A CONTEXT-AWARE ARCHITECTURE FOR

MOBILE KNOWLEDGE MANAGEMENT

Olaf Thiele, Hartwig Knapp and Martin Schader

University of Mannheim - Department of Information Systems

68131 Mannheim, Germany

Nicolas Prat

Essec Business School - Information and Decision Systems Department

Avenue Bernard Hirsch, B.P.50105 - 95021 Cergy cedex, France

Keywords:

Mobility, knowledge management and context-aware computing.

Abstract:

In many professional activities (e.g., medical diagnosis, construction, or sales), the ability to retrieve and

store knowledge in a mobile situation is crucial. This need, together with the progress of mobile devices, has

led to the emergence of mobile knowledge management. The mobile situation imposes speciﬁc constraints

on traditional knowledge management activities (e.g., knowledge retrieval or presentation). Therefore, a key

research question for mobile knowledge management is how context should be taken into account.

In this paper, we propose a context-aware knowledge management architecture for mobile environments. The

key aspect is the “Contextualizer” middleware that takes care of knowledge storage and retrieval as well as

contextually adapted knowledge presentation on mobile devices. In contrast to existing concepts, we suggest a

broader approach, where the middleware serves as the mediator to different mobile devices. We implemented

the architecture in a ﬁtted use case. Our roadside assistance use case demonstrates the architecture’s strength

for presenting both text and graphics stemming from several knowledge sources.

1 INTRODUCTION

The issues of knowledge management and mobility

have often been explored separately. In this paper, we

combine the increased importance of intellectual cap-

ital in today’s business life with the rapidly changing

properties and requirements of mobile devices. The

world of mobile consumer electronics has changed

enormously over the last years. While cell phones,

for example, display high-resolution video streams

and location-based services offer numerous interest-

ing opportunities, business life, on the other hand,

evolves in an analogous way but much slower. Busi-

ness applications are usually more specialized and

therefore more complex and more costly than stan-

dardized consumer software. Some applications like

e-mail messaging or word processors can be used for

both private and business tasks. Still, this kind of ap-

plication is mostly implemented in a proprietary way

(e.g. Blackberry). Research in this area mainly ad-

dresses just one kind of mobile device or application.

Most user interfaces are constructed especially for the

targeted device (e.g. special PalmOS devices; also see

(Bardram and Hansen, 2004)) or functionality is lim-

ited (e.g. WML; also see (Picco et al., 2000)).

Our generalized approach towards mobile knowl-

edge management aims at both, contextually adapted

queries to knowledge sources as well as contextually

adapted presentation of knowledge on mobile devices.

This includes modiﬁed communication with the mo-

bile client according to network variables. While

most research focuses on certain aspects of the prob-

lem, we present a complete architecture, which builds

upon previous work in the ﬁelds of knowledge man-

agement and mobility. A middleware component

called the Contextualizer serves as the main build-

ing block of the architecture. The Contextualizer is

complemented by components that reside on the mo-

bile device and in the back-end. This reduces the

resources needed to include a new mobile device or

knowledge source into the system. The main advan-

tage is that modiﬁcations to the knowledge source

(databases, unstructured knowledge sources, etc.) or

to the mobile devices need only be applied to the

Contextualizer instead of modifying all knowledge

sources and devices.

In the next section, we present the mobile knowl-

edge management environment and point out the

technical prerequisites. In Section 3, we highlight

our proposed architecture and explain its main com-

219

Thiele O., Knapp H., Schader M. and Prat N. (2006).

A CONTEXT-AWARE ARCHITECTURE FOR MOBILE KNOWLEDGE MANAGEMENT.

In Proceedings of the International Conference on Wireless Information Networks and Systems, pages 219-226

 SciTePress

ponents. Section 4 shows parts of our prototypical

implementation while we draw the outline of other

works in Section 5. Finally, we conclude with a re-

sume and an outlook.

2 MOBILE KNOWLEDGE

MANAGEMENT

ENVIRONMENT AND

PROPERTIES

In this section, we highlight the main aspects of

knowledge management and mobility relevant to our

work.

2.1 Knowledge Management

Knowledge management is often described as a sys-

tem of activities to permit the utilization of organi-

zational knowledge by the members of that organi-

zation (Hannig, 2002). Possible knowledge manage-

ment techniques and approaches usually fall into one

of the ﬁve categories: expert ﬁnder, virtual teamwork,

lessons learned databases, case-based reasoning, or

virtual/augmented reality (Derballa and Pousttchi,

2004). The ﬁrst approach is often used to identify and

contact a relevant expert for a speciﬁc problem. The

second one refers to work that is conducted by geo-

graphically distributed coworkers. The next approach

refers to the concept of maintaining positive and neg-

ative experiences in a problem-solving process (see

also (Probst et al., 1997)). The fourth looks for simi-

lar issues in the past and retrieves these solutions. The

last approach assists the user by means of virtual re-

ality. Due to the fact that knowledge management

is a vast ﬁeld of research, further deﬁnitions can be

found in the referred work or in the books by Daven-

port and Prusak (Davenport et al., 1997) or Nonaka

and Takeuchi (Nonaka and Takeuchi, 1995).

In this paper, we focus on four general knowledge

management processes that are relevant to our more

technical view. On the one hand, these are knowledge

presentation and acquisition, which evolve around hu-

man interaction with the system. Knowledge presen-

tation comprises both displaying data on the device as

well as interacting with it. If this interaction leads to

some sort of knowledge storage or if the user delib-

erately enters information through a form or just as

plain text, we categorize these activities as belonging

to knowledge acquisition. On the other hand, knowl-

edge retrieval and storage are relevant to our work.

We use these terms analogous to their technical coun-

terparts.

2.2 Mobility Aspects

Classical knowledge management applications were

programmed for desktop use. With the introduction of

mobility, sophisticated adaptation to the mobile con-

text becomes more important. In general, the unique

properties of mobile devices are a major inﬂuencing

factor in the proposed system. With a growing num-

ber of different device types, the heterogeneity grows

enormously as most devices differ in their capabili-

ties from PCs (e.g. screen size) and from other mo-

bile device types (e.g. cell phones and PDAs). Criti-

cal factors include the display size and resolution, the

memory size and accessibility, the processing power,

connection protocols like GPRS, UMTS, or EDGE,

interaction techniques like the stylus and supported

standardized programming interfaces like Java or C

(see (Lum and Lau, 2002) and more recently (Adipat

and Zhang, 2005)). Key properties are the ubiquity

and the ability to adapt to the user’s context, given

the current position and other preferences (e.g. user

proﬁles).

To sum up, beneﬁts of joining the ﬁelds of

knowledge management and mobility are ”any-

time, anywhere information access” (Grimm et al.,

2005), mobile-added values like ubiquity or context-

sensitivity (Derballa and Pousttchi, 2004), and auto-

matic context incorporation (e.g., knowledge is ac-

cessed in an adapted way and context variables such

as location are stored automatically).

2.3 Requirements

While the limitations of technology supporting

knowledge management activities have already been

discussed elsewhere (see for example (Derballa and

Pousttchi, 2004)), the requirements for mobile de-

vices and their communication partners are a ﬁeld,

which deserves further description. As mentioned

above, mobile devices are limited in several key fea-

tures that are normally fulﬁlled by common PCs or

workstations (e.g. screen size). Other mobile re-

strictions include network issues like bandwidth or

communication protocols. Because of this fact, we

introduced an intelligent middleware - the Contextu-

alizer -, which serves the function to convert a data

stream from the server side into a client-readable for-

mat and back. Therefore, the Contextualizer has a

rough understanding of device types with their prop-

erties and clients transfer their speciﬁc needs (e.g., ex-

act device identiﬁcation, resolution, display size, col-

ors, free memory, number of presentable letters in a

row, etc.) during the communication’s initialization

phase. The middleware, which will be presented in

more detailed in Section 3 receives a response from

one or more of the databases and converts it into the

client’s contextually adapted format. In the other di-

rection, the middleware has to translate a client’s re-

quest for knowledge into a database-understandable

format. This is also described in Section 3.

3 MOBILE KNOWLEDGE

MANAGEMENT

ARCHITECTURE

In this section, we introduce our mobile knowledge

management architecture and its underlying compo-

nents. Subsequently, we explain how the context

mapping works and afterwards, we present the Con-

textualizer middleware in more detail in Section ??.

3.1 Introduction

Our proposed architecture for mobile knowledge

management, shown in ﬁgure 1 on the next page, is

divided into four layers. The ﬁrst layer contains the

mobile users or mobile agents. This human agent

interacts with the system to access knowledge. The

human agent then utilizes this knowledge, transfers

it to others or enters some new knowledge back into

the system. The knowledge management activities,

knowledge presentation and acquisition are the inter-

face between layers one and two. The second layer

consists of Java-enabled mobile devices like mobile

phones, PDAs, or laptop computers. These three de-

vice types are the most common today and usually

all devices of a certain type share similar properties

(e.g., reduced set of keys on cell phones compared

to a complete keyboard on notebook computers). We

programmed client-side pieces of software for each

device type. They serve as small agents on the de-

vice, which automatically determine device proper-

ties such as screen resolution, color table, or band-

width. If the device supports determination of the lo-

cation (by GPS), this data is collected as well and sent

to the Contextualizer, which resides on the third layer.

The implemented middleware serves as the connect-

ing link between the adjoining layers - it is the main

focus of our work. It translates the data ﬂow coming

from the mobile devices to a database query and the

response is then adapted according to the mobile con-

text and sent back to the device. The interfacing ac-

tivities between the third and fourth layer are knowl-

edge storage and retrieval. The fourth layer contains

the knowledge source which is built up of different

database types. We chose the three most popular

types: relational (SQL), object-oriented (OODB), and

XML databases. Due to the fact that most of the in-

telligence and complexity of our application reside on

the third layer, we explain this component more de-

tailed in the next subsection.

3.2 Outline of the Contextualizer

The Contextualizer is a middleware component,

which serves as a mediator between the human agent

and existing knowledge bases. It takes care of com-

munication with the knowledge base (knowledge stor-

age and retrieval) as well as linking the server and

client side according to the context (knowledge pre-

sentation and acquisition). In general terms, if the

Contextualizer receives an XML query from a mo-

bile device, it parses the request and analyzes the con-

text according to the following four context elements:

user dependency, technological environment, situa-

tion, and task. Subsequently, all relevant databases

are queried, the results are transformed according to

the four context elements, and, ﬁnally, the result is

sent back to the client application. All four context el-

ements are described thoroughly in the following sub-

sections. Table 1 gives on overview of typical context

elements.

Table 1: Overview of Context Elements by Category.

Context Ele-

ments

Subelements

User Depen-

dency

Role in the Organization

(function, hierarchical po-

sition), User Preferences

(technical, application

speciﬁc)

Technological

Environment

Handheld Device (screen

size, processing power,

available memory), Net-

work (bandwidth, current

network load)

Situational

Elements

Time (time and time zone),

Location (automatically via

GPS or manually by zip

code)

Task-Speciﬁc

Elements

Heavily dependent on the

task and therefore hard to

pre-determine

Not all context elements need to be considered for

all knowledge management activities. The role of a

user within the organization, for example, has signif-

icant impact on what type of knowledge can be ac-

cessed (retrieval of knowledge) but is independent of

the knowledge presentation. Furthermore, context el-

ements may change during the execution of a task and

therefore need special attention. Again, the role of

the user will usually remain stable but the location or

Figure 1: Mobile Knowledge Management Architecture.

network load may well change. Thus, the distinction

between static and dynamic context elements is im-

portant to our architecture. Finally, some context ele-

ments need to be determined either on the client or the

server side. While the user’s role has to be retrieved

on the server side to prevent misuse, the location (e.g.

by GPS use) is usually retrieved on the mobile de-

vice. Some elements require server and client side

effort (e.g. measuring network load).

We carefully designed this integrated architecture

to meet all our needs. Lei and Georganas distinguish

three different areas (client, proxy, or server), where

an adaptation could be placed in a context-aware ar-

chitecture (Lei and Georganas, 2001). Several argu-

ments support our approach to include some logic on

the client and the back-end with a sophisticated me-

diator like the Contextualizer in between. Thus, we

combine all three approaches with an emphasis on the

proxy. For one, we need to collect as much context

parameters as possible. These are only available on

the client. Therefore, we need a minimum applica-

tion on the client if not a more complex agent soft-

ware. Furthermore, a uniﬁed language for access-

ing knowledge sources allows for selecting chunks

of knowledge as they are needed for the actual con-

text. Driving instructions could be textual, verbal,

sketch-based, or full-size maps. Performing all trans-

formations on the client, for example, would exceed

the processing power of most cell phones, whereas an

implementation centered around only one knowledge

source would hinder spreading the architecture to dif-

ferent tasks.

3.3 User Dependency

User dependent elements form the ﬁrst type of context

elements. They include technical preferences as well

as the agent’s role within the organization. Most tech-

nical preferences are represented through the user’s

proﬁle. A typical user proﬁle includes data on pre-

ferred color themes as well as favored font sizes. Pro-

ﬁle information is stored on the client and is trans-

ferred to the Contextualizer in the initialization phase.

In addition to the proﬁle, the role of the user within

the organization needs to be taken into account. On

the one hand, not all members of an organization work

with all available knowledge. The users’ role deter-

mines what knowledge they are allowed to and need

to access. Authentication and authorization is incor-

porated within the startup phase. For example, all

available data would lead to an information overﬂow

for executive members of the organization. They need

an abstract overview of the accessible knowledge.

3.4 Technological Environment

The technical environment comprises elements of the

device and the network. The device plays a major

role for the Contextualizer. Variables like the screen

size, processing power, available colors, or program-

ming interfaces are key determinants for putting the

knowledge into context. Two major types of applica-

tions need to be considered: text and graphical knowl-

edge representations. Text can be displayed in many

ways. Font size, text formatting and navigational el-

ements within texts (e.g. hyperlinks) as well as ways

for changing screens (e.g. scrolling) can be varied

between mobile devices depending on their capabili-

ties. As for graphical knowledge representation, dis-

play facilities vary a lot between devices. While some

devices like laptops or PDAs offer powerful standard-

ized programming interfaces, smaller devices like cell

phones or smartphones usually have limited custom

interfaces.

The network connection is another important inﬂu-

encing factor. The available bandwidth determines

the transferable data size. Pictures, for example,

need to be compressed or simpliﬁed before transfer.

Furthermore, interaction techniques usually rely on

broadband connections. But the available bandwidth

may differ due to increased network trafﬁc. In this

case, the bandwidth parameters need to be dynami-

cally adjusted. Finally, the costs for network trafﬁc

might be a limiting factor. Some users might prefer

a low-end knowledge representation over a more ex-

pensive one.

3.5 Situational Elements

Situational elements are those connected with time

and location. The time variable includes the actual

time as well as the corresponding time zone. De-

pending on the time of day, knowledge queries might

return different results. Moreover, the time zone is

important when retrieving time-critical information

from databases in far away countries.

The other important situational element is the loca-

tion. The position of the user on the globe is one of

the most prominent contextual elements and is used

in many mobile applications. The location can be de-

termined automatically (via GPS) or manually. The

position of the user is especially interesting for graph-

ical knowledge representation in maps or sketches.

The dynamic nature of this element is also impor-

tant in navigational tasks. Both the Contextualizer

and client need to keep up with a driving vehicle to

deliver timely knowledge.

3.6 Task-Speciﬁc Elements

All other context element implementations, as de-

scribed above, are universal and can be used for most

task types, but elements speciﬁc to the task need to

be adjusted for each task type. To ease the program-

ming effort needed for implementing new tasks, we

developed a semantic XML description format (sim-

ilar to Topic Maps), which can be used to seman-

tically express what knowledge should be retrieved

from which database. The semantic description is sit-

uated in the part named KM meta repository and fur-

ther includes information on interrelated topics. Basi-

cally, the client sends a request to the server, which

determines the resources to access by using these

descriptions. Within the Contextualizer, all data is

represented in a uniform XML format, regardless of

the data source (relational, object-oriented, or native

XML).

4 PROTOTYPE

IMPLEMENTATION

We implemented a sample use case prototype to val-

idate our mobile knowledge management architec-

ture and to illustrate its features. First, a techni-

cal overview of the implementation is given. Subse-

quently, we then present the basic idea of our use case

and we introduce the implemented clients as well as

the corresponding server side.

4.1 Overall Implementation

Generally, all applications (client and server side) are

programmed in Java. The many advantages of Java

include the wide variety of available client program-

ming interfaces as well as a free choice of the server

side operating system. The often mentioned perfor-

mance drawback is not as signiﬁcant for our applica-

tions but might play a role when using more or larger

images. The client-side application has been pro-

grammed using the J2ME programming interfaces.

The widely available components include APIs for

handling images as well as storing information on the

device for caching purposes. The server side and es-

pecially the Contextualizer were programmed using

J2EE. The Java 2 Enterprise Edition APIs offer the

capabilities for handling multiple connections with

clients as well as databases. The Contextualizer is

therefore scalable and performant.

4.2 Roadside Assistance Use Case

The main idea behind the use case is helping both pro-

fessionals and car drivers in assessing problems re-

lated to a car that broke down. The look and feel of

the application is depicted in ﬁgure 2.

The welcome screen is shown in the left half and

a typical GUI element in the right half of the screen-

shot. Due to the high screen resolution and processing

power of modern cell phones (here the Nokia6230i)

Figure 2: Screenshots from the application GUI.

we present these screenshots. The adaptations for

PDAs and laptop computers look similar to those

known from analogous applications (e.g. (Yue et al.,

2005)). The main idea of the use case led to two ba-

sic scenarios: In the ﬁrst one, a professional mechanic

drives to a remote location to solve a car breakdown.

In order to ﬁnd the location, he uses a map given by

our application. Our implementation is shown in ﬁg-

ure 3.

Figure 3: Illustrations from Scenario 1.

Both, the actual position and the location of the ac-

cident or breakdown are shown on the map. The cur-

sor keys may be used to navigate on the map. Once

the mechanic arrives at the destination, the applica-

tion helps him to spot the failure by querying several

databases (e.g. the manufacturer’s one and reports by

colleagues). The mechanic as well as the car driver

might use either a cell phone, a PDA, or a laptop to

access the knowledge management system. The sec-

ond scenario includes the driver of the car, who wants

to ﬁnd a nearby garage while the car is still on the road

or he intends to mark his position on a map. Further-

more, the driver might describe his situation using the

client, which sends the request to the Contextualizer.

There, the request is parsed and sent to the databases

according to the predeﬁned semantic map. The re-

sults are then transferred back to the mobile client. A

possible request to the system is shown in ﬁgure 4.

First, the user describes the problem in short key-

words. Afterwards, several databases are questioned

and the results are then transferred back to the device.

In both scenarios, some knowledge on possible solu-

tions is transferred back into the system. The outside

temperature, for was helpful in solving the problem

Figure 4: Screenshots of Scenario 2.

might ease future use of the system.

4.3 Contextual Adaptation

The implemented use case adapts knowledge accord-

ing to all four context elements. User dependent el-

ements include a user proﬁle and organizational role.

The user is able to choose a certain skin and save fa-

vored top menu items. The organizational role de-

termines, which parts of the system may be accessed

by the user. The mechanic has more access to dif-

ferent knowledge sources than a regular driver (e.g.,

excerpts from a professional database are not easily

understandable). Technical context elements in the

use case include the color, interaction style, screen

size, and bandwidth. The images sent to the device

are adapted according to the given color table. Cer-

tain keywords in responses are highlighted if colors

are supported. The graphics are sent in total if the

device supports interaction types like the stylus or

mouse movements. Only parts of a picture are sent

to a cell phone due to the limited interaction capa-

bilities and the smaller screen. The screen size and

bandwidth determine how much of the map is sent at a

time. Devices with a slow network connection receive

smaller pieces of the map. The bandwidth of a regular

cell phone posed a problematic situation when mov-

ing fast while viewing high resolution images. Sit-

uational elements include the time and place. Prob-

lems differ at night, especially in winter time. Fur-

thermore, a much smaller set of garages is open after

hours. The location plays a major role in this sce-

nario and is monitored constantly. The task speciﬁc

elements were mainly described above and included

map navigation and problem solving.

5 RELATED WORK

While knowledge management and mobility are well-

established research ﬁelds, the combination of both

raises questions and uncertainties. On the more tech-

nical side, for example, we face the problem that con-

nections between mobile devices and databases are

not dependable; a connection and session manage-

ment has to be introduced to guarantee a complete and

timely data exchange. Some (Holliday et al., 2002)

attend to the question how a database query might

be answered completely and discuss several discon-

nection modes while others (see for example (Chan

and Roddick, 2003)) dwell on weak or partial connec-

tion modes. Further work revolves around the ques-

tion how the amounts of data sent back and forth can

be minimized in order to avoid large data transfers in

non-broadband networks (see (Chang et al., 2004) or

(Lindemann and Waldhorst, 2004)).

Work similar to our approach was carried out by

Fagrell et al. as early as 2000 (Fagrell et al., 2000).

As mobile devices had much less processing power

back then, context elements such as user dependency

or automatic location detection were not considered.

The work of Wei and Prehofer focuses more on dis-

tributed context repositories, which are queried for

decision making (Wei and Prehofer, 2003). Derballa

and Pousttchi present the idea of mobile-added val-

ues (Derballa and Pousttchi, 2004). Their approach

is rather abstract and they do not offer a technical

implementation roadmap. Focusing on similar issues

are Grimm et al., who are working on the MUMMY

project (see (Grimm et al., 2005) or (Grimm et al.,

2002)). They developed applications for several tasks

(e.g. mobile facility management, trade fair informa-

tion). Finally, Adipat and Zhang develop an adaptable

framework for mobile knowledge management (Adi-

pat and Zhang, 2005). Their main focus lies on web

content adaptation according to user preferences.

Further research on contextual awareness is carried

out in the ﬁeld of human computer interaction. Dour-

ish (Dourish, 2004) gives a comprehensive overview

of the ﬁeld’s history and current research. An ap-

proach towards mobile knowledge management from

that research ﬁeld is presented by Shen (Shen, 2003)

or Bardram and Hansen (Bardram and Hansen, 2004).

Finally, work on mobile knowledge management ex-

ists that centers around peer to peer exchange of

knowledge. Schwotzer and Geihs present an architec-

ture for topic map exchange (Schwotzer and Geihs,

2002).

6 CONCLUSION

Mobile knowledge management is a key application

area for mobile computing in general and context-

aware computing in particular. In this paper, we have

presented a context-aware mobile knowledge man-

agement architecture. The key component of the ar-

chitecture is the Contextualizer. This middleware im-

plements the knowledge management activities (e.g.,

knowledge storage, retrieval, presentation, and acqui-

sition) by taking into account the context elements,

namely the user, the technical environment, the situa-

tion (i.e. time and place) and the speciﬁc task at hand.

Moreover, in contrast to existing approaches, the Con-

textualizer comprises a meta-repository, which de-

cides dynamically where the requested knowledge re-

sides. We have implemented a prototype of our mo-

bile knowledge management architecture and applied

it to a scenario.

Our prototype architecture needs to be extended in

order to take into account all mobile knowledge man-

agement activities (e.g. better knowledge acquisition

techniques like forms and recording audio). The scal-

ability of the architecture requires further testing (a

wider range of devices and media types) and applica-

tions to real-life situations (e.g. handheld audits). We

are currently working on these issues.

REFERENCES

Adipat, B. and Zhang, D. (2005). Developing adaptive

and personalized mobile applications: A framework

and design issues. In Proceedings of the Eleventh

Americas Conference on Information Systems (AM-

CIS 2005). Omaha, Nebraska.

Bardram, J. E. and Hansen, T. R. (2004). The aware archi-

tecture: supporting context-mediated social awareness

in mobile cooperation. In CSCW ’04: Proceedings of

the 2004 ACM conference on Computer supported co-

operative work, pages 192–201, New York, NY, USA.

ACM Press.

Chan, D. and Roddick, J. (2003). Context-sensitive mobile

database summarisation. In ACSC 2003. Twenty-Sixth

Australasian Computer Science Conference.

Chang, T.-Y., Velayutham, A., and Sivakumar, R. (2004).

Mimic: raw activity shipping for ﬁle synchronization

in mobile ﬁle systems. In MobiSys ’04: Proceedings

of the 2nd international conference on Mobile sys-

tems, applications, and services, pages 165–176, New

York, NY, USA. ACM Press.

Davenport, T. H., Prusak, L., and Prusak, L. (1997). Work-

ing Knowledge: How Organizations Manage What

They Know. Harvard Business School Press, Boston,

MA, USA.

Derballa, V. and Pousttchi, K. (2004). Extending knowl-

edge management to mobile workplaces. In ICEC

2004. Sixth International Conference on Electronic

Commerce, pages 583– 590.

Dourish, P. (2004). What we talk about when we talk about

context. Personal Ubiquitous Comput., 8(1):19–30.

Fagrell, H., Forsberg, K., and Sanneblad, J. (2000). Field-

wise: A mobile knowledge management architecture.

In CSCW 2000, pages 211– 220.

Grimm, M., Tazari, M.-R., and Balfanz, D. (2002). Lec-

ture Notes in Computer Science 2569, chapter To-

wards a Framework for Mobile Knowledge Manage-

ment, pages 326–338. Springer.

Grimm, M., Tazari, M.-R., and Balfanz, D. (2005). A ref-

erence model for mobile knowledge management. In

Proceedings of I-KNOW 05. Graz, Austria.

Hannig, U. (2002). Knowledge Management and Business

Intelligence, chapter Zwei Welten wachsen zusam-

men. Springer, Berlin.

Holliday, J., Agrawal, D., and El Abbadi, A. (2002). Dis-

connection modes for mobile devices. In Wireless Net-

works 8, pages 391– 402.

Lei, Z. and Georganas, N. (2001). Context-based me-

dia adaptation in pervasive computing. In Canadian

Conference on Electrical and Computer Engineering,

2001, pages 913–918 vol.2. IEEE Press.

Lindemann, C. and Waldhorst, O. P. (2004). Exploiting epi-

demic data dissemination for consistent lookup opera-

tions in mobile applications. SIGMOBILE Mob. Com-

put. Commun. Rev., 8(3):44–56.

Lum, W. Y. and Lau, F. C. M. (2002). A context-aware de-

cision engine for content adaptation. IEEE Pervasive

Computing, 1(3):41–49.

Nonaka, I. and Takeuchi, H. (1995). The Knowledge-

Creating Company : How Japanese Companies Cre-

ate the Dynamics of Innovation. Oxford University

Press.

Picco, G. P., Murphy, A. L., and Roman, G.-C. (2000).

Developing mobile computing applications with lime.

In ICSE ’00: Proceedings of the 22nd international

conference on Software engineering, pages 766–769,

New York, NY, USA. ACM Press.

Probst, G., Raub, S., and Romhardt, K. (1997). Wissen

managen, chapter Wie Unternehmen ihre wertvollste

Ressource optimal nutzen. Gabler, Wiesbaden.

Schwotzer, T. and Geihs, K. (2002). Shark - a system

for management, synchronization and exchange of

knowledge in mobile user groups. In 2nd Inter-

national Conference on Knowledge Management (I-

KNOW ’02), pages 149–155. Graz, Austria.

Shen, J. (2003). Utilizing mobile devices to capture case

stories for knowledge management. In CHI ’03: CHI

’03 extended abstracts on Human factors in comput-

ing systems, pages 688–689, New York, NY, USA.

ACM Press.

Wei, Q. and Prehofer, C. (2003). Context management

in mobile environments. In Proceedings of ANwire

Workshop, Paris.

Yue, W., Mu, S., Wang, H., and Wang, G. (2005). Tgh: a

case study of designing natural interaction for mobile

guide systems. In MobileHCI ’05: Proceedings of the

7th international conference on Human computer in-

teraction with mobile devices & services, pages 199–

206, New York, NY, USA. ACM Press.