Reusable SOA Assets Identification Using E-Business Patterns

Abstract

Keyword-based search techniques are currently used to identify assets packaged for reuse. Unfortunately, these techniques are not suitable for searching large asset repositories, as they are known of having low precision and recall. One way to improve the search accuracy, is to identify reusable assets using their architecturemodels, since models based on similar architecture decisions, will most probably lead to developing similar solutions. However, informality and heterogeneity in designing architecture models remains a major obstacle for this identification approach. To overcome this problem, this paper proposes use of e-business patterns for reusable asset identification instead of architectural models. As e-business patterns are common to every designer, they could be used as an architecture design reference for comparing assets architecture layouts. Therefore, we propose amodel to capture such architecture layouts in a machineunderstandable format using a graph of the adopted business,integration, and application e-business patterns, thereby assets could be rapidly identified based on the matching status of their architecture layout graphs. We believe the proposed approach provides better precision and recall when compared to unstructured keyword-based approaches.