Supporting software processes analysis and decision-making using provenance data

Abstract

Data provenance can be defined as the description of the origins of a piece of data and the process by which it arrived in a database. Provenance has been successfully used in health sciences, chemical industries, and scientific computing, considering that these areas require a comprehensive traceability mechanism. Moreover, companies have been increasing the amount of data they collect from their systems and processes, considering the dropping cost of memory and storage technologies in the last years. Thus, this thesis investigates if the use of provenance models and techniques can support software processes execution analysis and data-driven decision-making, considering the increasing availability of process data provided by companies. A provenance model for software processes was developed and evaluated by experts in process and provenance area, in addition to an approach for capturing, storing, inferencing of implicit information, and visualization to software process provenance data. In addition, a case study using data from industry’s processes was conducted to evaluate the approach, with a discussion about several specific analysis and data-driven decision-making possibilities.