| |
|
|||||
International Workshop on Requirements Engineering Patterns – REP’04September 6, 2004
|
||||||
| Agenda | Summary of Presentations | Pattern Analysis Working Groups | Program Committee | Contact | ||
The first International Workshop on Requirements Engineering Patterns (REP04) was held on Sept 06, 2004 in Kyoto in conjunction with the 12th International IEEE Requirements Engineering Conference, RE04. The workshop aimed to investigate the potential of using patterns for sharing RE knowledge and experience among organizations who are in the process of adopting RE.
The workshop was attended by twenty practitioners from industry and research. The morning session featured three presentations which examined the potential and the role of patterns for knowledge exchange. The afternoon was reserved for discussion groups: case studies which were contributed from the workshop attendees were examined for common RE practices with the attempt to express these practices as patterns.
Download Workshop Summary (PDF)
Agenda |
Morning Session
Afternoon Session
Summary of Presentations |
The three presentations of the morning session revealed three different viewpoints on requirements engineering.
Ryuichi Hosoya opened the workshop with "A Thought on the Possibility of Requirements Patternization from the Viewpoint of Alexandrian Patterns". He observed that many IT systems, especially commodity systems, are delivering good value (i.e. respond to their users' needs) although they have not been developed according to actual user requirement specifications; moreover, he challenged that many users are not even able to specify their real needs in advance. As on the other hand from an abstract viewpoint many IT systems are comparable in their structure and perform comparable tasks, he concluded that also the underlying requirements should have comparable structures. He proposed to capture these commonalities in requirements patterns which should then be used to define and justify IT systems. This way, the specification procedure could be enhanced while at the same time improving the overall specification quality.
Kathrin Lappe was introducing the results of "The Working Group on Requirements Engineering Patterns of the German Informatics Society". The group developed a method for comparing case studies from projects and extracting patterns of recurring successful requirements engineering activities, and it created a collection of fourteen RE patterns which have been identified by that method. The patterns are intended for teams which are in the process of adopting RE and help the teams to shorten the RE learning curve. Ideally, reading a pattern would correspond to a discussion with an (absent) RE expert. A Web-based RE pattern repository was under development which should enable access to RE patterns based on selected project tasks or specific roles or conditions.
Eiiti Hanyuda gave a brief position statement in which he recommended expressing requirements for information systems in terms of business models. He proposed to use a reference model - "(4+1)x1 Views on Business Process Modelling" - for verifying the requirements model and ensuring complete and consistent specifications.
The presentations, though following the same objective - finding a pattern-like mechanism for improving the requirements process - pointed out very different approaches to and interpretations of the topic: Patterns were envisioned to
This variety of interpretations was felt as a confirmation of the general demand for "something like patterns" for RE process improvement.
Pattern Review |
The afternoon session started with briefly reviewing two preliminary versions of patterns which have been contributed by the Working Group on Requirements Engineering Patterns:
The discussion first examined whether the patterns were understandable, instructive and applicable (resp. useful), and if they contained relevant information providing added value compared to e.g. textbooks. It was felt that the patterns were clearly understandable and addressing relevant topics, but should contain more information on when and how to apply them, supported by criteria for successful pattern application.
In the course of the discussion it was proposed to add rationales and cost-benefit-arguments to the patterns to help project managers acquiring and justifying RE resources. It was also observed that the two patterns were related in the sense that using prototypes is a method which could be moderated by a documentation manager, or better a requirements engineer. It was confirmed the attendees in their career have experienced similar situations as described by the patterns and that they made comparable observations.
Pattern Analysis Working Groups |
The major part of the afternoon was reserved for interactive work, aiming to practically exercise what has been discussed so far. The workshop attendees were organized into two working groups for sharing experience and identify new patterns. Two topics were agreed upon for the two working groups: one group discussed envisioning a common set of requirements with inhomogeneous stakeholders, the other addressed issues of requirements reuse. The two groups experienced very different successes.
The first group had a lively discussion, but was not able to identify pattern candidates as the reported experiences have been too different, including small-scale moderating efforts and building ontologies. The group then examined how the pattern approach fitted their discussion: it was felt that the notion of patterns has been beneficial to the conversation as it helped the presenters structuring their reports, and it was inferred that patterns would be helpful for authors in general, however it was not possible to conclude further benefits.
The second group had an intensive exchange of experience and identified two new pattern candidates - more candidates would have been found if the discussion could have lasted longer. The group followed the method from the Working Group on Requirements Engineering Patterns (WGREP) which has been introduced earlier in the day. Six different observations with substantial overlap were reported, including projects from automotive industry and suppliers, information system development, plant construction and computer hardware product development. The observations were then analyzed for recurring successful RE practices:
The group successfully used the proposed pattern analysis method, and especially the core element, the so-called pattern vector model has proven to be an appropriate instrument for capturing and communicating experience.
It is interesting to note that the group set out to discuss requirements reuse, which has also been the key element of the reported observations, but that during discussion it turned out that the commonalities of the projects lay in their way of requirements analysis. Similar effects have also been observed by WGREP.
Conclusion |
Did we meet the expectations? - was the final question to be answered by a brief feedback round. As reflected by the presentations, the term "Requirements Engineering Pattern" has been associated with a variety of meanings, all of which were felt relevant for RE. During the workshop the different approaches converged into a broader picture. The results of the afternoon working groups were felt to be encouraging for further RE pattern activities, and yes, the expectations have been vague, various and met.
Program Committee |
| Ian F. Alexander | ScenarioPlus, UK |
| Daniel M. Berry | U Waterloo, Canady |
| Lars Hagge | DESY, Germany |
| Eiiti Hanyuda | Mamezou, Japan |
| Frank Houdek | DaimlerChrysler, Germany |
| Natalia Juristo | U Madrid, Spain |
| Barbara Paech | U Heidelberg, Germany |
| Camille Salinesi | U Paris, France |
| Hironori Washizaki |
Nat. Inst. Informatics, Japan |
Workshop Organizers |
| DESY, Germany | |
| Frank Houdek | DaimlerChrysler, Germany |
| Barbara
Paech |
U Heidelberg, Germany |