Industrial Maintenance Modelled in SOCCA: an Experience Report Tineke de Bunje' Gregor Engels2 Luuk Groenewegen2 Aart Matsinger' Mark Rijnbeek2 'Philips Research Laboratories Prof. Holstlaan 4 NL-5656 AA Eindhoven The Netherlands The Netherlands 2Leiden University, Computer Sc. Dept. P.O. Box 9512 NL-2300 RA Leiden e-mail: { engels,luuk} @ wi.leidenuniv.nl e-mail: { bunje,matsingr } @natlab.research.philips.com Abstract A large industrial process, software maintenance, has been modelled by using the process modelling language SOCCA. The paper reports about the experiences with this trial. In particulal; feasibility, expressiveness, quality, and the overall benejits of a formal SOCCA model are discussed and compared to the formerly existing informal process description. In order to illustrate the results, a well chosen process model fragment from the larger model is outlined in detail. It addresses in particular the human-intensive coop- eration within the process and shows the seamless combi- nation of technical components and human agent components in the SOCCA model. The main conclusions from this trial are that formal SOCCA models are suited to model realistic industrial processes and that due to an intrinsic modular structure of a SOCCA model even huge models remain reasonably readable and understandable. 1. Introduction Research into software process technology (SPT) has made great progress during the last decade. On one hand, software process modelling concepts and languages have been investigated intensively. They range from activity-ori- ented to goal-oriented approaches, from high-level, abstract modelling languages to low-level, programming language like approaches. They focus on certain aspects of a software process such as synchronization of activities or try to offer comprehensive approaches to model all aspects of a soft- ware process in an integrated way. On the other hand, tool support for modelling activities as well as process-centred software development environments have been developed. An overview of all this is given by the proceedings of the European as well as International workshop series on SPT as well as by the proceedings of previous ICSP conferences (e.g. [21, 17, 221). In addition, the programme of any major conference in the field of software engineering comprises at least one session on SPT. Furthermore, research consortia have been funded like the ESPRIT basic research working group PROMOTER to bundle up research activities [9]. While the research agenda on SPT still contains a long list of open questions, researchers in the field are convinced that the results reached so far are mature enough to be tested and employed in realistic industrial situations. But, since great success stories on employing software process technology in industrial contexts are still missing, industrial software process improvement departments are still hesitat- ing to apply research results in industry. Nevertheless, industry is keen to learn what the benefits of applying soft- ware process technology might be. Therefore, the software industry is running trial applications on transferring soft- ware process technology to industrial software development processes. This article reports about such a trial, that was performed at Philips Research in close cooperation with the Software Engineering and Information Systems (SEIS) group at Lei- den University. The basic idea was to set up a formal model for a real-life industrial process with the process modelling language SOCCA (Specification of Coordinated and Coop- erative Activities). This language has been developed by members of the SEIS group during the last four years. While having been tested in several small-size case studies before, the case study described here is the largest and most complex one using SOCCA. This case study was mainly done during a master thesis project as reported in [ 181. During this case study, the objectives of the industrial partner, Philips, and of the research partner, the SEIS group at Leiden University were to gain more insight in the fol- lowing points: feasibility: the possibility to formulate a SOCCA model for a large-scale industrial process; expressiveness: the details of a SOCCA model not only in relation to the software development process, but also in relation to the embedding thereof in the more general business process; quality: the quality of the SOCCA model compared to 13 0-8186-7719-8196 $5.00 0 1996 IEEE Authorized licensed use limited to: IEEE Xplore. Downloaded on November 2, 2008 at 12:12 from IEEE Xplore. Restrictions apply.