Foundation of a Framework to Support Knowledge Management inthe Field of Context-Aware and Pervasive Computing.

Philipp Amann*/** and Gerald Quirchmayr**/****CTBTO PrepCom, Vienna International Centre, P.O. Box 1250, A-1400 Vienna, Austria

** University of Vienna, Institute for Computer Science and Business Informatics, Liebiggasse 4/6,A-1010 Vienna, Austria

***University of South Australia, School of Computer and Information Science, Mawson Lakes Campus,Mawson Lakes, SA 5095, Australia

philipp.amann@ctbto.org gerald.quirchmayr@univie.ac.at/gerald.quirchmayr@unisa.edu.au

Copyright (c)2003, Australian Computer Society, Inc. This paperappeared at Workshop on Wearable, Invisible, Context-Aware,Ambient, Pervasive and Ubiquitous Computing, Adelaide, Australia.Conferences in Research and Practice in Information Technology, Vol.21. Editors, Chris Johnson, Paul Montague and Chris Steketee.Reproduction for academic, not-for profit purposes permitted providedthis text is included.

The ideas discussed in this paper express the authors’ opinion. TheCTBTO PrepCom affiliation of Philipp Amann is mentioned todocument the excellent support and continuing input from thisorganization for his research.

AbstractIn this paper we propose a framework to combineKnowledge Management and context-aware andpervasive computing, emphasizing on synchronizationand adaptation issues of workflow processes in mobilesettings. The key aspect of the proposed framework is toenable adaptive, two-way interaction between context-aware systems and users in mobile settings. In contrast toexisting concepts, we aim at capturing active feedbackfrom users, which should contribute to the OrganizationalMemory, after being reviewed, evaluated and classified.Thus, users would not only act as consumers but also assuppliers of relevant information and knowledge to thesystem and other users. In addition, the concept includesexisting approaches to adapt to, for instance, differentQuality of Service levels in order to provide a maximumlevel of local autonomy. We suggest using the adaptationconcept to also support adjustments to cross-culturaldifferences in perceiving and communicating informationand knowledge. Our work is motivated by the need for adistributed, context-aware and pervasive computingframework to support maintenance and administrationtasks related to the International Monitoring System(IMS) of the CTBTO PrepCom, an internationalorganization located in Vienna, Austria.

Keywords: Knowledge management, WfMS, context-awareness, pervasive computing, local autonomy,synchronization, adaptability

1 Introduction

In this paper we combine and expand existing theoreticaland practical work in the fields of KnowledgeManagement (KM) and context-aware and pervasivecomputing, emphasizing on synchronization andadaptation aspects of workflow processes in mobilesettings. We believe that there is a need to combine theseefforts into a broader framework, which addressestechnical, but also organizational, cultural and socialaspects such as organizational and personal knowledge,

cross-cultural communication structures, skills,motivation, and experience. In particular, we areinterested in the technical, organizational and managerialmeans to represent and disseminate (local) knowledgeand to build up and provide access to the OrganizationalMemory (see Maurer 2001 as well as Kuppinger 2000 andBrown 1998 for similar concepts like Organizational IQand Organizational Knowledge) in a mobile environment,using context-specific knowledge and information. Thisincludes the definition of the technical foundation (e.g.definition of domain specific semantics, mobile workflowprocesses, etc.) but also tackling the challenges that aregenerally related to developing and introducing a systemthat requires users to 'hardcode' their knowledge andadapt to new processes and structures (Grudin 1994 andFahey 1998). We seek to utilize these insights byaddressing and including human factors in our conceptand pursuing an approach that is not primarily IT-driven.In particular, research on Communities of Practice (CoP),an interesting notion of the collective nature ofknowledge in organizations, constitutes an importantaspect of the foundation of the proposed framework(Brown 1998 and Hildreth 2000). In this paper, however,we will mainly focus on synchronization and adaptabilityaspects of mobile workflow processes. Similar to Thomas(2002), the framework aims at using the coordinationpower of Workflow Management Systems (WfMS) inmobile settings.

A key aspect of the proposed framework is to enableadaptive, two-way interaction between context-awaresystems and users in mobile settings - focusing onsupport of remote personnel. In contrast to existingconcepts already using active feedback from users toadapt accordingly Piekarski (1999), it is our intention toexploit not only explicit context information such asidentity, location, time, and preferences but also tacitinformation and local knowledge of users. This feedback,in turn, should contribute to the Organizational Memory,after being reviewed, evaluated and classified. Thus,users would not only act as consumers but also as

suppliers of relevant information and knowledge to thesystem and other users. In addition, the concept includesexisting approaches to adapt to, for instance, differentQuality of Service (QoS) levels in order to provide amaximum level of local autonomy (Alonso 1995 andAlonso 1997). We propose using the extended adaptationconcept to also support adjustments to cross-culturaldifferences in perceiving and communicating informationand knowledge.

The remainder of the paper is structured as follows.Subsection 1.1 provides background information on thecase study. In section 2, an overview of related work andrelevant literature is given. Section 3 is concerned withthe identification of open issues and the detaileddescription of the problem. Section 4 outlines theproposed solution and, in section 5, a summary of theexpected outcome and future work is given.

1.1 Background Information

Apart from academic aspects, our work is motivated bythe need for a distributed, context-aware and pervasivecomputing framework to support maintenance andadministration tasks related to the InternationalMonitoring System (IMS) of the CTBTO PrepCom1, aninternational organization located in Vienna, Austria.The organization is based on the Comprehensive NuclearTest Ban Treaty (CTBT), which prohibits all nuclear testexplosions in all environments.

The CTBT foresees the implementation of the IMS,which is a world-wide network of 321 monitoring stationsand 16 radionuclide laboratories to monitor compliancewith the provisions of the Treaty (see figure 1). The IMSis based on seismic, radionuclide, hydroacoustic andinfrasound monitoring technologies. These technologiesare employed to monitor underground, underwater andatmosphere environments, respectively. Radionuclidestations, for instance, can detect radioactive debris fromatmospheric explosions or vented by underground orunderwater nuclear explosions.

Figure 1: International Monitoring System

The collected data is sent to the Provisional TechnicalSecretariat (PTS) in Vienna via satellite links where it is

1 http://www.ctbto.org, accessed November 2002

being analyzed and combined to reports for memberstates. Global coverage is being ensured through theGlobal Communications Infrastructure (GCI), whichreceives and distributes data through a network ofgeostationary satellites and hubs (see figure 2).

The basic requirements of IMS stations in terms ofinstallation, data quality and transmission, up-time,security, communication and reporting procedures, andmaintenance procedures are defined in CTBTO internaloperational manuals (CTBTO PrepCom 2000:1, CTBTOPrepCom 2000:2, CTBTO PrepCom 2000:3 and CTBTOPrepCom 2000:4).

Figure 2: Global Communications Infrastructure

While the manuals define the daily work procedures andmaintenance tasks in a general way, there is a lot of tacitknowledge and information that is applied by stationoperators in their day-to-day work (e.g. calibration ofsensors). This knowledge is often created throughlearning by doing, especially if new equipment is beingdeployed. To capture and disseminate this localknowledge can be considered very important because ofthe special-purpose character of some of the equipmentfor IMS stations.

In addition to this routine work, station operators need toperform scheduled maintenance tasks and send reports ona monthly basis. This monthly report is considered to bea key element in the maintenance of IMS stations as it“[..] provides a consolidated, executive-level overview ofall activities and issues that may have impacted stationoperational capabilities over the previous month. Thisoverview is essential for effective management of theTechnical Secretariat cost management system andoverall system effectiveness statistics. To ensure theseprograms are successful, this report must reflect allstation issues as accurately as possible.[..]" (CTBTOPrepCom 2000:1, p. 30). As we will mention later, oneof the problems with these reports is their official statusand the political dimension related to it. Similar problemsare described, for instance, in the area of KM forcustomer support: “support-oriented information can bepolitically sensitive, because it often involves product andservice failing” (Davenport 1998, p. 206). Hence, certainrelevant information might not be found in these reportsdue to political reasons. Furthermore, most of thisinformation is currently not ‘actively’ used, owing to itslimited online availability.

2 Relevant Literature and Related Work

This section provides a survey of relevant literature andwork in the fields that are of importance to our approach.Due to the fact that Knowledge Management and mobileworkflow processes play a crucial role in the proposedframework, the following subsections are geared to thesetwo areas.

2.1 The Field of Workflow ManagementSystems

In order to support all facets of KM in mobile settings,the combination of GroupWare and WorkflowManagement Systems (WfMS), despite their differentcharacteristics, is suggested by several authors. The basicidea is to exploit the coordination aspects of WfMS andprovide tools to collaborate and communicate (Domingos1999 and Thomas 2002).

Although the Workflow Management Coalition’sWorkflow Reference Model2 foresees some basicconcepts to support mobile and distributed workflowmanagement (WfM) (e.g. electronic mail model, remoteprocedure call and message passing model), there iscurrently no adequate support for WfM in mobilesettings, which has led to several different approaches totackle the problems linked to mobile settings.

Research in this area is concerned with distributedworkflow frameworks, disconnected clients, survivabilityand synchronization of (mobile) workflow processes, tomention some (Alonso 1995, Cardoso 2001, Domingos1999, Heinl 1999, Jing 1999, Purvis 2000 and Reichert1999). The goal is to develop concepts that can deal withthe special characteristics of mobile settings withoutjeopardizing the overall workflow process, keepingpersistency but at the same time keeping a maximumlevel of local autonomy. Users of such solutions are, forinstance, healthcare, utilities, surveying, transportation,telecommunications, and emergency responseorganizations.

One of the identified restrictions of current WfMSpertains to the lack or insufficient support of incompletetask definitions. To lift this restriction is particularlyimportant, as task definitions in mobile settings are likelyto be completed/complemented by information or localknowledge only obtainable at physical locations.

Existing WfMS can be classified according to their WFdefinition scheme. For instance, one can distinguishbetween process and activity structures ranging from veryrigid and inflexible (pre-defined process and activitieswhich are typically found in administration) throughmore flexible structures (pre-defined activities and ad-hocprocesses) to very flexible, human-centered structures,which support ad-hoc processes and ad-hoc activities.Because of the time and expertise needed to defineworkflows, ad-hoc modeling is seen by some authors as acontradiction in itself (Alonso 1997 and Heinl 1999]).

2 http://www.wfmc.org, accessed November 2002

2.2 Relevant Frameworks, Approaches andTechnical Infrastructures

In the area of WfMS, several frameworks and approacheshave been described to support distributed and/or mobileenvironments. In Heinl (1999), for instance, acomprehensive approach to flexibility in WfMS isdescribed, focusing on adaptive workflows i.e. workflowsthat can adapt to exceptions and changes of businessprocesses. One of the central concepts of the approach isdescriptive modeling, which is based on late modeling.Using descriptive modeling means to omit certaindefinitions of the workflow type at modeling time, whichprovides the end user with more flexibility. Of course thisalso requires end users to have the necessary knowledge,skills and expertise to chose the proper action. Theimplementation of descriptive modeling, which has beenincorporated in the Mobile WfMS by the authors, is basedon a perspective oriented workflow model where eachperspective represents an independent modeling aspect.

Reichert (1999) focuses on the high-flexibility,maintainability, and scalability of enterprise-wide andcross-enterprise workflow management systems. Theauthors discuss adaptive WfMS, which support run-timechanges of WF instances i.e. adaptations at the executionlevel. In contrast to Heinl (1999), this approach focuseson the support of ad-hoc deviations from pre-definedworkflow schemas at the instance level, which may benecessary to handle exceptions or support late modeling.Some of these features have been developed as part of theADEPT project but due to the non-trivial character ofdynamic WF changes, a number of open issues likesemantic rollback, security, and handling of privilegesand roles still remain.

In Domingos (1999), a workflow architecture to managecollaborative work is discussed. The definition andimplementation is based on a case study, which tookplace in the electrical industry. The authors point out that“in mobile collaborative scenarios, pre-defined processesare too restrictive and there is high probability ofconflicts/divergences as well as communication failures,involuntary disconnection and more problems to dealwith high availability and reliability in resource sharing”(Domingos 1999, p. 4). The proposed workflowarchitecture is based on loosely defined workflowssupporting a hierarchy of decentralized organizationalworkflows with different levels of specialization andwork-control granularity. The basic idea is to not over-specify tasks. Some of the shortcomings are:

• Planner-Team approach requires considerablecoordination, which is put on the planner andrealized through election mechanisms;

• The actual integration of local updates to the globalstate of the WfMS is not described in detail;

• Lack of KM concepts.

Jing (1999) distinguishes between location-independentand location-dependent mobile workflow applications.Each type requires different mobile support (e.g. on-callfield service is spatial in nature and therefore location-dependent). Location-independent activities are

characterized by a high level of structure and pre-designbecause input and output is better predictable. Based ontwo different examples, the authors look into strategies tosupport workflow-specific and application-specificadaptations. The first type of adaptation performsoptimization by re-scheduling and sequencing multiplework activities dynamically whereas the latter type usesuser knowledge and application semantics like theconsideration of different QoS levels of the underlyingcommunication infrastructure for optimization purposes.

In Davies (1995), application-specific adaptation is alsoidentified as an effective way to support mobileapplications. At a more technical level, the authorsdescribe the implementation and use of a platformdesigned to support collaborative multimedia applicationsin a mobile environment as part of the Mobile OpenSystems Technologies (MOST) project. The purpose ofthe project is to study the requirements of field engineerswithin the U.K. power distribution industry. Throughexplicit binding of communication streams, the platformsupports testing of and adapting to different levels of QoSwithout having the client to take care of the actualtechnical implementation. The technical realization ofthe corresponding protocol – QEX – is described inDavies (1996). In addition, the system provides visual,context-specific feedback to the user on the state ofcommunications between all conference participants andcollaborative tools. Thus, the user can select theappropriate communication channel, based on a givenQoS level.

An interesting approach to survivability of WfMS, usingKM concepts, is described in Cardoso (2001). The idea isto adapt workflow processes in response to exceptions,which are handled by exception handlers using existingknowledge. The authors define survivability as “thecapability of a workflow management systems tomaintain a pre-established acceptable running mode andbehavior after the occurrence of unexpected errors,accidents, failures or attacks, in a timely manner and toallow the adaptation and evolution of the supportedprocesses in response to its surrounding environment”(Cardoso 2001, p.2). It is interesting to note that thisdefinition is geared to exceptional, catastrophic eventsand does not include the ‘natural’ disconnection thatoccurs in mobile settings. If no exception handler can befound to handle an exception human input is required,which will be stored in the Case Based Reasoning (CBR)system for future reuse. Thus, the exception resolutionprocess is actually the population process of the CBRsystem. While using an intelligent CBR mechanism toreuse existing knowledge in handling exception is apromising approach, the proposed solution does notelaborate on the actual semantic impact of dynamicchanges to running WF instances. The concept reliesheavily on the knowledge and input by humans in the‘learning stage’ but is able to reuse this information infuture scenarios. It does not, however, consider mobileaspects.

In Fagrell (2000), an architecture to support mobilecollaboration is outlined. It facilitates filtering of tasksusing agents, including several, heterogeneous backend

information systems, flexible and adaptable views oninformation, and task-dependent access to experts. It isdesigned to support autonomous and creative activities inmobile settings, providing collaboration through interestprofiles, which are automatically created for a taskthrough extracting keywords from that task. If interestprofiles (i.e. tasks) overlap, the user will see the name ofthe corresponding user and her availability (considered asan ‘expert’ for the task at hand). The prototype describedsupports news journalists - as a consequence, experts areselected based on authorship but not informationprovided by the user. Furthermore, the system requires astrong attitude of sharing and collaboration and does notsupport structured tasks. There is also no explicitfeedback component to capture relevant information andknowledge in the field.

The model proposed in Thomas (2002) aims atintegrating existing backbone information systems likeWfMS in mobile computing, in particular in outdoorwearable augmented reality computing to supportcollaboration across a number of application domains.To this end, an Outdoor Wearable Augmented RealityCollaboration System (OWARCS) is introduced. Amedical emergency scenario using an advancedcontrol/analysis room concept demonstrates the practicalrelevance of the OWARCS. Similar to notes andannotations as used, for instance, in Reddy (2001), theOWARCS supports augmented reality informationstickers (e.g. audio messages) to be placed on an objectvia an eye cursor. The authors identify synchronizationand coordination of processes running in parallel as wellas handling of communication interruptions as majordifficulties. As a solution, the authors propose to linktraditional WfMS to a message based enterprise bus likeELVIN (Fitzpatrick 2000 and Sutton 2001, see alsosection 3). The paper further outlines the differentcomponents of an advanced control room environment,emphasizing the importance of adaptability,synchronization, flexibility (e.g. rapid configuration tosupport different display and computing devices), androbustness. One important difference to the approachdiscussed here is that the OWARCS focuses on reusingexisting knowledge whereas we also intend to explicitlysupport the capture and dissemination of new knowledge.

Bialek (2001) details a project that supports a specificWF process (ordering and measuring of blood samples) ina hospital, using hand-held devices. The main goal is todefine a mobile, reliable solution, which provides amaximum level of local autonomy i.e. the mobile clientshould work continuously even in case of networkdisruptions. The communication concept between clientand server is using a Message Oriented Middleware(MOM), which stores and forwards messages in a reliableand persistent fashion. The concept also investigatesmeans to recover from different types of software faults.To this end, it utilizes watchdog processes to monitor theapplication modules. The proposed concept focuses onjust a particular WF process and considers adaptation atthe communication layer only. However, the technicalconcept is a very useful starting point for the technicalpart of our approach.

2.3 Knowledge Management and MobilityAspects

Research on KM in mobile settings employs concepts likelocal knowledge (i.e. knowledge specific to individualusers, places, procedures and situations; created throughlocal experience and used locally) and CoPs, which holdcollectively produced knowledge, to describe the specialcharacteristics of mobile, distributed KM processes.

The important role of local knowledge in mobile settingsis discussed, for instance, in Fagrell (1999). The authorsargue that particularly remote mobility, which is definedas remote users interacting with each other usingtechnology, is closely related to local knowledge.

Along with KM studies in mobile settings, there is a basicrecognition of the importance of individual and groupdifferences, including of course cultural differences.Hayward (1997), for instance, presents an interestingstudy on culture - defined as the collective programmingof the mind which distinguishes one group from another -and aviation safety. The need of KM to adapt to localsettings is also identified in Teece (1998). In this context,the concept of Boundary Objects, which are defined asartefacts (e.g. shared documents) used by communities totransfer knowledge over boundaries between them, hasbeen introduced. Boundary objects are generic enough tobe transferable but robust to retain their structure.However, their transfer often needs to be supported bytacit knowledge and expertise as they can be interpreteddifferently, particularly in international environments(Hildreth 2000). In Reddy (2001), the role of medicalrecords as boundary objects in a Common InformationSpace (CIS) is discussed. At a more theoretical level,Dourish (2001:1) discusses boundary objects in the formof process descriptions in WfMS.

Closely related to the work linked to mobile settings, isthe work on software agents, which are basicallyautonomous entities that can communicate with otherpeople and agents to fulfil complex tasks, sometimes onbehalf of a human being. As such they can be used, forinstance, in mobile, heterogeneous environments to workon tasks while the user is disconnected from anyserver/service as well as virtual team members and tutors(Bradshaw 1997 and Cabri 1998). The use of softwareagent technology can add significantly to the usefulnessand practical relevance of our approach. Especiallyinformation retrieval in the field, where continuos andpermanent network connection is generally not available,can be supported by software agents. The actualincorporation of software agent technologies is envisagedat a later stage, after the implementation and evaluation ofthe current approach.

3 Problem Description

Current shortcomings and limitations in the area ofcontext-aware and pervasive computing can be attributedto technical aspects (Sohlenkamp 2000) as well asinadequate or lacking organizational, managerial andpsychological concepts and measures, as discussed in

(Dourish 2001:2, Grudin 1994 and Shipman 1994). Thefindings include discrepancies in the way benefits areperceived by management and individuals, problems withinadequate and cumbersome information and knowledgestructures, difficulties in eliciting and/or revealingworkflow processes and the importance of managerialand key user support. Additional problems concern theneed to learn and adapt to new formalisms and thedisruption of social processes. These shortcomings andproblems prevent or limit the success of such systems byfailing to support and motivate users to activelyparticipate and to share relevant information andknowledge. Last but not least, technical restrictions (e.g.limited CPU power and physical memory, short batterylife, small screen size, etc.) have been the cause ofdifficulties and limitations in practical settings (Brodie2001 and Sohlenkamp 2000).

Another problem with most interactive computationaltechnologies is their complexity, which makes themextremely obtrusive elements of our workingenvironment – sometimes even requiring the adaptationand redesign of practices, organizational processes andphysical settings to accommodate usage (Dourish2001:2). This has led to a number of unsuccessful anddifficult projects like the UK air traffic control center atSwanwick, which took six years longer to becomeoperational than originally planned, was greatly overbudget and, above all, still has some severe technical anduser interface problems (e.g. font size used to labelaircrafts on the monitors is too small)3.

On the other hand, successful interactive technologieslike ELVIN (Fitzpatrick 2000) have created considerableattention, particularly because of the unobtrusivecharacteristics of applications using the ELVINinfrastructure like the TickerTape interface. Theimportance of these characteristics is also recognized byELVIN’s authors who stress that without the Tickertapeinterface “[..] it is doubtful whether its [ELVIN] potentialfor support of awareness and interaction would have beenrealised” (Fitzpatrick 2000, p. 27).

The reason for selecting the IMS as a proof of concept forour framework is connected to current infrastructural andtechnical difficulties pertaining to different equipment,several different information sources as well as thelocalization of knowledge (‘ know how’ but also 'knowwho') and the graphical distribution of stations. Severaldifferent software packages are used, for instance, tomaintain and administer stations and gather feedback.Furthermore, the geographical distribution of the IMSnetwork and the resulting cultural diversity brings about anumber of issues, which manifest, for instance, indifferent ways of communicating and perceivingknowledge and information. Last but not least, there is anexpressed need of station operators for adequate supportin disseminating relevant information and knowledgeamong them.

3 See, for instance:http://news.bbc.co.uk/hi/english/uk/newsid_1936000/1936464.stm, accessed May 2002

The subsequent paragraphs summarize the current issuesand difficulties pertaining to the administration andmaintenance of IMS stations:

General aspects:

• The geographical distribution of stations creates notonly technical challenges but also legal and politicalissues, which need to be resolved. Some of theequipment used to monitor environments has beenspecifically developed for the IMS (e.g. partiallyblinded sensors). As a consequence, it is not possibleto make use of previous experience with suchequipment. Thus, dissemination of relevantinformation and new knowledge among stationoperators is an important requirement;

• The IMS has special requirements in terms of up-time, data accuracy and security. In general, the PTShas special interest in analyzing unforeseen problemsand unscheduled repairs;

• Maintenance procedures, work routines, etc. arecurrently described at a high-level only. There is aneed to specify procedures i.e. workflow processes ata more detailed level;

• Software for station equipment and the equipmentitself is provided by different external companies,using different formats and guidelines.

Technical aspects:

• Currently there are several different informationsystems used to provide feedback on problems. Theinformation is mainly used for documentationpurposes and not or only limited available on-line;

• Not all stations have the same level of connectivity.Therefore, offline media (e.g. manuals, CDs) areimportant as well as achieving the highest level oflocal autonomy possible, especially for stations withbad connectivity;

• Any interactive, context-aware system needs to beintegrated into the existing communicationinfrastructure. It is necessary to reuse existing dataformats and aim at a rather lose integration in ordernot to cause any disruption to the existing softwaresystems;

• The existing information systems do not supportstation managers in an adequate and optimal way.One of the systems, the Experts CommunicationSystem (ECS), which we have designed andimplemented at the PTS in 2001/2002, has failed tosupport station operators due to both political reasonsand technical restrictions (secure access, not allrelevant information available, etc.). One of theproblems, as described also in (Grudin 1994 andShipman 1994), pertains to the dilemma of officialvs. unofficial rules. Because of the fact that ECS isprimarily used for exchanging diplomatic andpolitical views, station operators would not revealrelevant information as such information might causepotential political problems. Recent field studies witha dedicated 'Intranet' system for station operators,

provided and hosted by a commercial company,showed a much higher level of acceptance.

Human aspects:

• Due to the fact that the IMS is in the installationprocess some stations are more advanced in terms ofequipment installed and certified and sending data tothe PTS. As a consequence, the levels of experienceand local knowledge differ among station operators.Due to technological progress and the resultingdeployment of new equipment, this will remainrelevant even after the completion of the initialinstallation;

• The above mentioned cultural diversity influencesthe way context-aware guidance and support isperceived and knowledge and information iscommunicated. Another important aspect in thiscontext is the consideration of language barriers;

4 Foundation of Framework

As already mentioned, we propose a new frameworkbased on context-aware and pervasive computingconcepts to support KM in a distributed environment.We intend to provide decision support in the field andassist with the dissemination of relevant information andknowledge among geographically distributedcommunities with different cultural backgrounds.

Our emphasis is on communication and interactionaspects but we need to also investigate rule and process-based policies that are flexible enough to ensure amaximum level of local autonomy. For this purpose, weintroduce a flexible mapping of global processes to localversions, which depend, in turn, on locally available skillsand knowledge.

We further propose an adaptive approach at the technicallevel (e.g. different QoS levels of communicationchannels as described in Jing 1999) as well as at thesemantic level (e.g. use of profiles in combination withdescriptive workflow definitions) (Heinl 1999).Especially descriptive modeling is a good candidate toprovide the required structure of tasks and processes butalso the necessary flexibility to adapt to different levels oflocal expertise, thereby determining the level of localautonomy.

Process/Object Library

P1, O1.......

Pi, Oj........

Pn, Om

Station

Evaluation, Classification,Selection, etc.Load Process/Object

Situation Analysis Module

Store Process/Object

Station Operator

Station Feed

Context Specific Information, Guidance

Context Specific Information

Portable/WearableDevices

Evaluation, Quality Control,Codification, etc.

Feedback Module

Figure 3: Context Aware and Pervasive ComputingFramework - Technical View

The proposed framework represents an integratedapproach, which includes organizational, managerial andtechnological aspects. The key idea is not the inventionof a new technology but the intelligent integration andexploitation of existing concepts and solutions. Inparticular, we propose to integrate open standards toensure platform independence and facilitate thedevelopment and integration of new requirements,concepts and third-party tools.

As part of our effort, we need to address the followingissues:

• Context-specific provision and capture ofinformation and knowledge, using pervasivecomputing;

• Adaptation of mobile workflow processes to localsettings;

• Identification of re-synchronization requirements;

• Identification and implementation of adequatecaching and recovery strategies;

• Robustness of mobile workflow processes;

• Resolution of update conflicts;

• Appropriate user interface design.

As mentioned in section 1, we use IMS as a proof ofconcept. The main idea is to support station operators inthe field by providing context specific information andknowledge, using wearable equipment (see figure 3). Aspointed out before, the consideration of human aspectssuch as cultural differences, different levels ofmotivation, and language barriers will be crucial elementsduring the design and implementation phases.

The following subsections describe and define variousaspects and components of the proposed framework inmore detail. We particularly focus on synchronizationand adaptability aspects of workflow processes.

4.1 Application Scenario

Within the IMS context, four main user groups can beidentified:

• Stations operators;

• National Data Centre (NDC) personnel;

• PTS staff;

• Contractors.

The tasks of the first group, station operators, werealready described in section 1. NDCs are responsible forcollecting raw data, automatic and interactive processingof data as well as providing interactive access to themember state they belong to. They also provide technicalsupport to local station operators. PTS staff in the IMSdivision has basically the same tasks as NDC personnelbut must review data from all stations and provide theresults in a timely manner to all member states, ensuringa high level of quality. Furthermore, PTS staff isresponsible for the overall maintenance andadministration of the IMS, including the archiving of

data. The activities of PTS staff also comprise thedrafting of operational procedures, the testing ofequipment, the supervision and support of the installationof stations, the certification of installed stations, and themonitoring of stations, to list the most important. NDCpersonnel as well as PTS staff need to collaborate andcoordinate with local and global contractors like theoperator of the GCI and provider of third-party products.As a result of this division of responsibilities and thegeographical distribution, relevant knowledge andinformation is not equally distributed and not accessibleto all groups.

Because of this and the fact that all four groups areengaged in various communication processes duringnormal operations and emergency situations, they need tobe included in the framework, even though the primaryfocus is on station operators.

The following three operation types are of relevance toour framework, as they require support in the field as wellas collaboration and communication with others:

• Normal operation (e.g. routine operations as definedin the operational manuals, scheduled maintenancetasks, etc.);

• Handling of exceptional situations (e.g. disruption ofconnectivity, disruption of SW service, brokensensor, broken antenna, database crash, etc.);

• Handling of new situations (e.g. new equipment,experience with different new settings, etc.).

Station Operator

NDC Expert

Review and Classify Feedback

Synchronize Workflow

Authenticate/Log on

Perform routine maintenanceand administration tasks

PTS Expert

Define Workflow Process

Provide Context-specificinformation and tasks

Provide feedback Access Organizational Memory

Handle exceptions

Figure 4: Application-specific Use Cases

In order to meet the special requirements of the IMS, theframework must utilize a modular synchronizationconcept that supports different synchronization andadaptation strategies based on static information but alsocontext-specific information and local knowledgeavailable at the station. The rationale behind it being thatNDC and/or PTS experts should be able to either specifythe synchronization requirements of a given WF processat modeling time i.e. leave the server in full control orleave it up to the client to select the appropriatesynchronization concept, using information about thecontext of the user in connection with the task at hand.

The relevant actors and use cases are depicted in figure 4.

In order to map these use cases to the framework, it isnecessary to define possible strategies to achieve theenvisaged modularity and flexibility.

Roughly speaking, two basic strategies can be identified:

• If the level of local knowledge available at the stationis low, the server (i.e. NDC and/or PTS expert) willprefer to stay in full control and, thus, impose ratherstrict synchronization requirements on the client (i.e.station operator). As a consequence, the level of localautonomy will be low and no or just limited feedbackwill be required from the client;

• If there is a high level of local knowledge the serverwill define rather loose synchronization requirementsi.e. the user will have more flexibility in working ontasks, thereby achieving a higher level of localautonomy. However, the server might have moreinterest in feedback by the client to store anddisseminate this feedback, as appropriate, but also asa means of monitoring the overall WF process.

In order to achieve full flexibility and the highest level oflocal autonomy possible we choose adaptable approachesin the following areas:• Workflow definition;

• Adaptation to local knowledge and expertise (e.g.profile-based);

• Adaptation to local technical factors andrequirements (e.g. QoS).

4.2 Definition of Context in the ApplicationScenario

In Korkea-aho (2000), context is defined as “anyinformation that can be used to characterize the situationof an entity. An entity is a person, place, or object that isconsidered relevant to the interaction between a user andan application, including the user and the applicationthemselves” (Korkea-aho 2000, p. 2). Contextinformation can be identity, spatial information,environmental information (e.g. temperature, light, noiselevel, etc.), social situation, availability of resources,activity, etc.

In figure 5, we provide a high-level view on what wedefine as local context in a remote mobile setting. Theright-hand box represents the main primary factor of thelocal context of the mobile user (the station operator inour example). In our definition, primary factors compriseall knowledge and information sources as well as allcommunication channels that directly determine or affectthe local context of the user. In addition to thetechnology-based primary factors, the context of the useris influenced by ‘off-line’ knowledge, information anddata sources such as manuals, CDs, videotapes, etc.

Last but not least, face-to-face communication with otherusers in the same or similar settings represents a keyprimary factor of the user's context. Although werecognize the important role of face-to-facecommunication it is not the main focus of our workmainly because of the remote mobility aspect and our

global view on disseminating relevant information andknowledge among users in such an environment.Nevertheless we do agree on the importance of utilizinglocal knowledge as a result of social, dynamic andcooperative processes along with the authors of (Fagrell1999). Thus, face-to-face communication should beconsidered as a valuable source of information andknowledge transfer. With reference to our case study,this means that already existing measures like on-sitetraining, regular meetings of station operators and sitesurveys should keep their important role.

Station Operator

Global Process/Object Library

Local Process/Object Library

Mapping of Global to LocalPolicies and Procedures

Context-Specific

Knowledgeand Information

Updates/Synchronization

synchronous

EmailNewsgroupsDiscussiongr.

FAQ...

ChatVideo Conference

...

Common/GlobalServices

asynchronous

synchronous

asynch

ronous

P1, O1...

Pi, Oj...

Pn, Om

P1, O1..

Pi, Oj..

Pl, Ok

Local Policies and ProceduresUser Profiles

Global Policies and Procedures

Profiles

Figure 5: Context of Station Operators - PrimaryFactors

In addition to the above-mentioned primary factors, thereare a number of secondary factors (in the context of ourframework) that have a significant impact on the contextof the mobile user. These factors include culture, socialsettings, and education, among others. As mentionedearlier, we propose to utilize flexible mapping processesand adaptable workflow processes to take into accounttheir different values and qualities.

Using this definition of context as a basis, we areinterested in disseminating relevant information andknowledge in a context-aware manner. Thus, we need togather and evaluate context-specific information onstations to be able to provide problem-specific solutions,for instance. This includes state of health information asit is already produced and sent to the PTS. However, thisinformation is currently not used to actively induce anyworkflow processes or other activities on-site. Inaddition, we need to gather spatial, time and resourceinformation to be able to support station operators in thefield, using, for instance, wearable equipment.

4.3 Definition of Local Autonomy

A central aspect of our framework is the notion of localautonomy which, in our practical context, refers to theability of station operators to conduct scheduled androutine operations and to react on unscheduled technicalproblems in an independent way i.e. without or justlimited online access. Thus, local autonomy comprisesall activities and workflow processes that can be carriedout with limited system support. Consequently, the

notion of local autonomy is closely related to adaptabilityand synchronization of workflow processes.

We define local autonomy also in relation to globalpolicies and procedures that are mapped to local versions,depending on locally available knowledge as well as userprofiles (see figure 5). As a result, local autonomy canhave different levels of occurrence, thereby consideringlocal and individual differences. Ideally, the level ofautonomy should increase over time for standardoperations and exceptions as station operators gatherexperience and relevant knowledge and information.Clearly this means that the experience and knowledge ofthe station operator needs to be reflected in his or herprofiles, which is beyond the scope of this paper.

Apart from the proposed mapping of global policies andprocedures to local versions, local autonomy also requiresa local copy of the maintenance process and object libraryto provide context-aware information and knowledge.For this purpose, the proposed framework must foresee adistribution concept of the global knowledge andinformation repository. Depending on the level ofconnectivity, different update schemas have to bedefined, which, in turn, influence the level of localautonomy.

4.4 Synchronization and Adaptability Aspects

Synchronization and adaptation of workflow processeshas to be dealt with at several different levels: theinfrastructure level (OS, Hardware, communications,etc.); the workflow level (WfMS components); theschema level (workflow schema definition); and theinstance level (application execution of workflow tasks)(Cardoso 2001).

The following bullets list some of the open issues andproblems that need to be addressed in this context:

• Identification and classification of mobile workflowprocesses;

• Classification of global and local workflow processes- the system has to know in advance which subtaskscan in principle be processed in an autonomous way;

• Adaptation of workflow processes in relation toglobal and local policies and rules;

• Handling of disruption and definition ofresynchronization points and rollback procedures;

• Capture of local knowledge and experience as part ofthe workflow process.

The first two bullets are linked to the definition of thedifferent types of disruptions and their occurrence (e.g.before the start of a WF, during the execution and aftercompletion), which, in turn, has to be combined with thedifferent classes of tasks and processes (e.g. some mobileprocesses are more ‘sensitive’ to disruptions than others).This means that the framework must support thedefinition of different task and process classes at themodeling stage. We propose the following two basicclasses of WF processes and tasks, which can be extendedas needed:

• Normal – synchronization requirements depend onglobal and local settings and rules;

• Critical – cannot be executed and/or modified locallybut need to be synchronized with the WfMS.

We propose using a flexible workflow and task definitionapproach to define the basic levels of local autonomy butto allow for different local settings. This means that it isup to the designer of the workflow to indicate whether atask can be modified and executed locally, using, forinstance, late modeling, or whether it requires real-timesynchronization. The idea behind it being that it requiresin-depth understanding and knowledge to define andexploit the maximum level of local autonomy possible,considering also the different levels of local knowledgeand expertise.

In addition to specifying synchronization requirements,the framework must also support the context-specificgathering of relevant information and knowledge (similarto the before mentioned visual markers in Thomas 2002).Thus, the feedback component must be combined withthe workflow component to link this information to thecorresponding process. As discussed in the literature, it isimportant that the actual method used to capture feedbackis unobtrusive and does not interrupt the task at hand.This means, for instance, that the mobile user is notforced to provide feedback in a structured fashion. Aproven way of minimizing obstruction and disruption isthe use of voice recording (Thomas 2002) as well asvisual feedback (e.g. pictures and videos). Therefore,several technical means like voice recording andelectronic notes have to be provided to capture feedback,which is reviewed, evaluated and classified at a laterstage. We discuss the feedback component in moredetail in the next section.

4.5 Visualization and Interface Issues

Visualization plays an important role in KM due to theamount and complexity of information that needs to bedisplayed as well as the adaptation to specialrequirements and restrictions (e.g. limited screen size,unobtrusive provision of information for pilots). Optimaldesign of user interfaces is becoming ever more importantto transfer information and knowledge and, at the sametime, create use acceptance. This is particularly true forAugmented Reality (AR) systems, where visualizationtechniques play a paramount role (Azuma 2001, Brown1998, Fagrell 1999, Fagrell 2000, Hildreth 2000, Pappas2002, Piekarski 1999, Quirchmayr to appear, Sutton2001, Thomas 2000, Thomas 2002, and Tagg 2002).

For our purposes, we also need to look into visualizationand interface aspects regarding the capturing process ofknowledge and information. In particular, we have toinvestigate the following two areas:

• Visualization of context-specific information andknowledge in the field (e.g. availability of resources,quality of communication channel);

• Gathering of context-specific information andknowledge in the field.

We propose to base the visualization concept on deviceand platform independent standards and to separate itfrom any program logic. For this purpose, we follow ann-tier client/proxy-agent/server approach (as described inBialek 2001), using open standards to describe our data.The actual physical devices used to present information tothe user should not affect the storage and business logiclayers on the server but only define the formattingprocess and the set of available functions (depending onphysical attributes like screen size and computing power).From a design point of view the actual interface, incombination with the physical devise used, has to beunobtrusive and intuitive in use to minimize disruption ofany work and social processes (Azuma 2001, Grudin1994, Quirchmayr to appear, Shipman 1994, and Thomas2002).

The more interesting part of the visualization conceptpertains to the interface design to capture context-specificinformation and knowledge. Apart from any automatedmechanisms based on the actions of the user (automatedclassification, automated inference, etc.), we need toprovide an active way of capturing information andknowledge in the field. Again the technology and designused has to be unobtrusive and not too structured andformalized. As pointed out in Shipman (1994), mostexisting examples suffer from enforced formalization i.e.the usage of premature, not suitable structures; cognitiveoverhead; and disregard of individual differences andsituational structures. The authors identify gradualformalization and restructuring as one important remedyto overcome these problems.

4.6 Integration in Existing InformationSystems and Security Infrastructure

Explicit knowledge and information is stored (orsometimes ‘hidden’) in various different informationsystems, using different formats and structures. Access tothese information systems normally requires specialclients, which can make searching over severalinformation sources and combining and accessing morethan one information source a difficult and time-consuming task. While it can be argued that the numberof clients to access the various information systemsshould be minimized (using, for instance, a portal as asingle point of information access), it is in general notfeasible and/or reasonable to unify information systemsbecause of the different requirements, scope and specialsolutions provided by them.

A major aspect of the proposed framework is therefore itsopenness and flexibility in interfacing to otherinformation systems. To this end, the use ELVIN(Fitzpatrick 2000) and its extension to facilitate supportof mobility by introducing a proxy concept (Sutton 2001)is foreseen.

From a practical point of view, the framework mustprovide connections to the following existing informationsystems:

• Document Management System (accessible viaCORBA and Java Beans);

• Database Of the Technical Secretariat (DOTS) toretrieve information on station equipment (accessiblethrough Java Server Pages and Java Beans);

• State of Health Monitoring System.

As security plays a crucial role in the IMS, a system-widePublic Key Infrastructure (PKI) has been put in place todigitally sign and encrypt all data coming from stations.This ensures that the data has not been tampered with.Assuming that maintenance specific information andknowledge are confidential as well, the framework needsto utilize the PKI as well.

Finally, interoperability with the existing structure andsyntax of control commands for IMS stations has to beguaranteed (CTBTO PrepCom 2001).

4.7 Organizational, Managerial andAdministrative Aspects

As stressed elsewhere, we aim to include organizationaland managerial aspects to support the acceptance of theframework, ensure its practical relevance and induce thenecessary individual efforts to actively use it.

The following subsections summarize key aspects that weintend to address in our framework.

4.7.1 Integration of Key Users and Decision-Makers

Many authors have identified the importance of includingusers in the design process, and the development and testphase. In Grudin (1994), for instance, this is referred toas managing acceptance, which basically means tointegrate key users but also to manage the needs anddemands of managers and non-managers. While acontext-aware framework that utilizes and adds to theorganizational memory may immediately appeal to amanager, he or she may underestimate the downsides interms of unwelcome extra work for the users, disruptionof existing and functioning processes, the need to revealtacit knowledge, and so on.

Thus, it is required to include decision-makers as well askey users in the design and development phase. Inparticular, the support of key users is crucial to thesuccess of the proposed framework, as they are the maincustomers of the system and ‘suppliers’ of relevantinformation and knowledge. The identification of keyworkflow processes and the definition of structures andformalisms to support users but also to capture their inputare critical design aspects, which will require the supportand input of key users. They should be chosen accordingto their experience and knowledge but also their ability tocreate opinion, as they will substantially contribute to theoverall perception of the system.

At the management level, it is crucial to get the requiredsupport (or critical mass as it is put in Korkea-aho 2000)to get access to the required resources and to ensure thenecessary funding.

4.7.2 Review Processes

Systems that aim to capture (individual) knowledge andspecific information generally require review proceduresto ensure the accuracy and relevance of capturedknowledge and information. Furthermore, it is necessaryto classify and link new knowledge and information toexisting repositories, which can be a difficult and time-consuming task. There are a number of differentcomputer-supported classification techniques (e.g. graphtheoretic techniques, artificial neural network techniques,evolutionary approaches, etc.) that can support the reviewprocess but, for the time being, it can be assumed that thefinal quality check still lies with human beings.

The importance of reviewing processes is generallyaccepted in the literature. Lundberg (2000), for instance,suggests investigating into the methodologies for buildingnetworks of co-operating partners and the determinationof necessary support systems to handle the contributingand reviewing process effectively and efficiently. Similararguments can be found in (Grudin 1994, Maurer 2001and Shipman 1994).

However, as mentioned before, it is important to designreview and evaluation processes in a way that they do notcreate excessive burden or cognitive overhead for thosecarrying out the processes. It is also suggested to assignroles to people for periodical reviews (Fagrell 1999).

For the design of the review processes we intend to alsouse adaptive review and structuring processes tominimize obstruction and disruption of workflowprocesses.

5 Conclusion and Future Work

We have discussed the foundation of a framework forcontext-aware and pervasive computing that supportsknowledge management in mobile environments. Thedesign of the framework is based on existing conceptsand focuses on synchronization and adaptability aspectsof mobile workflow processes as well as gatheringfeedback in mobile settings.

We have further argued that it is crucial to abandon theprevailing IT-driven view in this area and also focus onthe psychological and cultural aspects of interaction andcommunication. To this end, we include managementand key personnel in the design and deployment processand focus on representation and localization mechanismsthat support different levels of local knowledge andcultural background in a flexible way.

From a practical point of view, the prototypeimplementation is expected to yield a higher level ofquality of maintenance and administration as well as asignificantly higher level of sharing and dissemination ofrelevant information and knowledge among stationmanagers.

The following bullets summarize the expected results andbenefits at the different levels:

Organizational and Managerial level:

• Support for synchronization and adaptability ofworkflow processes;

• Personalization of information and localization;

• Minimizing the drain of knowledge by capturing anddisseminating relevant information and knowledge;

• Definition of adaptive processes for mobile support,reviewing and feedback;

• Means to complement face-to-face communication;

• Consideration of multi-cultural aspects.

Maintenance level:

• Access to context-relevant information, using anydevice (e.g. laptops, PDAs, etc.);

• Single-point of access i.e. one interface to accessinformation and data from various sources such asdocuments, databases, image repositories, etc.;

• Combination of online and offline media;

• Possibility to store/codify maintenance specificinformation and knowledge.

Technical level:

• Definition of domain specific ontology to supportcodification, exchange and retrieval of maintenance-specific information and knowledge;

• Capability to store, share and retrieve informationand knowledge pertaining to the maintenance andadministration of IMS stations;

• Platform and device independent representation ofdata/information (emphasis on wearable equipment);

• Integration of heterogeneous and distributed data andinformation sources;

• Definition of object and process repository to providecontext-specific information and knowledge.

The next steps will involve the detailed technical designof the proposed framework and the implementation of aprototype to be tested with an IMS station.

6 References

Abowd, G.D. et al (1997): Context-awareness in wearableand ubiquitous computing, Proc. of the FirstInternational Symposium on Wearable Computing(ISWC), Boston, MA, 1997, 179-180.

Alonso, G. et al (1995): Exotica/FMDC: HandlingDisconnected Clients in a Workflow ManagementSystem, Proc. of the 3rd International Conference onCooperative Information Systems, Vienna, Austria, 1995.

Alonso, G. et al (1997): Functionality and Limitations ofCurrent Workflow Management Systems, Special Issueon Cooperative Information Systems, IEEE Expert, 1 (9),1997.

Azuma, R. et al (2001): Recent Advances in AugmentedReality, IEEE Computer Graphics and Applications, 21(6), 2001, 34-47.

Bates, J. (1998): The State of the Art in Distributed andDependable Computing, A CaberNet Sponsored Report,1998.

Bialek, R. P. and Peter Brøndum, P. (2001): Thin,Disconnected Clients on a Hospital IT Network, master’sthesis, Computer Science Department (DIKU), Universityof Copenhagen, Denmark, 2001.

Botkin, J. (1999): Smart Business – How KnowledgeCommunities can revolutionize your company. NewYork: The Free Press, 1999.

Bradshaw, J.M. et al (1997): Roles for AgentTechnologies in Knowledge Management: Examplesfrom Applications in Aerospace and Medicine, Proc. ofAAAI Spring Symposium on Artificial Intelligence inKnowledge Management, 1997, 9-16.

Brodie, J. and Perry, M.J. (2001): Designing for Mobileand Distributed Work: Technology Use in RemoteSettings, EPSRC Progress Report, 2001.

Brown, J. S. and Duguid, P. (1998): OrganizingKnowledge, in Cole, R. E. (ed) Knowledge and the Firm,40 (3), 1998, 90-111.

Cabri, G., Leonardi, L. and Zambonelli, F. (1998):Mobile Agent Technology: Current Trends andPerspectives, AICA, Napoli, Italy, 1998.

Chandrasekaran, B., Josephson, J.R. and RichardBenjamins, V. (1998): Ontology of Tasks and Methods,Proc. of the 11th Knowledge Acquisition Modeling andManagement Workshop, Banff, Canada, 1998.

Cardoso, J. et al (2001): Survivability Architecture forWorkflow Systems, Proc. of the 39th Annual ACMSoutheast Conference, Athens, Georgia, 2001.

Chard, A. M. and Sarvay, M. (1997): KnowledgeManagement at Ernst & Young, Graduate School ofBusiness, Stanford University, 1997.

CTBTO PrepCom (2000:1): Operational Manual forHydroacoustic Monitoring and The InternationalExchange of Hydroacoustic Data, classified document,CTBT/WGB/TL-11/3/Rev.8, 2000.

CTBTO PrepCom (2000:2): Operational Manual forInfrasound Monitoring and The International Exchangeof Infrasound Data, classified document,CTBT/WGB/TL-11/4/Rev.8, 2000.

CTBTO PrepCom (2000:3): Operational Manual forRadionuclide Monitoring and The InternationalExchange of Radionuclide Data, classified document,CTBT/WGB/TL-11/5/Rev.8, 2000.

CTBTO PrepCom (2000:4): Operational Manual forSeismological Monitoring and The InternationalExchange of Seismological Data, classified document,CTBT/WGB/TL-11/2/Rev.8, 2000.

IDC Documentation (2001): Formats and Protocols forContinuous Data CD-1.1 Revision 0.2, classifieddocument, 2001.

Davenport, T. H. and Klahr, P. (1998): ManagingCustomer Support Knowledge, in Cole, R. E. (ed)Knowledge and the Firm, 40 (3), 1998, 195-208.

Davenport, T. H. and Hansen, M. T. (1999): KnowledgeManagement at Andersen Consulting, Harvard BusinessSchool, 1999.

Davies, N. et al (1995): Supporting CollaborativeApplications in a Heterogeneous Mobile Environment,Special Issue of Computer Communications on MobileComputing, 19, 1995, 346-358.

Davies, N. et al (1996): Distributed Systems Support forAdaptive Mobile Applications, ACM Mobile Networksand Applications, Special Issue on Mobile Computing -System Services, 1 (4), 1996.

Domingos, H.J. et al (1999): A Workflow Architecture toManage Mobile Collaborative Work, Actas do 1ºEncontro Português de Computação Móvel, 1999.

Dourish, P. (2001:1): Process Descriptions asOrganisational Accounting Devices: The Dual Use ofWorkflow Technologies, Proc. of the 2001 InternationalACM SIGGROUP Conference on Supporting GroupWork, Boulder, Colorado, 2001, 52-60.

Dourish, P. (2001:2): Seeking a Foundation for Context-Aware Computing, Human-Computer Interaction, 16 (2-3), 2001.

Fagrell, H. and Ljungberg, F. (1999): Exploring Supportfor Knowledge Management in Mobile Work, Proc. ofthe sixth European Conference on Computer-SupportedCooperative Work, Copenhagen, Denmark, 1999, 259-275.

Fagrell, H. and Forsberg, K. (2000): Sanneblad, J.,FieldWise: A Mobile Knowledge ManagementArchitecture, Proc. of CSCW’00, Philadelphia, PA, 2000,211-220.

Fahey, L. and Prusak, L. (1998): The Eleven DeadliestSins of Knowledge Management, in Cole, R. E. (ed)Knowledge and the Firm, 40 (3), 1998, 265-276.

Fitzpatrick, G. et al. (2000): Supporting PublicAvailability and Accessibility with Elvin: Experiencesand Reflections, Journal of Collaborative Computing (ToAppear), 2000.

Garvin, D. A. (1997): A Note on KnowledgeManagement, Harvard Business School, 1997.

Grudin, J. (1994): Groupware and Social Dynamics:Eight Challenges for Developers, Communications of theACM, 37 (1), 1994, 92-105.

Hansen, M. T., Nohria, N. and Tierney, Th. (1999):What’s Your Strategy for Managing Knowledge? HarvardBusiness School, 1999, 106-116.

Hayward, B. (1997): Culture, CRM and aviation safety,ANZSASI, Asia Pacific Regional Air Safety Seminar,Brisbane, Australia, 1997.

Heinl, P. et al (1999): A Comprehensive Approach toFlexibility in Workflow Management Systems, Proc. ofthe international joint conference on work activitiescoordination and collaboration, San Francisco, CA,1999, 79 - 88.

Hildreth, P., Kimble, C. and Wright, P. (2000):Communities of Practice in the Distributed InternationalEnvironment, Journal of Knowledge Management, 4 (1),27-38.

Jing, J. et al (1999): Workflow and ApplicationAdaptations in Mobile Environments, Proc. of the SecondIEEE Workshop on Mobile Computing Systems andApplications, New Orleans, Louisiana, 1999, 62-69.

Korkea-aho, M. (2000): Context-Aware ApplicationsSurvey, Internetworking Seminar (Tik-110.551), Sjökulla,Finland, 2000.

Kuppinger, M. and Woywode, M. (2000): Vom Intranetzum Knowledge Management. München Wien: CarlHanser Verlag, 2000.

Luff, P. and Heath, C. (1998): Mobility in Collaboration,Proc. of CSCW’98, Seattle, Washington, 1998, 305-314.

Lundberg, B. G. (2000): The academic organization: anapproach to knowledge management, in Carlsson, S. A. etal (eds) Decision Support through KnowledgeManagement. Edsbruk, 2000, 209-222.

Maurer, H. (1999): The Heart of the Problem: KnowledgeManagement and Knowledge Transfer, Proc. ofENABLE'99, Espoo, Finland, 1999, 8-17.

Maurer, H. and Tochtermann, K. (2001): VierKomponenten des informationstechnischenWissensmanagements: Unternehmensgedächtnisse,Wissensretrieval, Wissenstransfer undWissensvisualisierung und Wissensmanagement undKreativität, Management von nicht-explizitem Wissen:Noch mehr von der Natur lernen, 3, Ulm, Germany,2001, 83-100.

Nonaka, I. (1994): A Dynamic Theory of OrganizationalKnowledge Creation, Organization Science, 5 (1), 1994,14-37.

Pappas, C. et al (2002): A Mobile E-Health System Basedon Workflow Automation Tools, Proc. of the 15th IEEESymposium of Computer-Based Medical Systems,Maribor, Slovenia, 2002.

Piekarski, W., Gunther, B. and Thomas, B. (1999):Integrating Virtual and Augmented Realities in anOutdoor Application, Proc. of the 2nd Int'l Workshop onAugmented Reality, San Francisco, CA, 1999, 45-54.

Purvis, M., Purvis, M. and Lemalu, S. (2000): AnAdaptive Distributed Workflow System Framework,Proc. of the Seventh Asia-Pacific Software EngineeringConference (APSEC’00), Los Alamitos, CA, 2000.

Quirchmayr, G. and Tagg, R. (to appear): AnArchitectural Concept for Knowledge Integration in Inter-Administration Computing, to appear.

Reddy, M. C., Dourish, P. and Pratt, W. (2001):Coordinating Heterogeneous Work: Information andRepresentation in Medical Care, Proc. of the EuropeanConference on Computer Supported Cooperative Work(ECSCW'01), Bonn, Germany, 2001.

Reichert, M., Bauer, T. and Dadam, P. (1999):Enterprise-Wide and Cross-Enterprise WorkflowManagement: Challenges and Research Issues forAdaptive Workflows, Proc. of the Informatik'99Workshop "Enterprise-wide and Cross-enterpriseWorkflow Management: Concepts, Systems, Applications,Paderborn, Germany, 1999.

Shipman, F.M. and Marshall, C.C. (1994): FormalityConsidered Harmful: Experiences, Emerging Themes andDirections, Technical Report ISTL-CSA-94-0802, XeroxResearch Center, Palo Alto, CA, 1994.

Sohlenkamp, M., Prinz, W. and Fuchs, L. (2000):PoliAwac: Design and Evaluation of an AwarenessEnhanced Groupware Client, AI & Society Journal, 14,2000, 31-47.

Sutton, P., Arkins, R. and Segall, B. (2001): SupportingDisconnectedness - Transparent Information Delivery forMobile and Invisible Computing, Proc. of IEEEInternational Symposium on Cluster Computing and theGrid, Brisbane, Australia, 2001.

Tagg, R. and Quirchmayr, G. (2002): An ArchitecturalConcept for Knowledge Integration in Inter-Administration Computing, Proc. of the 3rd Workshop onKnowledge Management in Electronic Government(KMGov 2002), Copenhagen, Denmark, 2002.

Teece, D. J. (1998): Capturing Value from KnowledgeAssets: The New Economy, Markets for Know-how, andIntangible Assets, in Cole, R. E. (ed) Knowledge and theFirm, 40 (3), 1998, 55-79.

Thomas, B. et al (2000): ARQuake: An Outdoor/IndoorAugmented Reality First Person Application, Proc. of theFourth International Symposium on Wearable Computers(ISWC'00), Atlanta, Georgia, 2000.

Thomas, B., Quirchmayr, G. and Piekarski, W. (2002):Through walls communication for medical emergencyservices, International Journal of Human ComputerInteraction, 2002.