Establishing pattern sequences using stochastic processes with an application to organizational patterns
摘要
Software developers who solve recurring problems employ similar practices, which are documented and reused as patterns. Organizational patterns can fix organizational issues. Security patterns can be used to implement security measures into software or an organization. Expected solutions to complex problems of establishing security in the organization can be embodied in pattern sequences. Sequences of patterns that are expected to be used in practice cannot be established by applying stochastic processes, because multiple sequences may have the same probability of use. The symmetry of the relationship between patterns can be calculated using stochastic processes, and these symmetries can be used to distinguish pattern sequences that are truly expected to be used. This article presents two methods of establishing pattern sequences using stochastic processes. Both methods, based on stochastic trees and Bayesian networks, can be used to establish pattern sequences that are expected and unexpected in practice. The method that combines stochastic trees and symmetry of relationships between patterns can be used to establish sequences of conditionally independent patterns. The method that combines Bayesian networks and symmetry of relationships between patterns can be used to establish sequences of conditionally dependent patterns in multiple pattern languages or catalogs without constructing multiple stochastic trees. We used these methods to establish 19 sequences of organizational and security patterns. We show how expected sequences of organizational and security patterns can establish security in an organization. These two methods can also be used to establish expected sequences of patterns for engineering software for the cloud.