PROV-Process: proveniência de dados aplicada a processos de desenvolvimento de software

Abstract

The software development process can be defined as a set of activities, methods, practices and transformations used to develop and maintain the software and its related products. A simplified description of this process is called process model, which defines the activities for the development of software, product specifications of each activity and the indication of the roles of the people involved. The implementation of these processes generates important data on it. The proper analysis of the history of this data may result in the discovery of important information, which can contribute to the understanding of the process and therefore contribute to its improvement. The word provenance refers to the origin or source a particular object. In computer terms, provenance is a historical record of the derivation of data that can assist in the understanding of the data and / or the current record. This dissertation presents a proposal for an architecture that, through the use of data source models, combined with an ontological model and data mining techniques, aims to identify improvements in software development processes and present them to the project manager. This tool, by importing the process execution data, feeds a relational database, modeled based on a provenance model. These data are loaded into an ontology model and into a data mining file. Upon this loading, the data are processed by an inference machine, considering the ontological model, and also by an algorithm that integrates classification and association rules in data mining. The result of this analysis can presents points to improvements in the software development process. The proposed architecture is based on related work, which selected from the execution of a systematic review.