Ontology-Oriented Software Effort Estimation System for E-commerce Applications Based on Extreme Programming and Scrum Methodologies. (14th February 2019)

• Record Type: Journal Article
• Authors: Adnan, Muhammad; Afzal, Muhammad; Asif, Khadim Hussain
• Is Part Of: Computer journal. Volume 62:Number 11(2019)
• Page Start: 1605
• Page End: 1624
• Publication Date: 2019-02-14
• Subjects: cosmic function point -- e-commerce application -- functional size measurement -- knowledge reasoning -- knowledge management -- ontology -- software effort estimation system -- Scrum -- XP; Computers -- Periodicals; 005.1
• DOI: 10.1093/comjnl/bxy141
• Languages: English
• ISSNs: 0010-4620

Abstract

Abstract: Presently, software industry is severely suffering from inaccurate effort estimation and inadequate unstructured or semi-structured project history management. In fact, both are difficult to accomplish and hence badly impact the software projects. We proposed improvements in the effort estimation and the project history management of e-commerce projects focusing on Extreme Programing (XP) and Scrum methodologies using ontology models in our software effort estimation system. Proposed system infers suitable estimate in the form of time, resources and lessons learnt as per the project leader's requirements by using description logic and HermiT reasoner. To validate our approach, we have performed a case study comprising 20 Business-to-Consumer (B2C) web projects and performed comparative analysis on the collected efforts in both XP and Scrum contexts by applying (Mean Magnitude of Relative Error) MMRE and PRED(25) prediction accuracy measures. Likewise, software functional size of understudy e-commerce projects was measured using COSMIC functional size measurement methodology. Regression analysis of relations among actual COSMIC function points, estimated effort, and actual effort spent for the projects show better significance-F and R 2 values for our approach. The comparative results show that overall proposed approach provides accurate estimates and significantly improves over planning poker and delphi methods by 10% and 30%, respectively.