DNRF Chair: Professor Emilia Mendes

The Software Effort Estimation (SEE) is one of the most important areas in Software Engineering, being investigated for 50+ years; however, despite its importance, there are still significant shortcomings, which this research will address.

The first shortcoming relates to the lack of an all-encompassing software effort estimation taxonomy. Taxonomies enable scientists to share a common vocabulary and understanding, ease cognitive reasoning by using a higher-level abstraction than the instances themselves and help identify research gaps and opportunities.

The second shortcoming relates to the lack of a comprehensive synthesis of evidence of software effort estimation research in Software Engineering, including both completeness (SEE corpus) and depth (the means for deep understanding of the SEE complex phenomenon). Such comprehensive synthesis of evidence is paramount to inform both research and practice.

Therefore, the purpose of this research project is to significantly improve our scientific understanding of Software Effort Estimation, by means of a pioneering solution including an effort estimation gold standard taxonomy and the meta-synthesis of the effort estimation corpus using a technique that quantifies the relationship between variables probabilistically. We coined this meta-synthesis solution probabilistic meta-synthesis of evidence.

The research involves the proposal a gold standard taxonomy for SEE, its use with Natural Language Processing and Artificial Intelligence to classify and extract data from a large corpus of SEE studies, and finally the use of such classified and extracted data, in combination with Bayesian Network, to build a probabilistic synthesis of evidence for SEE.