Is that true...?

Abstract

This paper presents a theoretical perspective on patterns derived from epistemology and theory of science. We argue that patterns are specific kinds of theories and that the process of pattern mining is similar to scientific discovery. Exploring the concepts of induction, deduction and abduction with respect to patterns, we reflect upon common methods of pattern mining in the pattern community. This allows for a critical discussion of the level of confidence and corroboration of patterns. We suggest new research questions on the mining and evaluation of patterns. For the scientific scholar the paper offers arguments that pattern mining is a research process with outcomes as reliable and sound as other scientific procedures. This justification is needed to establish the pattern approach as a scientific methodology beyond the scope of the pattern community. For the pragmatic pattern practitioner, e.g. users and authors of patterns, this paper encourages the critical reflection on the pattern concept. Patterns are not triedand-true per se, just like theories they have to be subjected to empirical tests. Understanding the epistemological nature of patterns is crucial to derive criteria for pattern quality, e.g. the degrees of corroboration, and the limits of objectivism -especially since patterns are not only descriptive documentation but normative instructions, designed to have an impact on shaping our environments.

Improving students' learning in software engineering education through multi-level assignments

“…Design patterns are usually discovered in the process of pattern mining, the main task hereby is to "expose the invariant structure and discriminate it from the surface structure (the non-essential features)" [Kohls and Panke 2009]. There are different approaches to pattern mining, whereby inductive inference-the observation and analysis of good examples [Alexander 1979]-is probably the most common one and typical for qualitative research [Kohls and Panke 2009].…”

“…There are different approaches to pattern mining, whereby inductive inference-the observation and analysis of good examples [Alexander 1979]-is probably the most common one and typical for qualitative research [Kohls and Panke 2009]. This approach was also applied in this work.…”

“…This offers a few advantages compared to more traditional practice reports or theoretical descriptions [Laurillard 2012] and is a valid scientific format for formulating a theory about a concrete experience [Kohls and Panke 2009]. Christopher Alexander states that "Each pattern describes a problem which occurs over and over again in our environment, and then describes the core of the solution to that problem , in such a way that you can use this solution a million times over, without ever doing it twice" [Alexander et al 1977].…”

Using pattern mining for competency-focused education

“…This requires some certain level of experience in this domain. There are different techniques which can be successfully applied for pattern mining, among which the retrospective approach, social approach, artifactual approach, and pattern-mining workshops [11]. The last technique often is used in combination with other techniques, and can therefore be considered as one of the most applied techniques for collecting patterns.…”

“…The last technique often is used in combination with other techniques, and can therefore be considered as one of the most applied techniques for collecting patterns. Common to all approaches is the use of induction, also typical for qualitative research [11].…”