Matrix-based project dataset parsers

Abstract

There are several existing project datasets, which involve separate data sources for simulated and real projects, individual and multiprojects, and single- and multimodal attributes. In addition, their file structures are heterogeneous; therefore, scholars can usually use only one dataset to test a proposed scheduling or resource allocation algorithm. Since the internal structures of these projects are also very different, it is difficult to ensure that an algorithm optimized for a given type of project will also perform well on projects with other structures. The proposed parsing method supports researchers in:reading several types of projects: simulated, real, individual, and multiprojects, as well as single- and multimodal attributes;•considering the priorities of activities and the flexibility of their dependencies, which is essential for modeling the structural flexibility employed by agile, hybrid, and extreme project management approaches;•building a large project database for testing and comparing different scheduling and resource allocation algorithms.