Research On Approach Of EPMM-Based Software EvolutionProcess Automation

Software process execution is the integration point of software process engineering and software projects engineering, which is also the engine of the activities of software projects engineering. The activities of software development projects is actually guided and controlled by software process execution. Software process execution is the key issue in the field of software process research. The execution of software evolution process guides and controls the software evolution activities.Execution of the software evolution process can be automated, semi-automated or manual. The software evolution process automation is a software evolution process which is completely automated execute by computer.Because of the participation of humans, it is of great difficult to achieve complete automation in the near future. This thesis discusses the approaches of software evolution process automation based on EPMM(EvoIution Process Meta-Model) that interact with the external environment. For one thing, the approaches are considered as an important technique which is used to boost productivity of software and at the same time improves the quality and the efficiency of software evolution. For another, the approaches can dynamically create the knowledge base of EPM (Software Evolution Process Model) and its execution model (EXECUTED_EPM:Executed EPM) which provide the data for improving the original EPM.This thesis,which combines the advantage of software automation technique and logic programming languages of creating rule-based expert system, proposed automated methods and related technologies of software evolution process model. The use of this method to convert software evolution process model to execution model which represented by facts and rules of Prolog, and also constructs knowledge base of software evolution process. Performing the execution model of software evolution process can be considered as reasoning on this knowledge base and dynamically updated knowledge in knowledge base of EXECUTED_EPM during execution process of EXECUTED_EPM, this is also the automatic accumulation process of knowledge of software evolution process.The thesis focuses on EPMM and EPM generated by modeling methods. The selection criterionof activity classification approach is presented and the classification approach for EPM activities is proposed through systematically analyzing the dynamic characteristics of EPM. The thesis proves the activity type set in EPM is complete set. The method of creating EXECUTED_EPM is proposed through analyzing logic programming language based rule representation and reasoning method. In the thesis, the EXECUTED_EPM instantiation method is proposed, the EXECUTED_EPM rule engine is designed and the converter from EPM to EXECUTED_EPM is implemented. The main research works of this thesis include the following aspects:(1) Propose the method of the activities’classification. While systematically analyzing the floating process of the activities’control information of the process level in EPM, this thesis study the execution mechanism and resources allocation policy of the pre-condition resources and the post-condition resources of the activities in EPM and propose the selection criteria of classification. The thesis also proposes pre-condition classification method, the post-condition classification method and pre-and-post-condition classification method of the activities.(2) It is proved that the premise type and post-condition type of any EPM obtained by EPMM based white-box modeling method are complete sets. Using mathematical induction method, the pre-condition type and post-condition of software evolution process model obtained by using four basic block to refine the initial activity are firstly proved to be complete sets. Then, prove that the pre-condition type and post-condition of new software evolution process model obtained by white-box modeling method at any level are complete sets.(3)The conversion method for EPM process layer to EXECUTED_EPM is proposed and the transformation rules and algorithm is designed. The transformation rules are proposed for each equivalence class obtained by the pre-condition type and post-condition type. The Transformation rule set is proved to be the complete set and the algorithm for converting EPM to EXECUTED_EPM is implemented.(4) The conversion method from the activity level and the task level to EXECUTED_EPM and the hierarchy of the transformation rules of EXECUTED_EPM are proposed. This thesis also proposes the input resources and output resources of the activity level and the transformation rules of roles of the task level and also the transformation algorithm is designed. The conversion method and algorithm of the global level are proposed.(5) The rule engine of the execution model of EPM is designed and implemented. The resources preparations method and the algorithm of EXECUTED_EPM are proposed. The method and the rules of capturing the concurrent executing activity in EXECUTED_EPM are proposed.(6) The model transformation tools of software evolution process is designed and implemented.