Both disciplines – AI and SE – share the need to answer with structure (either called variables or data structures, or information or knowledge) when confronted with a real world scenario. This article discusses how SE Knowledge may be maintained and constantly cultivated among software engineers, adopting the well known Wiki concept based on approaches focussing ontologies. However, the main objective is not to destroy the desired lightweight character that Wikis stay for and have rather been successful for. In the long run it should result in a higher quality of software systems due to a better availiability and viability of SE-specific knowledge. 1 Creating systems for system creators The success of a system requires the art to really understand the problem it is supposed to solve. This understanding needs to be tranferred into a system with both, a considerable match of the system features with the necessities of the scenario in which it is used and with fatal flaws absent. System implementations are vulnerable to apparently minor mistakes, so that even close to perfection, artifacts can completely fail; a single minor mistake —a line of code, or an asset in the knowledge configuration— may destroy everything. [Bat02a] This does not only refer to a possible system malfunction. “Nonuse is a major reason for systems’ not achieving their designers’ objectives.” [SMC95] Consequently, SE and AI share the vital interest, we call reuse, i.e. to analyse the parameters of successful projects and to remember mistakes, in order not to commit them again — be it based on knowledge assets, expertise, code snippets, or anything else (cf. [JAD02,Kar95,MYA95]). Approaches to reuse are well known from knowledge management. In their details they feature a great bandwidth and —because also here the system must match the scenario— equally so do the approaches and systems to solve them. [AvE04,GDK04,JAD02,MA04] The problem approached in this paper is how to create a product plus respective methodical knowhow that can support teams of software engineers by increasing reuse. (Note, that all of the above certainly applies to this endeavor.) Like almost always in KM and SE, because the scenarios of reuse always comprise the people, i.e. individuals and their social system, the problems (of reuse) cannot be formally denoted (cf. “wicked problems” [RW73,Con05]). Note, that the problem at hand is even not one specific problem, but a whole cohort of problems. However, the partners in the consortium of the project RISE [pc05] want to use the product inhouse, but the industry partners, also really want to market it. They dispose of a Content Management System (large scale, with a strong case based retrieval (CBR) search [VWB98]) and a search engine (based on the vector space paradigm (VSP) [AHK03]), respectively, and want to combine their systems with Wikis. With one of the about 200 currently available Wiki systems [EG05] some of the 19200 software producing companies in Germany [SWR00] will probably be rather successful. It is further known, that there exist problems of reuse that cannot be successfully addressed with systems at all. A third party of companies do not have reuse problems or do not try to get better. For the fourth party a pure (“standard”) Wiki did or does not succeed, some additional intelligence is required — modules and/or measures. The cases where such modules can be based on CBR or VSP are clearly in target. To sum up, the solution sought is: a Wiki, extra modules, and a method. 2 Storage of “Knowledge” – the OMIS promises A broad spectrum of problems can be tackled with Organisational Memory Information Systems (OMIS) — systems that provide the service of a central storage of information to a group of people. We have seen very tame and successful projects, e.g. a company where the files of finished projects were frequently lost. Their knowledge problem was easily solved by simply providing them with a rather simple database for their files of finished projects. OMISs are one, if not the main technical means of knowledge management since years. [ABH98] Basically, the new paradigm of a Wiki is also an OMIS, however, features an aditional strong collaborative component. [EG05] A milestone of successfull uses of OMISes, published in the scientific literature [O’L98], are the strategy databases of large consulting firms, where three firms have been explicitly examined: Arthur Andersen’s Knowledge Space, Ernst & Youngs Leading Practices K-Base, and Price Waterhouse’s Knowledge View. 3000 to 5000 strategy documents respectively have been reported to be continuously maintained with great diligence. Every consultant returning from a project in the field is set free for up to one week, gets assistance by dedicated senior technial writers and processes and prepares his experience in order to enter it into the strategy database and keep it up to date. Consultants, starting with the very 4 Even more so, as frequently not the best solution is the most successful. first time in the field, continuously experience how much their personal success depends on this database, so that there is a high enough motivation to invest the considerable required effort, time, and money into its maintenance. [O’L98] We are aiming at those cases, where such a formal and disciplined procedure is not possible. 3 Towards group Knowledge by means of a Wiki 3.1 Wiki: Capturing and Maintaining Knowledge Wikis allow everybody to provide his knowledge easily, i.e. informally. The whole of the usergroup manages the process of polishing it. They are already in wide use, also for real business e.g. for project management. [EG05] The collaborativeness that Wikis add to the basic OMIS, promises to outperform the pure storage of information in an OMIS or the pure support of collaboration. In some cases Wikis have remarkable success. “T-Systems, SAP, Web.de, British Telecommunications, Walt Disney Corporation, Motorola und Yahoo. These companies use Wikis in their intranets. ”[Gmb05] The current manifesto of the German green party was collaboratively edited based on a Wiki. [Jel05] Detailed examinations of the parameters specificly of Wiki success and failure are underway. However, there are points that can be made now already. “Perceived enjoyment and perceived ease of use are stronger determinants of intentions to use [information systems] than perceived usefulness.”[vdH04] The joint editing supports that users intensely deal with the material of the others and that the final result is agreed or accepted. Where technology fails, a Wiki lets its users do what is required. E.g. it is taken care of improving, say a convoluted contribution. No knowledge assets are systematically excluded, contributions can freely adapt to necessities. Going beyond business information, e.g. with homepages of people, a company wiki could improve the efficiency of the communication (so that the business profits). For the controller point of view, it might be interesting to note, that the Wikis movement seems to be sustainable and further that, continuously, there are many new feature requests collected from users many of which are implemented. However, the RISE consortium, all pilot users of Wikis and observers of the Wiki landscape, have witnessed a number of Wiki projects fail completely. 5 The others can find interesting references in [JAD02]. 6 Wiki is not genuinely new, as there were web-based so-called content management systems like e.g. ZOPE before. Funnily, Rhizome, who do not label themselves as a Wiki, are considered a Wiki from others. Recently on the first and greatest event of the Wikipedia community, the Wikimania http://wikimania.wikimedia.org/wiki/Programme, it was mentioned to us in the discussion in the workshop “Wikipedia as a learning community” (http://en.wikibooks.org/w/index.php?title=Wikimania05/CL1), that Tim Berners Lee in his original implementation not of the Semantic Web, but of the World Wide Web, had designed the http-protocol so that people viewing pages, could also edit and save them. This is a remarkable fact. The Wiki was almost invented then already. When the content grows, it can become very hard to maintain a good structure. Without a good structure, information resources get lost in chaos, and people are frustrated by searching. [Ton05,Dav04] The latter sections treat approaches to transcend the usability of current Wikis to problems, where the maintenance of the content and structure of a Wiki are hardly possible today. In order, these are the promises of Semantic Wiki, Wikitology, and of some main methodical measures.
更多
分享
引用
