Ontology-based methodology for error detection in software design

Abstract

Improving the quality of a software design with the goal of producing a high quality software product continues to grow in importance due to the costs that result from poorly designed software. It is commonly accepted that multiple design views are required in order to clearly specify the required functionality of software. There is universal agreement as to the importance of identifying inconsistencies early in the software design process, but the challenge is how to reconcile the representations of the diverse views to ensure consistency. To address the problem of inconsistencies that occur across multiple design views, this research introduces the Methodology for Objects to Agents (MOA). MOA utilizes a new ontology, the Ontology for Software Specification and Design (OSSD), as a common information model to integrate specification knowledge and design knowledge in order to facilitate the interoperability of formal requirements modeling tools and design tools, with the end goal of detecting inconsistency errors in a design. The methodology, which transforms designs represented using the Unified Modeling Language (UML) into representations written in formal agent-oriented modeling languages, integrates object-oriented concepts and agent-oriented concepts in order to take advantage of the benefits that both approaches can provide. The OSSD model is a hierarchical decomposition of software development concepts, including ontological constructs of objects, attributes, behavior, relations, states, transitions, goals, constraints, and plans. The methodology includes a consistency checking process that defines a consistency framework and an Inter-View Inconsistency Detection technique. MOA enhances software design quality by integrating multiple software design views, integrating object-oriented and agent-oriented concepts, and defining an error detection method that associates rules with ontological properties.