Pattern mining is a term used in the pattern community to describe the process of identifying or discovering patterns. To identify new patterns, usually an informal or ad hoc process of finding patterns (e.g., in existing software systems) is used. This paper reflects on lessons learned and methods used regarding the integration of software pattern mining with qualitative research methods during our work on a pattern language for process-driven and service-oriented architectures (SOAs). This pattern language aims at closing the conceptual gap between business architecture and software architecture with a focus on process-driven solutions. In this paper we emphasize the notion of understanding patterns as sociological phenomena of problem solving behavior. We further introduce a systematic approach for pattern mining based Glaserian Grounded Theory techniques. This approach has been applied for mining the pattern language for process-driven SOAs. This work may also contribute to a better empirical grounding of software pattern mining. We will illustrate our approach using the pattern language for process-driven SOAs as a pattern mining case study.

Top