Supporting Cooperative Learning of Process Knowledge on the World Wide Web

Weigang Wang, Jörg M. Haake, Jessica Rubart, Daniel TietzeIPSI – Information Publish and Systems Institute

GMD – Germany Research Center for Information TechnologyDolivostr. 15, 64293 Darmstadt

{wwang, haake, rubart, tietze}@darmstadt.gmd.de

AbstractThe WWW makes learning materials widely accessibleand provides an environment where people can learnacross time and space. However, the simple read-onlyinformation structure on the Web provides little or verylimited guidance to learners, especially when they wantto learn how to do something through interaction withcomputers and other people. We suggest overcoming theproblem by introducing a graphical hypermedia-basedprocess representation and a cooperative processenactment support. The hypermedia-based processstructure is accessible on the WWW. It contains a richset of associated materials with which people canseamlessly interact while they are systematicallycarrying out the process they are learning under theguidance of computers or tutors. A hypermedia basedprocess support system and a use case are presented.The use case shows that such a system can not onlyprovide traditional learning support, but also offersnovel cooperative hypermedia based support for thelearning of process knowledge.

1. Introduction

In today’s global economy, organizations arebecoming increasingly distributed. Furthermore, newforms of virtual organizations are emerging. As aconsequence, teams carrying out business processes in acollaborative way are increasingly distributed andconstantly changing. In addition, the business processesto be executed are changing in our global economy.Together, these trends require those teams

(1) learn what business processes they shall executecollaboratively,

(2) learn how they can execute these processeseffectively in a distributed team,

(3) carry out the work collaboratively following theintended process, and

(4) adapt the process to the needs of the team and theactual situation. Thus, a platform is needed whichsupports teams in the description and definition ofprocesses, the learning of these processes, and theadaptation and execution of these processes.

The Internet and the WWW make informationwidely accessible and provide an environment withinwhich people can learn and perform work whenever theywant and wherever they are. However, the simple read-only information structure of the Web creates severalproblems: Firstly, it provides learners with little or verylimited guidance, especially when they want to learn howto do something through interaction with computers andother people. The simple HTML-based informationstructure can not represent a process structuregraphically. Therefore, it is difficult for learners to getan overview of the process they are learning. Secondly,the read-only restriction prevents team members fromchanging a process description, either individually orcooperatively, based on the changing situation and whatthey have previously learnt. This restriction alsoprevents learners from getting interactive support fromcomputers and other learners or tutors.

In this paper, we describe how to use and augmentthe WWW to overcome the above problems. Ourapproach to supporting teams in the description anddefinition of processes, the learning of these processes,and the adaptation and execution of these processes canbe outlined as follows:

• Use advanced hypermedia structures, i.e., extendedlinks and composite nodes, to enrich the informationstructure of the current Web, as is required forprocess modeling;

• Integrate process support into the advancedhypermedia structure that is superimposed on thecurrent Web, as is required for supportingcollaborative process execution;

• Use the authoring and cooperation capabilities of acooperative hypermedia system to support cooperativemodeling, changing, and execution of both learningprocesses as well as the business processes to belearnt;

• Make the cooperative hypermedia-based processsupport accessible on the Web; and

• Provide traditional plus cooperative hypermedia-based process support for the learning of processknowledge.

The remainder of the paper is organized as follows:In section 2, our enabling technology, the cooperativehypermedia-based approach to process supports, ispresented. Section 2.1 describes the principle of ourflexible cooperative hypermedia based process support.Section 2.2 presents a cooperative process support system(XCHIPS) implemented with the approach. After thesetechnical descriptions, section 3 uses a use case scenarioto describe how the support for the learning of processknowledge is provided by XCHIPS. In section 3.1, a usecase scenario is described. In section 3.2, the learningprocess for process knowledge and our support for thelearning process are described. Section 3.3 presents oursupport for traditional learning skills (or teachingresources) on process knowledge learning. This sectionalso reveals how the cooperative learning support canenrich the traditional means for the learning of processknowledge. Section 4 discusses related work. The paperconcludes with a summary and plans for future work.For a quick grasp of the cooperative learning aspects ofthe paper, readers may also go through section 3 beforethe more technical oriented section 2.

2. Cooperative hypermedia approach toprocess support

In this section, the concepts of cooperativehypermedia and the principle of cooperative hypermediabased process support are presented. Then, a prototype

system implemented with the cooperative hypermediabased approach is described.

2.1. Flexible, cooperative, hypermedia-basedprocess support

A work procedure in a workflow system is defined bya workflow model composed of a set of discrete worksteps with explicit specifications of how a unit of workflows among the different steps [2]. In general, aworkflow coordination model can be defined as a directedgraph, (N, L), with a node set N representing individualsteps in the procedure and an edge set L representing thecoordination structure among the tasks [8].

The hypertext concept distinguishes informationcomponents (nodes) which are connected by relationships(links) [13]. Using links, linear as well as nonlinearnetwork structures can be formed. In addition to thebasic notion of nodes and links, one can introduce typesof nodes and types of links. These types can be used tocapture application or domain semantics, e.g., bydetermining allowed types of nodes as link end points ofspecific types of links. In addition to simple nodes, manyhypertext systems introduced composite nodes(composites) that can contain other nodes and links.Thus, they can be used to form aggregated subnets withinthe hyperdocument, which lead to the possibility oflayered graphs or networks. Hypermedia extends thehypertext concept by allowing any kind of multimediainformation to be the content of nodes. Collaborativehypermedia now adds to the hypermedia concept thepossibility of sharing a hypermedia workspace amongmany people.

Both the logical structures of a workflow process anda hypermedia network can be seen as a directed graph.This makes the correspondence between many workflowconcepts and hypermedia concepts straightforward. Forinstance, tasks correspond to nodes. Control and dataflow connectors between tasks correspond to links.Process execution corresponds to guided tours through ahypertext graph. The key difference between these worldsis that process structures have their own computationalsemantics, such as task state transitions and flowdependencies, while hypermedia structures usually haveno such semantics.

Our approach to modeling processes is to incorporatetask-related attributes, such as state and time, intohypertext nodes, and incorporate control flow and dataflow semantics into hypertext links [5, 10]. Thus, a

Figure 1. Process structure for organizing a conference in an XCHIPS browser

process is represented as a set of hypermedia task nodesconnected by process links. Ordinary hypermedia links,nodes, and media objects are application-specific. Theyare usually not directly accessible to a workflow system.

However, in this model they are represented in a unifiedhypertext model and therefore accessible by our system.Such a hypermedia-based process support system canprovide support for a wide range of processes from veryinformal ones that guide people to perform their tasks tovery formal workflow processes with automatic executionsupport. The composite structure of nested nodes ofdifferent types can provide different coordination supportfor differently structured subtasks. Such a hypermedia-based process description can contain a rich set ofassociated materials with which people can seamlesslyinteract while they are carrying out the primary task.More technical details on the cooperative hypermediabased process support can be found in [5, 10, and 11].

2.2. The XCHIPS system

Based on the hypermedia-based process support

approach, we have developed XCHIPS. XCHIPS standsfor eXtensible Cooperative Hypermedia Integrated withProcess Support. The XCHIPS system contains a set oftools (components) for people to create and usehypermedia objects cooperatively. It is implementedusing Java and a Java-based Dynamic CollaborationEnvironment (DyCE) [12]. DyCE is a framework forcreating mobile groupware components that are loadedon demand. It provides dynamically replicated-shareddata as well as transactional support for access to andmodification of this shared data. Additionally, DyCEprovides dynamic extension of the working environmentin the sense of adding groupware components at runtime.

2.2.1. Components and shared objects. The XCHIPScollaborative components developed on DyCE can beregistered to a server and transferred across the Internet.These components may work as Java applications or Java

applets that are executable in standard WWW browsers.In the XCHIPS system, both the hypermedia-basedprocess structure and its associated information arerepresented in object-oriented shared data models of thecollaborative components. These data models are storedon one or more DyCE servers and replicated to clientapplications when needed. They can also be importedfrom and exported to documents in XML.

2.2.2. Process structure representation. In XCHIPS,the process structure is represented by composite nodesand extended XML links. Composite nodes aregraphically presented as labeled icons and the extendedlinks are graphically presented as labeled arrow linesbetween the node icons. The common WWW embeddedlinks (i.e. the HTML links or simple XML links) can beused to point to the content information of a task orassociated reference materials. They are presented as linkmarkers with a special icon image. Composite nodesmay contain other nodes, links, and information objects.An XCHIPS browser has been developed to present andedit the graphical process structure, its associated contentobjects and HTML links. The state information of tasknodes is color-coded in their iconified views (white forinactive, yellow for enabled, green for active and brownfor completed). Tasks are performed by menu operationsactivated on the task nodes. For the common WWWembedded links, when activated, their content pages willbe presented in standard Web browsers (i.e., Netscapeand IE).

2.2.3. The XCHIPS browser. There are two objectpalettes on the XCHIPS browser: a tool palette on theleft-hand side and an object palette on right-hand side ofthe browser (see Figure 1). The tool palette containsgroupware tools and information organization tools. Thegroupware tools include a User Guide Browser, anExample Browser, a Paste tool, a Search tool, a Chattool, and a Whiteboard tool. The informationorganization tools include hypermedia objects forcreating Notepads, Embedded Links and Folders. Theobject palette on the right-hand side contains hypermediaobjects used to represent workflow processes. Theseobjects are “Start”, “Finish”, “Milestone”, “Task”,“Actor”, and “Meeting” nodes, and “precede”,“coordinate”, “change”, “supervise”, “assign to”, and“participate” links.

2.2.4. Sessions. The XCHIPS browser is itself agroupware tool. In addition, all tools activated from itstool palette or its content pane are also groupware tools.These groupware tools are used in sessions. A session is

formed of a collection of people, tools, and contentobjects. The groupware tools can be activated in thesame session (thus becoming accessible to all userswithin this session) or in a new session (providing asimple transition between coupled cooperative andindividual work). Users can invite other users to jointheir working session by activating the Invite user buttonat the upper right corner of the session browser. Userscan also use the query tool to search for all the activesessions and request to join selected sessions.

2.2.5. Working modes. The XCHIPS browser can belaunched in private, loosely coupled or tightly coupledworking modes, and users may switch among thesemodes within a collaborative working process. In theseworking modes, some aspects of shared objects, such asthe current node and the position of the scrollbar of thecontent pane, are coupled in different ways. For instance,in a loosely coupled cooperation mode, the navigationaction of a user does not affect the browsers of othercooperating users. However, if users work in a tightlycoupled cooperation mode, all users follow thenavigational actions of other group members in the samesession.

3. Supporting cooperative learning of processknowledge in XCHIPS

In this section, we first introduce a use case scenario.Then we use this scenario throughout all the subsectionsto show

• how the process for learning how to do something(i.e., the learning process) and the process on doingsomething (i.e., the business process to be learnt) issupported, and

• how both traditional support and newer flexible,cooperative hypermedia based process support for thelearning of the process knowledge is provided by theXCHIPS system.

3.1. Use case scenario

Peter is a distinguished scientist. He is invited toserve as the next conference chair of a conference series.He has had no such experience before. Therefore, hewants to learn how to organize a conference, asconference chair. Peter's friend John, who lives inanother country, has served as conference chair for thisconference series. Therefore, Peter contacts John for

advice. Coincidentally, John has documented theprocess for organizing the conference in the XCHIPSsystem, so he simply sends a URL to Peter. The scenariois continued in the following sections in italic font.

3.2. Support for the learning process and thebusiness process to be learnt

There are a number of different types of learning,such as learning to remember something, learning tounderstand something, learning a skill and learning howto do something. The learning process for learning howto do something is a simple process involving three steps[1]:

(1) determine the purpose,

(2) identify the procedures involved, and

(3) practice the task.

The following sub sections provide more details on thesesteps and describe how they can be supported by theXCHIPS system.

3.2.1. Determining the purpose. A clear understandingof the purpose of what we are trying to learn maymotivate us to learn. The XCHIPS system supports theexplicit formulation of goals and allows learners toaccess this information for facilitating a betterunderstanding. A hypermedia-based process structuremay contain the process definition and other relevantinformation resources; the goals can be expressed astextual annotations to the process structure. Afterreceiving the URL from John, Peter types it into the URLfield of the Web browser on his computer desktop. AnXCHIPS browser appears and its content page containsa graphical representation of the process structure fororganizing a conference (see Figure 1). On the top ofthe content pane, there is a textual annotation that says,"This is a process for organizing a conference. Theconference committee members are expected tounderstand and follow the process under the direction ofa conference chair ". Peter is very happy, as this is justwhat he wants.

3.2.2. Identifying the procedures. There are many waysto identify the procedures of how to do something. Oneway is to extract them from written instructions. Anotherway is to examine documented examples. The third wayis to watch someone else doing it or interview someonewho knows how to do it. The identified procedures onhow to do something can be externalized in a graphical

process representation, maybe first on paper and then incomputer using the hypermedia-based processrepresentation. In the XCHIPS system, the identifiedprocedures on how to do something can be externalizedin a graphical hypermedia-based process structure, inwhich tasks are represented as task nodes and precedencerelationships between tasks are represented as processlinks. Other associated information, such as team androle based organization structure and referentialmaterials can be represented using ordinary hypermedianodes, links, and other content objects. When John was toserve as a conference chair for the first time, he had thesame learning need. He spent a lot of time reading textson how to organize conferences, examining variousconference Web pages, and talking to experiencedpeople. When he finally figured out how to organize aconference, he documented the process in the XCHIPSsystem and used the process structure as a systematicguidance for learning and execution support.

3.2.3. Practicing. We will learn how to do somethingsuccessfully only by practicing it. We can practice thetasks in a process following the identified procedures. Asdescribed in the next section, Peter has studied theprocess structure and practiced it virtually using bothtraditional learning support and cooperative hypermediabased support provided by XCHIPS.

3.3. Traditional plus new kind of support

The headings of the following subsections are namedafter the widely used learning skills or teaching resourcesfor learning how to do something [1]. In eachsubsection, we first briefly introduce one or two suchtraditional learning skills. Then, we describe how theyare supported in the XCHIPS system. Finally, wecontinue the scenario to further highlight such support,especially the newer ingredients for cooperative learning.

3.3.1. Written instruction and examples. One way oflearning how to do something is by following writteninstructions. These instructions summarize the mainpoints to be learnt and provide us with backgroundreference material. In the XCHIPS system, suchinstructions are provided as annotations and referentiallinks at the side of the graphical process structure (seethe “xref” on the upper right corner of the content panein Figure 1. This “xref” contains an URL, which pointsa Web page describing the established regulations of theconference series). Additional information about how touse the process structure is provided in hyperlinked on-

line documentation. This can also be referenced fromwithin the hypermedia-based process structure.

Another way of learning how to do something is toexamine and follow good examples. The hypermedia-based process structure can serve as an example.Learners can navigate through the process structure toget an overall picture of how to do something. They canalso simulate the process execution to see how it proceedswhen the actual execution starts.

Peter glances at the process structure and quicklygets an overview over how to organize a conference. Hethen has a look at the on-line help information of thesystem, and starts to navigate through the processstructure to examine the content of the tasks and the subprocess structures, such as the sub process for “developconference program”. He also reads the instructionsand reference materials attached to the process. Afterthe first go, he returns to the root task node of theprocess structure, and triggers the process executionanimation. The root task node first turns to “yellow”(for ready), and then to “green” (for active), then itscontaining process structure appears. The start task inthe process structure turns to “yellow” and then to“green”, and finally to “brown” (for completion). Thenthe following tasks of the start task turn to “yellow”,“green”, and “brown”. The animation continues untilall sub tasks of the tasks are completed. Peter gets afeeling that a conference has been going smoothlytowards its successful completion.

3.3.2. Demonstration and practice in a guided tour.One of the best ways of learning how to do something isto watch someone else doing it. A demonstration by atutor is useful in that learners can see a model of correctbehavior provided by the tutor, which they can thenimitate. With demonstrations, the tutor is also availableto answer any questions that the learners might have.

Peter has some questions to ask. He invites Johninto his working session in the XCHIPS system. Johnvolunteers to give a guided tour for Peter. As they workin a tightly coupled mode, the navigation performed byJohn also affects Peter’s browse. They use the Chat toolprovided by the system in the same session as theirXCHIPS browser, so that they can see both the processstructure and the Chat window at the same time. Johnexplains what he has done for each task during the tourand answers many questions asked by Peter.

3.3.3. Role-play and record-keeping. One of the mostefficient ways of learning how to do something is to

practice it. Role-play exercise is a situation in which oneacts out, or performs certain skills and behaviors in asimulated situation. Each learner is given a certain roleto perform and he or she is free to develop his or her roleat wish, or according to guidelines that are provided.Role-play exercise is a valuable learning strategy in thatit allows learners to practice new skills in a controlled,safe setting, before they are used in real environment.

Record-keeping of the role-play exercises shouldgreatly improve our learning of skills because the recordsprovide feedback of how we have performed and howothers see us. Because all the information attached to aprocess is kept in the shared hypermedia space,newcomers can navigate the hyperspace to examine theprocess structures and the contents used or created in theprocess. Also, the process structure can be reused forperforming similar tasks. In the XCHIPS system, theactors of a process take roles named with specifiedresponsibilities. These roles are visualized and assignedto specific tasks at the proper levels where they areinvolved in the nested process structure.

Peter makes a clone of the process structure, doessome more practice, and makes some tailoring to hisneeds. He introduces the process structure and theXCHIPS system to his will-be program chair Tom. Theywork together on-line in the system, taking theconference chair and program chair roles. Peter workson many tasks, but only at a high level. Tom takesresponsibility for developing a conference program.Peter also tries the Treasurer role. They edit the“execute conference program” task jointly (Figure 1shows an awareness indicator that Peter and Tom areworking on this task). They split the task into two tasks:one for “Refine conference program” and another for“Execute conference program”. They agree that it is agood idea to have the system used by all their conferenceand program committee members. Peter feels that he iswell prepared to organize and chair the comingconference.

4. Related work

Cooperative learning can improve several cognitiveand social aspects of the learning process [6]. In adistributed cooperative learning setting, learners needsome means to communicate and coordinate theirlearning activities. In the CSILE system, a certaincooperative learning method is modeled, and learners areguided through the process by specific rules [9]. Work in

[6] and [7] provide some scripted learning support aslearning protocols. The scripts are implementedprocedures that can be executed automatically or ondemand. The scripted learning support is some form offormal process support. Comparing to scriptedprocedures, the hypermedia based process supportprovided by XCHIPS is more flexible. The XCHIPSsystem can support a wide range of coordination means;from very informal shared artifact-based means to veryformal processes that are suitable for differentcooperative learning settings.

Process support is primarily provided by workflowsystems. However, most workflow systems are nothypermedia-based and they often only provide users witha work list. It is not always easy to figure out theoverview of a process structure and the current states ofits containing tasks. Some workflow systems alsoprovide a graphical process editor. However, the editor ismainly for process designers, rather than end users of theprocesses. The same is true for most scripting and agentbased workflow systems. Our hypermedia based approachleads to a system dealing with processes and informationstructures (as the subject of processes) as two views on aunified data model. This permits smooth transitionsbetween these two views, and supports cooperative workon defining and manipulating process and informationstructures. Unlike those workflow systems, which stillmaintain the strict separation between process anddocument data, the XCHIPS system can provide muchmore flexibility.

The user interface provided by most Web-basedworkflow systems uses the simple HTML-based pages, inwhich only few components, such as Forms, providesome interaction support. Compared with HTML, Java isa full-fledged programming language, its developmentkit provides a rich set of interface components forcreating application user interface. Therefore, we haveimplemented our user interface in Java. Java and XMLhave provided our system with a better accessibility andvisualization support. In our approach, we still useHTML links to refer to our exported XML documentsand other Web pages, and use Web browsers to invokethe XCHIPS system so that we can take advantage of thewide accessibility of the WWW.

A few previous hypertext systems have providedsome form of process support. For example, Anecdote [4]provides support for a storyboarding process and thework in [3] provides an automaton-based hypertext forhypertext browsing and software engineering processes.

However, their focus is not on cooperative modeling andexecution of user-defined processes. The CHIPS system,a predecessor of XCHIPS, provides hypermedia-basedsupport for cooperative process definition and execution[5, 10, and 11]. However, in the past it had not beenapplied to the learning of process knowledge and asmany systems in this category, it is not accessible on theweb. Other XCHIPS extensions to CHIPS include theXML importing/exporting facilities and the dynamicsoftware component extensibility, which allows the newlydeveloped groupware components to be added to thesystem and made available to all its distributed users atrun time. The advanced hypertext links and compositionstructure in XCHIPS enrich and complement the currentWeb information organization structure. They provide aprocess enabled, graphically presented semantic networkthat is superimposed on web pages. Moreover, XCHIPSenables synchronous as well as asynchronouscollaborative editing of shared processes and theirassociated information.

5. Conclusions

In this paper, we described XCHIPS, which providescooperative hypermedia-based process definition andenactment support for modeling, disseminating, andevolving process knowledge, and how it can be used forsupporting cooperative learning of process knowledge onthe WWW. XCHIPS presents the hypermedia-basedprocess structure in a graphical form over the WWW. Inaddition, the hypermedia-based process structurecontains a rich set of associated materials with whichpeople can interact while they are carrying out thebusiness process they are learning. The processes can beexecuted automatically, under the control of computer, ormanually, under the control of the process performers.The explicit process representation and enactmentsupport can be used to guide people to practice and carryout the process they are learning. A prototype systembased on the approach has been implemented. The usecase presented in this paper shows that the system cannot only provide traditional learning support, but alsooffers novel cooperative hypermedia based support for thelearning of process knowledge.

This paper focuses on the support for the learning ofprocess knowledge in general, and the type of learning onhow to doing something in particular. Our experiencesuggests that such cooperative hypermedia-based processsupport is also helpful for other types of learning, such aslearning to memorize something and learning to

understand something. The XCHIPS system is one of thetools we will bring into and extend in the EXTERNALproject (IST1999-10091) funded by the CEC.EXTERNAL focuses on the engineering and operation ofnetworked organizations, and the management of processknowledge. Next, we will broaden and apply ourapproach to process knowledge management, integrate itwith other EXTERNAL tools, apply it to other types oflearning, and evaluate this approach in three real-worlduse cases.

Acknowledgments

We thank Martin Wessner for his helpful commentson a previous version of this document.

References

1. D. Hind, Transferable personal skills, BusinessEducation Publication Limited, 1989

2. C. Ellis, and G. Nutt, Modeling and enactment ofworkflow systems. Advances in Petri Nets 93, (June1993), 1--16.

3. R. Furuta, and D. Stotts, Interpreted collaborationprotocols and their use in groupware prototyping. InProc. of ACM CSCW'94 (Oct.,1994), pp.~121--313.

4. K. Harada, E. Tanaka, R. Ogawa, and Y. Hara,Anecdote: A multimedia storyboarding system withseamless authoring support. In Proc. of ACMMultimedia'96 Nov. 1996), pp.~341--351.

5. J. Haake and W. Wang, Flexible Support forBusiness Processes: Extending CooperativeHypermedia with Process Support, in Informationand Software Technology (41) 6, 1999, pp. 355-366

6. B.M. Mancini, R. H. Hall, M.A. Hall, and B.Stewart (1998). The individual in the dyad: aqualitative analysis of scripted cooperative learning.Journal of Classroom Interaction, 33 (1), 14-22.

7. M. Wessner, H. Pfister, Y. Miao, Using LearningProtocols to Structure Computer-SupportedCooperative Learning. In: Proceedings of the ED-MEDIA, Seattle, Washington, June 19-24, 1999, pp.471-476.

8. G. Nutt, The evolution towards flexible workflowsystems. Distributed Systems Engineering Journal, 3,4 (Dec. 1996), 76--294.

9. M. Scardamalia and C. Bereiter, Computer Supportfor Knowledge-Building Communities. InKoschmann, T. (Ed.): CSCL: Theory and practice ofan emerging paradigm. Mahwah, NJ: Erlbaum, 249-268, 1996

10. W. Wang and J.M. Haake, Flexible Coordinationwith Cooperative Hypermedia, Proceedings of ACMHypertext'98, pp. 245-255, June, 1998

11. W. Wang "Team-and-Role-based OrganizationalContext and Access Control for CooperativeHypermedia Environments", in Proceedings of ACMHypertext'99, pp. 37-46, Feb 23-25, 1999.

12. D.A. Tietze, Ralf Steinmetz, Ein Framework zurEntwicklung komponentenbasierter Groupware. Toappear in: Proceedings der Fachtagung D-CSCW2000, München, September, 2000

13. J. Nielsen, Hypertext and Hypermedia. AcademicPress, Inc., San Diego, 1990