PRomisE2 - Recording and Displaying Situated Process Information in Knowledge Management Applications Marcel Hoffmann Informatics & Society, University of Dortmund Germany
[email protected] Thomas Herrmann Informatics & Society, University of Dortmund Germany
[email protected] Marc Diefenbruch ExperTeam AG, Dortmund Germany
[email protected] Thomas Goesmann Fraunhofer ISST, Dortmund Germany
[email protected]
Abstract: Employing collaboration technology for Knowledge Management some inherent challenges have to be met. One important prerequisite is that users adjust their patterns of interaction and cooperation and agree on the adjustments of others so they can practise cooperative knowledge management to their satisfaction. In dynamic environments, however, the patterns of cooperation must be continuously reassessed and revised. This paper suggest an approach to record and display process information in a way that makes transparent the patterns of interaction and produces navigable process displays that can be incorporated into work routines. We introduce a technical architecture, describe a concrete usage scenario and reflect on potential applications. Keywords: Knowledge Management, Process Oriented Navigation, Information Systems Category: H.3.5, H.4.3, H.5.3, J.4, K.4.3
1 Motivation Organizational Memory Information Systems (OMIS) [Stein and Zwass 95] have a certain characteristic: the way the software is used is not exhaustively inscribed in the technology. Effectively, technology for creative knowledge work – by nature – leaves alternatives for adoption. As a consequence, “acceptance in practice” needs to be supported by participatory design strategies, training, continuous improvement and internal marketing activities. In the case of collaborative systems – such as CSCW or CSCL (Computer Supported Cooperative Work / Learning) – more problems will
arise: The user’s routines are inconsistent. Mutual expectations do not match each other. Similar to the adaptation of the technical design to the users’ requirements, the users and their cooperation need to adapt to the new media, too [Orlikowski 92]. Surveys on KM detected that European companies consider user motivation as a critical factors for success in improving organizational learning and setting up systems for sharing knowledge [Mertins et al. 99]. Accordingly “incentivising” strategies are en vogue at many places. From our experience, however, we conclude that helping users to agree on and to continuously revise and adjust their patterns of interaction may bring about positive experiences of success, too. CSCW researchers generally agree that this strategy represents to a more sustainable approach to support acceptance of collaboration technology than making their usage mandatory or stimulating acceptance with extrinsic incentives [Grudin and Palen 95]. When it comes to coordinating different users the articulation of interaction patterns in terms of processes is a well-tried approach. We define processes as constellations of activities or events, which change the state of parts of a compound socio-technical system and are partially linked by causal or temporal relations. Due to the “untidy nature” of knowledge work [Davenport et al. 96], however, the transfer of process engineering methodology and technology to KM still seems to be difficult. Drawing on experiences with workflow and with document-centered KM systems, our work suggests a new way to employ process descriptions in knowledge work.
2 Requirements for Process Oriented Knowledge Management With regard to the technical support of process oriented KM we have distinguished two different perspectives up to now [Diefenbruch et al. 00]. • Active control of knowledge intensive business processes with workflow technology. • Process-oriented navigation of information systems storing process knowledge. In both cases various kinds of process knowledge are relevant and need to be stored and retrieved with regard to the current situation, e.g. instructions and guidelines, diagrams, or other work resources like applications, documents, templates and information systems. The approach introduced in this paper falls into the second category. Controlling the execution of activities on grounds of a formal description of the business processes, workflow systems store a lot of information about the current usage situation. Therefore workflow-based solutions are capable to filter resources relevant to the current situation. Unfortunately, many other applications that are applied in KM, like for instance groupware, document management, or content management systems, “are less aware” of the user’s current task and context. As a consequence users of these systems spend a lot of effort to locate and to access process related resources. Classification schemes, which are put in place to provide task specific support, do not meet this challenge. This approach usually imposes additional burden for knowledge suppliers explicitly attributing resources to processes or tasks. Furthermore, explicit categorization of contents with regard to process causes
maintenance costs and may hamper flexible process development. After all, the most important requirement to ensure acceptance of KM systems is to minimize all users’ burden from contributing and retrieving information. Contextualized and situated supply and the support of “articulation work” [Schmidt 99] are therefore critical requirements to focus on.
3 Supporting Situated Supply and Articulation of Process Knowledge PRomisE2 (the PRocess oriented OMIS navigator II) is a collection of software-tools designed to create, store, share, annotate and navigate process diagrams and process resources, building on top of a groupware, document management or an advanced management applications that provide functionality to store and retrieve process relevant documents or data. PRomisE2 uses two kinds of schemas to store process related information. Firstly, process patterns (pps) formulated in a simple XML process description language list activities, roles, and entities and record relations among these elements of the process. The process description language implements a subset of the SeeMe diagramming language [Herrmann & Loser 99] allowing to describe flexibly sequenced activities as well as their relation to entities, like documents or other artifacts and to roles. Any element of a process description can be specified by comments, a vagueness attribute indicating the level of completeness and level of certainty of the correctness, and a function call or URL. Another attribute describes the validity of the element with respect to context variables. This validity attribute works with the second kind of PRomisE2 data, the translations files (tls) that are used to substitute generic process information or to translate information from one context into another. . XML Document Current Situation
XSLT Parser
Situation variable
XSL Stylesheet Select and situate process
HTTP Client XML Document process pattern (pp) Validity attribute Generic node
XSL Stylesheet Generate process diagram
document Frames[0]
Frames[1] PRomisE2
Process Display
SVG Process viewer function CreateProcessStep
XML Document Translations (tl) Translation pattern
Figure 1: PRomisE2 architecture
3.1 Displaying Process Knowledge PRomisE2 provides situated process knowledge with XSL stylesheets that, firstly, select process descriptions (pps) matching the current usage situation, secondly, translate these patterns by substituting generic contents with contents matching the current usage situation, and thirdly, generate functions calls which render a svg process digram Figure 1 shows how the elements of PRomisE2 work together. In the first step the xsl-stylesheet select and situate process is comparing situation variables of the current situation and the process patterns’ validity attributes to identify patterns valid in the user’s current situation. In our sample implementation the user’s identity, her group membership, and data on the object the user was working on when she called PRomisE2 are available as situation variables. In the second step translation files (tls) are called to substitute generic information in the patterns. Finally the generate process diagram stylesheet is generating a sequence of functions calls which compose the svg image. The svg image is displayed in an additional window or frame of the user’s http client and shows descriptions of processes that match the user’s current situation as represented in the situation variables. Building on XML and using the XML related standards XSLT and SVG (W3C recommendations) PRomisE2 can be employed on top of many web-based systems. The current experimental implementation of PRomisE2 builds on the commercial KM system Livelink ® (www.opentext.com) that facilitated the implementation which functionality to export configurations of the system and stored contents in XML format. However, PRomisE2 can be adapted to other web based systems, too.
Figure 2: Sample PRomisE2 process display
Diagrams displayed by PRomisE2 inform the user about her scope of action and subsequent activities she may trigger. Furthermore elements of the diagram can be used to navigate to process related contents. For instance, if the symbol of a document is clicked the corresponding document is displayed. The experimental implementation allows navigating to contents stored in the Livelink system or on any other site in the WWW. Additional application servers’ functionality and contents can be integrated, too, most easily, of course, if the server provides an http-interface. Figure 2 shows a sample output of the displaying process knowledge procedure. In this case PRomisE2 displays the predefined process of interactions during a seminar. The process pattern “SeminarIuGFIT092002” was selected by the select and situate process stylesheet since it was labeled to be valid for any node characterized as part of the seminar. The display provides a navigation image of the process. Clicking on the role symbols, for instance, a vcard for the user is displayed; entity symbols lead to either the selected document or the folder containing the document displayed in the upper frame. 3.2 Recording process descriptions with PRomisE2 Since we believe that seamless transition between the primary work tasks and the creation and adaptation of process description is essential for the success of processoriented KM, we are currently integrating functionality to generate and adapt process patterns. The current prototype supports the creation of process patterns using diagramming tools and, of course, XML editors. Designing prototypes for an automatic generation of process patterns we draw on existing event logging functionality. Extending the event logging to associate events with process pattern identifiers we can derive and continue process patterns from ongoing use of the system. However, our experience shows that to generate valuable process patterns we cannot rely on automatisms alone. Therefore, we are now concentrating on two scenarios for interactive articulation of process knowledge: A. User adaptations of automatically derived process patterns (e.g. deleting, adding, annotating, renaming, and joining elements). B. Recording processes of user interactions as blueprints for similar situations. In the first scenario a servlet-engine that listens to and executes editing events, like deletion, annotation or addition of process elements (activities, roles or entities) provides simple editing functionality. Notably, we are distinguishing user adaptation requests from process editing requests. While user adaptation requests are translated into additional validity attributes that assure the unchanged appearance of the process for other users, the process editing requests change the process patterns for all users. For the second scenario we have implemented a process recorder, that associates events caused by a user with a process pattern, translates those events into a process step (user, activity and object), and appends this triad to a process patterns. Passed to another user the process pattern identifier functions as a baton and allows to record cooperative processes, too.
4 Discussion We tested the PRomisE2 framework for supporting acquisition processes in a consulting company and for supporting knowledge processes in a seminar. Both studies relied on predefined processes generated by an editor. The displays were generally accepted as useful navigation aids. However, the studies discovered requirements for user adaptation and user configuration of process patterns, too. Therefore we will now evaluate how the described functionality for recording process descriptions changes the usage. In the early days of workflow systems [Ellis and Wainer 94] criticized the technology for being too rigid and getting in the way of work procedures rather than supporting practice. We still observe that most workflow-systems do not provide enough flexibility to control creative knowledge processes. PRomisE2 provides navigable images of work processes rather than controlling their execution. Obviously this approach limits the functionality. PRomisE2 cannot compare the actual progress of a process instance against a pattern, but it guides to resources, recommends potential partners, and suggests certain activities. In contrast to workflow models that are generally implemented as scripts that impose certain activities on users, the PRomisE2 display resembles what [Schmidt 99] referred to as maps that represent resources activity draws upon. We believe that these resources provide valuable assistance for users of KM applications and offer an innovative, process oriented perspective on knowledge management.
References [Davenport et al. 96] Davenport, T. H.; Jarvenpaa, S., L.; Beers, M.C.: “Improving Knowledge Work Processes”; Sloan Management Review 34, 4 (1996). 53-65. [Mertins et al. 00] Mertins, K.; Heisig, P.; Vorbeck, J.: “Knowledge Management. Best Practices in Europe”; Springer, Heidelberg (2000) [Diefenbruch et al. 00] Diefenbruch, M.; Hoffmann, M.; Misch, A.; Schneider, H.: “Situated Knowledge Management – on the borderline between chaos and rigidity”; Proceedings of PAKM (2000). 8-1 - 8-7. [Ellis 94] Ellis, C.A.; Wainer, J.: “Goal-based models of collaboration”; Collaborative Computing. 1, 1 (1994). 61-86 [Grudin and Palen 95] Grudin, J.; Palen, L.: “Why Groupware Succeeds: Discretion or Mandate?”; Proceedings of ECSCW (1995), Kluwer, Dordrecht (1995). 263-278. [Herrmann and Loser 99] Herrmann, T. ; Loser, K.-U.: “Vagueness in models of sociotechnical systems”: Behavior & Information Technology 18, 5 (1999). 313-323 [Hoffmann et al. 02] Hoffmann, M.; Goesmann, T.; Kienle, A.: „Analyse und Unterstützung von Wissensprozessen als Voraussetzung für erfolgreiches Wissensmanagement“; Abecker et al. (eds.): „Geschäftsprozessorientiertes Wissensmanagement“; Springer, Heidelberg (2002). To Appear. [Orlikowski 92] Orlikowski, Wanda: “The duality of technology: Rethinking the Concept of Technology in Organizations”; Organization Science 3, 3 (1992). 398-427.
[Stein and Zwass 95] Stein, E. W.; Zwass, V.: “Actualizing Organizational Memory with Information Technology”; Information Systems Research 6, 2. (1995). 85-117. [Schmidt 99] Schmidt, K.: “Of maps and scripts - the status of formal constructs in cooperative work”; Information and software technology 41 (1999). 319-329.