The growing complexity of software has posed challenges for the design, understanding and maintenance of software. Software developers often spend a lot of effort on understanding how software components are connected. Through analysis of the data obtained from a real software application, this paper demonstrates that network analysis approaches can help understand software component interconnections and their impact on software quality. Additionally, this paper shows that there is a correlation between the number of bugs found in an application and a set of measures used in component (interconnection) networks. Furthermore, a case study showed that the software performance improved over two releases as a result of improving the component network.

An increasingly important class of software-based systems is platforms that permit integration of third-party components, services, and tools. Service-Oriented Architecture (SOA) is one such platform that has been successful in providing integration and distribution in the business domain, and could be effective in other domains (e.g., scientific computing, healthcare, and complex decision making). In this paper, we discuss our application of SOA to provide an integration platform for socio-cultural analysis, a domain that, through models, tries to understand, analyze and predict relationships in large complex social systems. In developing this platform, called SORASCS, we had to overcome issues we believe are generally applicable to any application of SOA within a domain that involves technically naïve users and seeks to establish a sustainable software ecosystem based on a common integration platform. We discuss these issues, the lessons learned about the kinds of problems that occur, and pathways toward a solution.