynamic, loose aggregate, complex, and in such an environment, the implementation of the software under the supervision of unprecedented difficulty. In this environment, the implementation of the monitoring and management of software has great difficulty. Electronic transactions in such an open network environment, the user in electronic transactions often need to deal with unknown customers; the behavior of users in electronic transactions mainly was reflected through the interaction of software. the software information is on the core of the software run-time behavior information,analysis the software run-time behavior, and predict the development trend of the future action with the analysis results. It can predict the behavior of abnormality and cheat effectively in the entity's strange interactive process and intervene the unreliable behaviors in advance so that serious consequences can be avoided, its is of great significance to the development of electronic transactions.The paper has the following work:(1) In the aspect of monitoring the behavior Information, with the dynamic AOP techniques to monitor the interactive event generated by the relevant software entity .When describing and analyzing software behavior, mapping the state changes monitored by the system to the event semantic meaning with the event as the basic unit.(2) In the aspects of the analysis of behavior sequence, responding to the problem m of methods for the existing behavior sequence analysis main concern non-repetitive sub-sequences, and behavior sequence of the template not simple enough, the paper presents a marked and complex behavioral sequence analysis method. In behavior analysis process, extracting the behavior sequence with the clustering technology, automatic identifying and removing repeat sub-sequence, recording the content and location of remove part; Excavating motif of the remove part, using regular expressions describing the behavior template to simplify the behavior template, this method improved the efficiency of behavior analysis effectively.(3) In the aspects of behavioral predicting, responding to prediction methods have been most suitable for traditional distributed software systems, and focused on the credibility of the interaction results, not given a good solution to the problem of the prediction of run-time behavior interaction, the paper presents a method of software behavior predicting based on Hidden Markov Model-based, extracted the behavior template by behavior sequence analysis as the observe sequence, predicting behavior with unknown events of good and evil with the backward algorithms, warning against malicious behavior timely,predicting the intention of the non-malicious behavior; Getting the best behavior intention sequence of the observation sequence with the forward algorithm, getting the occurred behavior intention sequence with the Viterbi algorithm and calculate the follow-up behavior intentions. The method can accurately predict the next possible event; can also take the initiative to predict the trend of follow-up behavior.(4) A prototype system of electronic transactions was designed, and an application case was specified, the performance testing and analysis verify the feasibility and effectiveness of the method.