Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

The DeltaProcess approach to systematic software process change management

: Soto, M.
: Rombach, D.; Liggesmeyer, P.; Bomarius, F.

Fulltext urn:nbn:de:0011-n-1342659 (2.3 MByte PDF)
MD5 Fingerprint: 5365e89d9bfa27e34e3a14aad40c017e
Created on: 7.7.2010


Stuttgart: Fraunhofer Verlag, 2010, XI, 172 pp.
Zugl.: Kaiserslautern, TU, Diss., 2009
PhD Theses in Experimental Software Engineering, 31
ISBN: 3-8396-0140-1
ISBN: 978-3-8396-0140-2
ISSN: 9424-8737
Dissertation, Electronic Publication
Fraunhofer IESE ()
Software Engineering Process Group (SEPG); Software-Prozessdokumentation; Modellierungsexperte; Modellierungsforscher; software engineering; software process; software process modeling; model comparison; versioning model

Software process models evolve over time, for reasons ranging from process improvement to adaptation to new laws, norms and standards. Models commonly used in industry are often large and complex, containing sizable amounts of textual material as well as intricate networks of internal and external relations. For this reason, managing the changes made to an industrial process model is a difficult endeavor, comprising concrete tasks such as collaborative team work, review of changes for quality and compliance purposes, explicit maintenance of compliance to norms and standards, and creation and maintenance of model variants, among others. The vast majority of these tasks are not well supported, neither methodologically nor technologically. Indeed, most of these activities must currently be performed in a mainly manual, tedious, and error-prone way. Arguably, this lack of support seriously limits the adoption of process improvement methods in an organization.
The software development community has extensive experience in dealing with problems similar to those just mentioned. The usual solution in the software domain involves the maintenance of a version history, and, especially, the use of comparison algorithms to determine the differences (changes) between versions of a program. Although the traditional software version bookkeeping concepts (revisions, variants, branches, etc.) can be easily adapted to the process modeling case, the algorithms used to compare software versions are limited to text files and are thus unsuitable for the complex structure of process models.
In order to fill this gap, this dissertation proposes the DeltaProcess approach for process model comparison. The central features of this approach are (a) its ability to deal with models of the size and complexity typically found in industry; (b) its genericity, namely, its ability to support a wide variety of model schemata, thanks to the use of a general meta-metamodel; (c) its flexibility with respect to the types of changes that are recognized, based on formal, declarative comparison patterns, allowing for automated,focused difference analysis of process models; and (d) its flexibility to integrate arbitrary comparison algorithms and visualization mechanisms.
The DeltaProcess approach was validated in the context of three studies related to the maintenance and tailoring of large industrial process models. The validation was aimed at showing that (a) DeltaProcess is efficient enough to handle process model instances of the size and complexity typical in industry, (b) DeltaProcess can be adapted with reasonable effort to support new model schemata, and (c) DeltaProcess can be adapted with reasonable effort to support new types of changes. The validation-based on several realistic comparison tasks applied to hundreds of industrial model instances-shows comparison and merge times in the range of a few seconds to a few minutes, and adaptation efforts, for both new schemata and new types of changes, of less than two weeks of work for the majority of tasks.