Design And Implementation Of A Software Process Modeling Supporting Tool

Software process model is a static model that describes the relationship between developers,activities and artifacts involved in the software development process.Modeling software process is not only helpful for researchers and process managers to gain a comprehensive understanding of software process,but also can help to find the problems and mistakes in software process,so as to further improve the software process.As the software life cycle continues to expand,automation tools and practices are driving shorter software delivery cycles.In addition,a large number of automation tools supporting development are widely used,which leverages software process modeling at finer granularity and with higher fidelity.However,the existing modeling methods of software process are faced with a series of challenges.First,the data recording and modeling objectives are inconsistent,and the data recording of the software repository is mainly for the attribute information of the artifacts rather than the behavior information of the process,resulting in the inability to directly obtain the data describing software process.Second,there is undesired data quality and validity for modeling,which is reflected in a large number of inconsistent and incomprehensible information items in the data records.Third,the construction of process models is of low efficiency and poor fidelity.The construction of process models largely depends on the modelers' experience and their cognition of the process enactment.The cognitive modeling process is of low efficiency and associated with strong subjectivity,and it is difficult to restore the real process in an objective and concrete way.In order to cope with the challenges of software process modeling,a set of automatic construction methods of software process model is developed,and based on these methods,a tool named “Software Process Analyzer”(SPA)is implemented in this thesis,which is used to assist modelers in data processing and process modeling.For Challenges 1 and 2,the data processing module of SPA can extract,clean,integrate and transform data from different data sources with different contents and formats to generate a unified process event log.For challenge 3,SPA's process modeling module integrates multiple process mining and model evaluation algorithms,which can recover realistic and concrete process models.In addition,SPA can also provide a series of functions for process parameter analysis,such as parameter statistics and clustering,to facilitate modelers to more intuitively understand the detailed process of the target project.Finally,in order to validate the feasibility of the method and the validity of the tool,this thesis selects the Continuous Integration(CI)in software engineering as a case study.Using the proposed methods,we can acquire and process the CI process logs,conduct conformance check and parameter analysis,further validate the effectiveness of the corresponding software process model.Based on the case study,it can be concluded that the SPA tool can effectively assist modelers in modeling software process.