Scope complexity management when planning public sector software projects: adaptive framework with deep neural network science

Abstract

“Scope Complexity Management in Planning Public Sector Software Projects: Adaptive Framework with Deep Neural_Network Science” Planning for public software projects is a daunting task, and various challenges come along in this phase. While in most cases, planning is put under the umbrella of management with the focus on people and technology expectations, the alignment of project goals with stakeholder expectations is another hurdle, potentially inducing the realization of project value and success of the project. As a measure, the emerging practice within the software producing industry is upfront management for understanding stakeholder expectations and value determinants. Although value proposition on these projects has been widely studied, there is still a high reoccurrence of perennial project management challenges that constrain success. These are mostly traced back to how the project was planned. Within the commercial software development circles, planning is for mitigating budget excesses, schedule slippages, and value propositions that are relatively non-protracted. However, when it comes to public sector software projects, it is an uphill task. These projects are characterized by having a plurality of stakeholders usually originating from different backgrounds and problem domains. This non-uniformity and plurality create a disaster as planners fail to align expectations to project goals. To address this challenge, this study developed an “Adaptive Scope Complexity Clarification Framework (AFSCOP)” which stands to provide appropriate guidance in planning public sector software projects. AFSCOP manages scope complexity when planning public sector software projects using Recurrent Neural Networks with Network Science (RNN_ Network Science). Predictive modelling using Artificial Intelligence (with Recurrent Neural Network) Network Science (with Network Models via Community Detection Algorithms) and Requirements Engineering were used to design, determine and assess the complexity within models and requirements engineering architecture (REA). The proposed Adaptive framework (AFSCOP) was partially tested on Uganda Electronic Medical Records (EMR). The results demonstrated accurate and reliable performance for the techniques of requirements classification with Recurrent Neural Networks and requirements clustering with the Scale-Free Network model. The theoretical, practical, and methodological contribution of AFSCOP is modelling the magnitude of complexity on the product side than the general project scope dimension. In part, it extends to examine how clarity of product scope and project goals can be harmonized as a crucial observable planning metric. As a data-driven approach, it is instrumental in identifying and predicting potential project success inhibitors, thus making a significant contribution to the field of systems engineering, software development, and project management.