USING TASK HISTORIES TO SUPPORT PERSON-TO-PERSON

KNOWLEDGE EXCHANGE

Extracting and using Contextual Overlap and Levels of Expertise to Connect

Knowledge Workers

Jörg Schmidl and Helmut Krcmar

Department of Informatics, Technische Universität München, Boltzmannstr. 3, Munich, Germany

Keywords: Expert mediation, Expert recommendation, Knowledge networks, Yellow page systems, Person-to-person

knowledge exchange, Social capital theory, Transactive memory systems, Cognitive motivation theory,

Problem solving theory.

Abstract: Knowledge within organizations is increasingly distributed, which raises the challenge to connect the right

individuals for knowledge exchange when needed. In this contribution we analyze this challenge in detail

and propose a concept to connect the right individuals by relying on the task histories of the knowledge

workers. We first investigate relevant theoretical models such as transactive memory theory, social capital

theory for knowledge exchange and a model based on socio-motivational and problem solving theory to find

relevant constructs. We then analyze the relevant state-of-the-art to find that all approaches have some

limitations with respect to the theoretical models. Our proposed solution to the challenge builds on using

task histories for the matching, and we show how it can be used to determine contextual overlap and level of

expertise – the first one is an adequate indicator for willingness to interact while the second one is an

indicator for ability to have a fruitful interaction. We then describe a case study in which we employed our

concept in a three month timeframe with 93 individuals. A survey after the case study shows that our

assumptions concerning the relevance and benefit of context overlap are substantiated.

1 INTRODUCTION

With the increase in relative importance of

knowledge for the success of a company there also is

a change in the role the single individual plays.

Nowadays, workforce increasingly consists of

knowledge workers and hence suitable support of

their work becomes more important (Davenport,

2011). According to Drucker (1988) knowledge

workers are specialists in their profession who

govern their work on their own adapting their

performance to feedback from their environment.

However, owing to the increasing diversification

and radical advancements in knowledge, knowledge

workers’ specialization is naturally limited to certain

areas – there are no universal geniuses like Leonard

DaVinci anymore. Hence an organizations is often

seen as “[...] a society of knowledge workers who

are interconnected by a computerized infrastructure”

(Holsapple, 1987) and a fundamental challenge for

organizations lies in the systematic coordination of

knowledge in this network (Quinn, 1992).

Leveraging the potential of this network of

knowledge workers can be done in two stereotypic

ways (Davenport, 2011): Either, the general goal is

to give knowledge workers access to as many and as

diverse sources of information, including fellow

knowledge workers, and assume that they will

handle and integrate the information autonomously.

Or, alternatively, the information delivered to the

knowledge worker is governed to a larger degree by

structured processes and systems.

While structured delivery is well-suited for tasks

that follow a routine, pursuing a free access model

assumes that knowledge workers know what

information they can use, how to manage it and how

to find it. However, “[…] workers may know how to

use technology tools, they may not be skilled at

searching for, using, or sharing the knowledge.”

(Davenport, 2011) and hence if possible some

structure should be imposed to guide the knowledge

workers. This also applies when accessing

Schmidl J. and Krcmar H..

USING TASK HISTORIES TO SUPPORT PERSON-TO-PERSON KNOWLEDGE EXCHANGE - Extracting and using Contextual Overlap and Levels of

Expertise to Connect Knowledge Workers.

DOI: 10.5220/0003661300770086

In Proceedings of the International Conference on Knowledge Management and Information Sharing (KMIS-2011), pages 77-86

ISBN: 978-989-8425-81-2

 2011 SCITEPRESS (Science and Technology Publications, Lda.)

knowledge by interacting with others that serve as

source of help for a concrete challenge, i.e. in

person-to-person knowledge exchange.

However, owing to specialization and the

resulting diversification of knowledge in companies,

finding suitable experts is a challenge, further fueled

by geographic distance, time-zone differences and

large pools of potential candidates typically found in

larger organizations. Therefore, the limited human

attention puts a natural limit to the ability to

collaborate (Qureshi, 2006) while at the same time a

knowledge worker’s attention becomes a crucial

resource that should be handled with care when

searching for interaction partners (Ye, 2008).

Suitable solutions should therefore induce only

small effort for the knowledge seeking individual

and at the same time, in a global view, should limit

the attention consumed for possible interaction

partners. Both taken together can increase the

likelihood of two knowledge workers exchanging

information and hence contribute to fostering

knowledge exchange in organizations.

The remainder of this contribution is structured

as follows. In chapter 2 we discuss relevant

theoretical models that describe person-to-person

knowledge transfer processes. In chapter 3 we

investigate approaches currently employed in

corporate settings to facilitate person-to-person

knowledge transfer. We especially relate them to the

constructs of the theoretical models discussed in

chapter 2 and indicate where they face challenges.

Chapter 4 describes our concept for addressing the

challenges found in chapter 3 that also complies

with the constructs described in the theoretical

models. A discussion about the conceptual design

choices of the concept is dealt with in chapter 5. The

subsequent chapter describes the evaluation of our

concept in a case study, while chapter 7 concludes

the contribution with a summary of results and an

outlook to further research.

2 THEORETICAL BASIS

FOR PERSON-TO-PERSON

KNOWLEDGE EXCHANGE

When knowledge is exchanged from person to

person a social process between the two actors starts.

Therefore, to understand the antecedents and

constituting steps in this social process, appropriate

models from sociology and psychology that describe

this process need to form the basis for any further

design. For our discussion we will rely on one model

from social psychology, another from sociology and

a third one from cognitive psychology.

2.1 Transactive Memory Systems

According to the theory’s originator transactive

memory systems can be described as “[...] a set of

individual memory systems in combination with the

communication that takes place between

individuals” (Wegner, 1985). Individuals use others

as their “external storage” by remembering pointers

to those that possess relevant knowledge, i.e. by

establishing know-who, instead of remembering the

procedural or factual knowledge itself. Transactive

memory system (TMS) theory also encompasses

processes that determine who is responsible to store

new knowledge on behalf of the group and processes

to later disseminate it within the group.

While the theory tries to explain knowledge

transfer processes between individuals, its unit of

analysis are groups that consist of individuals

acquainted to each other. Especially the necessity of

individuals to assume responsibility for the group

necessitates a binding element that socially

motivates the individual to contribute. In the original

TMS theory this binding element is personal

acquaintance, i.e. an established social fabric that

manifests in strong relationships between a group’s

individuals. In non-co-located, dispersed

organizations, close relationships across team or

department are very rare, which is why personal

acquaintance may not act as a binding element.

Appropriate approaches for supporting person-

to-person knowledge exchange need to have a

suitable surrogate for those strong interpersonal

relationships that still allows to establish a form of

TMS.

2.2 Social Capital Theory

Social capital (see for example (Lin, 2001) for an

overview) relates to an individual’s previously

established connections to (known) others, their

strength and reliability and the individual’s ability to

take benefit out of this network. The theory has been

adapted to explain knowledge exchange processes

(Nahapiet, 1998) which subsequently has been

applied to empirically study this process in

electronic networks of practice (Wasko, 2005; Law,

2008). Here, the original concept of social capital

had to be relaxed. While in its original form for

social capital to build up, it is necessary to know the

other individual so that later one’s own effort for an

individual may be reciprocated by this individual,

KMIS 2011 - International Conference on Knowledge Management and Information Sharing

electronic networks of practice are effectively

anonymous and individuals do not know each other.

Therefore, in electronic networks of practice other

constructs surrogate for this. Statistical evidence

could be found among others for the following

constructs (Wasko, 2005; Law, 2008): 1) The more

communication threads an individual has with others

the more likely he is to contribute, which is

subsumed under the concept network centrality 2)

During interactions the more one can rely on a

shared language the higher the willingness to

contribute 3) The more an individual can identify

with the network or more precisely identify with the

interaction partners in this network the more likely

he will contribute.

Approaches that foster knowledge exchange

have to adopt these three aspects.

2.3 Model based on Problem Solving

and Cognitive Motivation Theory

Olivera et al. (2008) developed a model to describe

how and why people contribute in distributed

organizations through IT-mediated means. They

argue that to understand the contribution behavior,

two strands of theories have to be combined:

theories of problem solving (Newell, 1972) and

cognitive motivation theories (Kanfer, 1990). The

model distinguishes three subsequent mediating

mechanisms. The first, awareness, relates to a

person recognizing an opportunity to contribute. In

the second, searching and matching, the individual

determines whether and how his knowledge is

sufficient to help another individual. The third

mechanism, formulation and delivery, deals with

formulating and communicating the individual’s

knowledge to help the other. Each of these

mechanisms inflicts costs for the individual who can

possibly help. The necessary overcompensation of

these costs is described by constructs from cognitive

motivation theory. The relevant ones are: Self-

enhancement, that is fueled by e.g. liking to express

one’s expertise (Wasko, 2000) and living up to one’s

self-identity (Constant, 1996); Exchange motivations

which is the equivalent to expectation of (individual

or generalized) reciprocity as discussed in relation to

social capital theory; Instrumental motivation which

refers to rewards such as recognition.

Appropriate approaches to support person-to-

person knowledge exchange should hence try to

support the aforementioned three mechanisms and

build upon the three means of motivation.

3 STATE-OF-THE-ART IN

PERSON-TO-PERSON

KNOWLEDGE EXCHANGE

In light of the theoretical models describing the

antecedents and constituting steps of person-to-

person knowledge exchange, in this chapter we want

to investigate the properties of contemporary

approaches supporting person-to-person knowledge

exchange.

There are diverse approaches and tools that quite

directly or more indirectly fall under the umbrella

term Knowledge Management. Binney has arranged

them according to a spectrum from transactional

systems to innovation-supporting systems (Binney,

2001) while others, e.g. (Böhmann, 2002) suggested

to use the SECI model (Nonaka, 1994) to impose

structure on the set of approaches and tools. Using

the respective structure’s dimensions and contrasting

those with the situation we look at – person-to-

person knowledge transfer, for complex, highly-

adaptable, knowledge-intensive tasks in a distributed

setting – we find that three approaches fall into a

comparable category with respect to our research:

Yellow Page Systems, Expert Recommender

Systems, and Knowledge Networks.

3.1 Yellow Pages Systems

A yellow page system (YPS) contains lists of the

individuals in an organization along with their

competencies, knowledge and skills in those areas

that are relevant to the organization. A knowledge

seeker may search for required knowledge and will

be presented with those individuals that match the

request. If the system contains levels of proficiency,

the result may also be ranked. The profiles contained

in the system are often manually maintained, while

some data may be extracted from directory services

(Krcmar, 2010) or Human Capital Management

systems (Gronau, 2004). Also, keeping those

profiles up-to-date is normally a manual process.

TMS’ backbone are interpersonal relationships

that in distributed settings need to be surrogated by

other means. The query mechanism in YPS typically

only operates on the level of expertise to find

relevant matches, neglecting the previous

relationship between actors and hence there is no

obvious surrogate for interpersonal relationships. On

the other hand, the application of social capital

theory on knowledge exchange suggests that many

communication threads increase likelihood to

contribute. YPS have no means to increase this

USING TASK HISTORIES TO SUPPORT PERSON-TO-PERSON KNOWLEDGE EXCHANGE - Extracting and using

Contextual Overlap and Levels of Expertise to Connect Knowledge Workers

number nor is it typically tracked. Also, shared

language, another positive influence for knowledge

contribution according to social capital theory, is not

part of the matching of knowledge requester and

potential helper. Identification with potential others

is not part of the matching, but result lists of queries

may contain affiliation and other socio-metric

information that may serve this purpose. In terms of

socio-cognitive and problem solving theory, YPS

have no means for creating awareness on the side of

the potential helper for requests of a knowledge

seeker – it follows a pull interaction schema.

However, match of request and expertise tends to be

high, if the profiles are up-to-date. YPS include no

measures that facilitate the formulation of responses,

though. Also, the motivation factors such as self-

identity are not specifically supported by the

matching delivered by YPS. Reciprocity on the other

hand is often a motivator in YPS settings as the

individuals are acquainted after the interaction.

Many of the aspects that theory predicts to be

important for supporting person-to-person

knowledge exchange are not present in YPS. Along

with the relatively high effort for keeping YPS up-

to-date, they seem to leave room for improvement.

3.2 Expert Recommender Systems

Expert recommender systems (ERS) can be seen as

the next evolution step of YPS. Tasks such as profile

generation and mediation of communication are

automated. ERS help in the following way: When

someone seeks an expert, he wants to know if there

is an expert that can answer the user's questions, but

also what level of expertise the user has and how it

compares to others, if there are others that also fulfill

the criteria and how the person can be reached (Seid,

2003). The automation of expert determination is

achieved by deriving levels of expertise in relation

to queried knowledge items from sources that may

be scanned for expertise evidence. Those sources

can be communication-based, such as e-mail

messages, document-based, such as websites or

electronic documents stored on intranets or

interaction-based where software usage is utilized as

source of expertise evidence.

Many of the aspects that theory predicts to be

important are however not present in ERS.

Considering suitable surrogates for TMS’

interpersonal relationships, ERS do usually not

include means for this. The query mechanism in

ERS typically only operates on the level of expertise

to find relevant matches, neglecting the previous

relationship between actors and hence there is no

obvious surrogate for interpersonal relationships.

Only few attempts can be found to somewhat

remedy this downside e.g. (Serdyukov, 2009). On

the other hand, the application of social capital

theory on knowledge exchange suggests that many

communication threads increase the likelihood to

contribute. ERS have no means to increase this

number nor is it typically tracked. Also shared

language, another positive influence for knowledge

contribution according to social capital theory, is

typically not part of the matching of knowledge

requester and potential helper, again with only few

exceptions. As was true for YPS, identification with

others is not part of the matching, but results of

queries may contain affiliation and other socio-

metric information that may serve this purpose. In

terms of socio-cognitive and problem solving theory,

ERS have no means for creating awareness on the

side of the potential helper for requests of a

knowledge seeker – it also follows a pull interaction

schema. However, match of request and expertise

tends to be high, if the expertise extraction

mechanism fits the users’ expectations. ERS include

no measures that facilitate the formulation of

responses, though. Also, the motivation factors such

as self-identity are not specifically supported by the

matching delivered by ERS. Reciprocity on the other

hand is often a motivator in ERS settings as the

individuals are acquainted after the interaction.

Many of the aspects that theory predicts to be

important for supporting person-to-person

knowledge exchange are also not present in ERS,

while due to its decreased effort they appear more

promising than YPS.

3.3 Electronic Networks of Practice

Electronic networks of practice (ENP) are a

geographically distributed group of individuals that

are engaged in a shared practice. However in

contrast to other forms of knowledge networks, the

group of individuals can be large, virtually limitless

in size, the individuals are loosely knit, but may not

know each other at all nor necessarily do they expect

to ever meet face-to-face (Brown, 2001; Wasko,

2005). By relying on IT-mediated communication,

ENP allow quick and effortless access to a broad

source of expertise through a wide variety of

knowledge carriers (Teigland, 2003).

In relation to TMS’ backbone – interpersonal

relationships – the interactions in an ENP support

the creation of weak ties. Those ties are far less

reliable and pronounced than those between

acquainted individuals; however, they are a suitable

KMIS 2011 - International Conference on Knowledge Management and Information Sharing

surrogate for distributed setting where people

normally never meet face to face. In contrast to the

two previous approaches, ENP do not offer explicit

mechanisms to find suitable interaction partners. It is

rather up to the individual to find relevant

individuals or, more often, relevant outlets within

the ENP, such as a forum concerned with his

knowledge request. Social capital theory’s indication

that many communication threads are beneficial can

be supported in ENP, as many of them feature

mechanisms to be informed by updates in parts of

the ENP, e.g. forums, which are relevant to the

individual. This measure can also increase

awareness of knowledge requests on the knowledge

bearer’s side. In ENP, shared language establishes

over time and with more and more interactions. The

same holds for the ability to identify with others that

also increases over time while being a member of

the ENP. However, the challenge of identifying with

individuals without previous interaction history

remains, especially since socio-metric information is

often not available in ENP. Another challenge lies in

matching available and requested expertise: If the

requester does not know where or who to ask he is

hindered and if the question is addressed to the

wrong individuals, answers are unlikely and effort is

wasted on the side of all affected individuals.

Requests for knowledge and responses are also not

associated with the work context of neither

requesting nor replying individual and hence to

some degree decontextualized, which may affect

ease of request and response formulation.

Reciprocity is often present in ENP - in its

generalized form, though. Also self-identity, another

motivation factor, may be reinforced by the ENP

itself, as other like-minded individuals are likely to

find one’s interactions in the ENP.

While many of the social factors of the

theoretical models can be matched onto features of

ENP, there remain challenges such as facilitating the

searching, finding and matching of interaction

partners or limiting consumption of awareness.

4 TASK HISTORY AS BINDING

ELEMENT

The theoretical models discussed in chapter 2

described the mechanisms that allow knowledge

transfer to happen mainly from the point of view of

the contributor, i.e. the knowledge bearer helping the

knowledge seeker. The knowledge request was

treated as a given prerequisite. However

understanding the knowledge seeker’s intention is

relevant as well. Seid (2003) analyzed which

circumstances lead someone to consult an expert.

First, someone might need access to information that

is not documented. Second, someone might not be

able to exactly specify what he needs to know, rather

the dialogue with an expert acts as the process to

facilitate information acquisition. Third, someone

might want to utilize an expert to be more efficient.

Someone with advanced expertise can handle tasks

faster than novices can: Therefore, relying on the

expert improves the initial individual’s efficiency.

Fourth, often users do not want a context-free,

general piece of information but rather need a

contextualized, situated interpretation of more

general knowledge that the expert might poses.

Fifth, someone might simply prefer relying on social

interaction instead of using anonymous media like

documents. Therefore, sufficient levels of expertise

on the side of the potential helper is important for

the knowledge seeker and hence for the knowledge

transfer to start and to be successful.

On the other hand, next to being able to help, the

helper needs to be willing to help and the requester

needs to be willing to ask this individual for help.

TMS address this aspect by stressing the importance

of established interpersonal relationships. In social

capital theory identification with others and use of

shared language expresses this aspect. The cognitive

psychology model expresses this aspect in the

constructs self-identity and recognition. In a

distributed setting interaction partners are

unacquainted and do not have a previous history of

interactions. Nevertheless, an appropriate

mechanism to determine suitable interaction partners

that reflects the constructs of those theoretical

models is necessary.

As both, ability and willingness to help, are

relevant for knowledge exchange to happen on the

knowledge bearer side, and both willingness to

accept help and ability to understand the offered

help are necessary on the requester side, a suitable

glue needs to be found. We suggest using the

histories of tasks performed by and knowledge

bearers as this binding element.

History of tasks refers to the tasks a knowledge

worker is currently engaged in or has been

performing in the recent past. Nowadays, large

portions of a company’s operations are supported by

information systems and a large extend of

knowledge work is as well. In some organizations

that are customer service-focused up to 75 % of a

knowledge worker’s tasks are IT-supported

(Makolm, 2007). Therefore the current and previous

task context of a knowledge worker is often

USING TASK HISTORIES TO SUPPORT PERSON-TO-PERSON KNOWLEDGE EXCHANGE - Extracting and using

Contextual Overlap and Levels of Expertise to Connect Knowledge Workers

adequately reflected in IT system use that can be

extracted from the logs that those systems create for

administrative purpose.

The history of tasks can serve both purposes that

we elicited to be important: ability and willingness

to engage in knowledge transfer for both, knowledge

seeker and knowledge bearer. The more often a

knowledge worker has performed a specific task, the

higher the chance that he has proficiency in

performing the task. Therefore the number of times a

knowledge worker performs a task, can be used to

determine his proficiency related to this task – an

assumption that is often taken as valid (Seid, 2003).

On the other hand, the history of tasks, especially its

very recent or current part, gives indication of the

knowledge worker’s current work context. If the

work context of the knowledge seeker and the one of

the knowledge bearer overlap, they are more likely

to engage and benefit from interaction. This

assumption is backed by fundamental results from

socio-psychology, with one of its clearest results

being, that one likes others that are similar to oneself

(Zimbardo, 1983) and that we identify with those

that are similar to us (Tajfel, 1986; Turner, 1987).

Other psychological results further support this

aspect. Similar attitudes were shown to predict

interpersonal attraction (Byrne, 1971) and joint

interest and mutual trust also correlate (Ziegler, 2007).

While not nearly as expressed, this contextual

overlap is a surrogate for the interpersonal

relationships that TMS has as its backbone, for a

setting in which personal acquaintance is scarce or

non-existent. Also the relevant constructs of social

capital theory can be supported. Being in similar

work context increases the ability to rely on the

same task-specific terminology and hence use of

shared language is possible, as is the identification

with the other as reasoned above. Context overlap

also translates nicely to the constructs used in the

model relying on problem solving and socio-

cognitive theory. When work contexts overlap, the

knowledge bearer can more easily determine how

his knowledge matches with the request and the

likelihood that it does is higher, as the request relates

to what he currently does or has done just recently.

Also, being in similar work contexts facilitates

formulation of responses as it is possible to rely on

shared terminology. The motivational aspects of the

model map to context overlap as well. Being

recognized as expert is a strong motivator that is

even higher when the recognition comes from

individuals that are similar to one. This is in line

with social comparison theory (Festinger, 1954) that

states that we want to be better than our reference

group of similar peers. Similarly, self-identity is

more pronounced when one can help in areas that

are relevant to oneself, which applies for tasks that

one is currently doing or has done just recently.

5 EXPERTISE VS. CONTEXTUAL

OVERLAP

In the previous chapter we have argued that task

history may serve the purpose of identifying levels

of expertise and at the same time may act as

surrogate for determining willingness to interact due

to similarity in task context and hence situation and

previous history. Finding suitable interaction

partners essentially is a filtering task, as otherwise

requests for help could just be broadcasted to all

individuals in an organization. When filtering, the

question shifts to determining which dimensions to

filter on and which filter values to set for them. As

we argued, level of expertise as well as contextual

overlap are relevant and form the two dimensions

we may use for filtering. Figure 1 illustrates the four

different ways to configure the filtering values.

Figure 1: Filtering thresholds operating on contextual

overlap and level of expertise.

One way to filter out non-suitable individuals lies in

choosing only those that exhibit a minimum level of

expertness. This minimum value can be absolute,

e.g. only individuals who have performed a task

more than ten times, or it can be relative to the

knowledge seeker, e.g. only individuals that have

performed a task at least five times more than the

knowledge seeker. If an individual is below this

threshold (rectangle II and IV), he is filtered out,

while the ones that are above are eligible but we

may choose to rank them according to the second

dimension and only consider a fixed number of them

III

IIIIV

Contextual overlap

Level of expertise

Minimal context

threshold

Minimal expertise

threshold

Minimal combined

threshold

KMIS 2011 - International Conference on Knowledge Management and Information Sharing

that have most contextual overlap.

Another way to filter lies in defining a minimum

value of contextual overlap to find those individuals

that are possibly willing to interact with the

knowledge seeker and that also can do so with only

small effort because their mental models are alike.

Again, the ones that are below the threshold

(rectangle III and IV) are filtered out, while the

remaining ones are all eligible, but we may choose

to rank them according to the first dimension and

only consider a fixed number of them that have

highest levels of expertise. However, this may

induce the problem that also arises in expert

recommendation systems. Experts with much higher

levels of expertise face a mental challenge when

interacting with (relatively seen) laypersons. Their

mental models about the relevant topics are

“compressed”, single facts are aggregated into larger

chunks and abstractions are used to condense the

relevant knowledge. To help, experts have to unpack

these chunks, which is a high mental effort as they

must undo their previous learning to understand the

requester’s problem context (Bromme, 2004).

A third way of filtering combines the thresholds

defined on the level of expertise dimension and on

the contextual overlap dimension. In this case, those

individuals that have insufficient expertise (rectangle

III and IV) and those that do not share sufficient

context (additionally rectangle II) are filtered out,

while the ones in rectangle I are suitable candidates.

A final way of filtering lies in defining a

combined threshold that takes into account level of

expertise and contextual overlap at the same time.

Individuals who are very similar to the knowledge

seeker but have only low levels of expertise may be

suitable, while also individuals that do not share

much commonalities with the knowledge seeker but

are very knowledgeable may be suitable candidates

as well. Therefore, in both cases one dimension

might compensate the lack in the other. However,

individuals that neither have a sufficient contextual

overlap nor sufficient levels of expertise (darker area

in the lower left part) are filtered out.

6 APPLICATION OF CONCEPT

To evaluate the concept of using histories of tasks to

foster knowledge exchange between knowledge

workers, we implemented a prototype that utilized

the concept and applied it in a case study. Within the

timeframe of three month we sought to foster

knowledge exchange among the knowledge workers

that were executing knowledge-intensive tasks in an

SAP system. More precisely, the 93 case study

participants were tasked to design a company‘s

organizational setup in the SAP system – a complex

system configuration task with multiple options and

the challenge to master the system and its

interactions in addition. We offered an interface in

the operational system that combined features of

expert recommendation systems and knowledge

networks where the overlap in task context served as

the glue (see Figure 2). The similarity of tasks was

determined in analogy to the discussion in chapter 4:

We analyzed the history of transactions, the SAP

concept of tasks, to find those individuals that have

common task contexts and find those that are,

relatively seen, experts whenever an individual seeks

support from within a certain task context.

Figure 2: Using task similarity to combine Expert

Recommendation and Knowledge networks.

The prototype worked as follows. Whenever an

individual wanted to interact with another to find

help for solving a challenging problem, he could,

directly in the SAP system, call up a program.

There, the individual would formulate a question

and send it out without specifying recipients. In the

background, the program then distributes the request

to the “right” individuals, based on the current

context of the requester and the features, i.e. context

and level of expertise, of the receivers. Only those

that are suitable (see previous chapter for filtering

options) were informed that their expertise was

being asked for along with the message itself. The

response of the knowledge bearers was also

automatically distributed, so that the individual

could focus on response formulation knowing that

the requester was in a similar context.

We implemented additional functionality into the

prototype, e.g. a forum to collect past interactions

with task-oriented structuring, facilitated message

creation, notification systems for possibly relevant

messages and other features. In this contribution we

only focus on and describe those that deal with the

core concept of using histories of tasks for

determining levels of expertise, contextual overlap

Expert

Recommender

Knowledge

Networks

Task Similarity

USING TASK HISTORIES TO SUPPORT PERSON-TO-PERSON KNOWLEDGE EXCHANGE - Extracting and using

Contextual Overlap and Levels of Expertise to Connect Knowledge Workers

and their perceived value for the participants.

7 EVALUATION

After the three month case study, we surveyed the

participants to find out about their perceived value

of the concept we applied. Among other items that

related to the additional features of the prototype, we

included a number of items that asked for the

participants’ perception of contextual overlap and

level of expertise. We received 18 fully filled

surveys. This corresponds to a return rate of roughly

20 % - a normal value for online surveys.

Figure 3: Influence of being seen as expert on contribution

behavior.

The socio-cognitive model indicated that reputation

was a strong motivator to respond to requests for

help. We utilized the concept of level of expertise as

one filtering dimension. Also the prototype indicated

to the receiver of requests that he was determined as

expert and therefore received the request.

Consequently, we wanted to find out how being seen

as expert influences the individual’s motivation. The

results as shown in Figure 3 indicate, that for about

28 % of the respondents being seen as expert

increases the likeliness to contribute, while for 22 %

it does not and 50 % were undecided on the effect.

Apparently, gain in reputation does not motivate all

participants likewise.

We also wanted to see what the influence of the

second dimension we used, contextual overlap of

requester and responder, would be on the

contribution behavior. Figure 4 illustrates that

roughly four out of ten survey participants felt that

they would respond more often if the work contexts

match. Interestingly, this is a higher value than for

being seen as expert and suggests that contextual

overlap is more important for the potential responder

than the level of expertise.

Figure 4: Influence of context match on contribution

behavior.

In our argumentation in chapter four we further

argued, that a shared context not only increases the

willingness of individuals to respond to requests but

also that the shared context facilitates the

interaction. We especially argued that the shared

context allows relying on shared terminology that

facilitates the formulation of messages.

Figure 5: Facilitation effect of similar context on request

formulation.

In our survey we also asked whether it is easier for

the knowledge seeker to formulate requests for help,

when knowing that the potential receivers will be in

a similar context. As illustrated in Figure 5, 11 % of

the survey respondents strongly agreed that it is

easier to formulate requests in this case and another

22%

50%

11%

Contribute more when seen as

expert

applies fully

applies

undecided

does not

apply

does not

apply at all

N = 18

33%

61%

Respond more often if requests

match context

applies fully

applies

undecided

does not

apply

does not

apply at all

N = 18

11%

33%

44%

Formulation easier if receiver in

same context

applies fully

applies

undecided

does not

apply

does not

apply at all

N = 18

KMIS 2011 - International Conference on Knowledge Management and Information Sharing

33 % agreed, while only 6 % did not or strongly not

think so, respectively. This indicates that for a large

portion of participants, knowing that the receiver is

in a similar work context helps them in

communicating their request and starting an

interaction with them.

Similarly, we wanted to determine the possible

facilitation effect on the responders’ side. Figure 6

shows whether the survey participants thought that

knowing about the similar context of the initial

requester would help them in formulating answers to

the knowledge request. While half of the responses

indicated that the participants were undecided

whether or not this knowledge would help them in

formulating answers, 28 % agreed it would and

17 % strongly agreed. With nearly half of the

respondents indicating the value of knowing the

contextual circumstances of the receiver as being

high, this appears to be a relevant feature.

Figure 6: Facilitation effect of similar context on answer

formulation.

Independent of the facilitation effect and the

increased willingness to interact, we also asked the

survey participants whether they perceived the

overlap of work context with their interaction

partners as generally valuable. Figure 7 shows that

while 50 % were undecided whether or not they

perceived this match to be important, 33 % did and

11 % did so strongly yet only 6 % indicated that they

did not. The inclusion of context overlap in the

mediation mechanism therefore seems to be a

suitable design choice.

Figure 7: General importance of contextual match.

8 CONCLUSIONS

In this contribution we addressed the challenge of

connecting knowledge workers to foster knowledge

exchange. Starting with three theoretical models that

describe the antecedents and process of knowledge

exchange we determine those constructs that apply

for our setting: distributed, non-acquainted

knowledge workers that interact across temporal,

physical and organizational borders. Subsequently,

we looked at state-of-the-art approaches that support

this setting and contrasted those approaches with the

constructs of the theoretical models to find that the

contemporary approaches do not support all

constructs. We hence suggest using a different

concept that relies on the history of tasks at its core.

Using this concept we describe how it can be used to

determine willingness and ability to support fellow

knowledge worker. Then, we described how we

implemented the concept in a case study with 93

individuals and describe the results we could obtain

by surveying the individuals after the three month

case study. We found that contextual match is at

least as important for the participants as the level of

expertise when interacting with other knowledge

workers. Also the survey results support our

expectation that formulation of messages among

knowledge workers is facilitated by contextual

overlap. Additionally, the participants found it

important to have a work context match with their

unknown interaction partners and indicated that

knowing that there is a contextual overlap motivates

them to contribute more. Our results appear

promising, but may be substantiated by replicating

the case study setting with more individuals and by

17%

28%

50%

Formulation easier if requester

in same context

applies fully

applies

undecided

does not

apply

does not

apply at all

N = 18

11%

33%

50%

Contextual match with receiver

is important

applies fully

applies

undecided

does not

apply

does not

apply at all

N = 18

USING TASK HISTORIES TO SUPPORT PERSON-TO-PERSON KNOWLEDGE EXCHANGE - Extracting and using

Contextual Overlap and Levels of Expertise to Connect Knowledge Workers

using different host systems for embedding new

prototypes and organizational settings to better

understand its limitations and benefits.

REFERENCES

Binney, D. (2001): The knowledge management spectrum

- understanding the KM landscape. In Journal of

Knowledge Management 5 (1), pp. 33–42.

Böhmann, T.; Krcmar, H. (2002): Werkzeuge für das

Wissensmanagement. In M. Bellmann, H. Krcmar, T.

Sommerlatte, Wissensmanagement, Symposion

Publishing GmbH, Duesseldorf.

Bromme, Rianer; Jucks, Regina; Rambow, Riklef (2004):

Experten-Laien-Kommunikation im Wissens-

management. In Gabi Reinmann, Heinz Mandl (Eds.):

Psychologie des Wissensmanagements. Perspektiven,

Theorien und Methoden. 1. ed. Göttingen: Hogrefe,

pp. 176–188.

Brown, J. S; Duguid, P. (2001): Knowledge and

organization: A social-practice perspective. In

Organization science 12 (2), pp. 198–213.

Byrne, D.E (1971): The attraction paradigm: Academic

Press.

Constant, D.; Sproull, L.; Kiesler, S. (1996): The kindness

of strangers: The usefulness of electronic weak ties for

technical advice. In Organization science 7 (2), pp.

119–135.

Davenport, T. H. (2011): Rethinking knowledge work: A

strategic approach. In McKinsey Quarterly. McKinsey

& Company

Drucker, Peter (1988): The coming of the new

organization. In Harvard business review.

Festinger, L. (1954): A theory of social comparison

processes. In Human Relations, 7(2) 117-140. 7 (2),

pp. 117–140.

Gronau, N.; Uslar, M. (2004): Integrating Knowledge

Management and Human Resources Via Skill

Management. In Journal of Universal Computer

Science: Proceedings of I-Know ‘04, pp. 135–142.

Holsapple, C. W; Whinston, A. B (1987): Knowledge-

based organizations. In The Information Society 5 (2),

pp. 77–90.

Kanfer, R. (1990): Motivation theory and industrial and

organizational psychology.

Krcmar, H. (2010): Informationsmanamgent, Berlin:

Springer.

Law, S. P. M.; Chang, M. K. (2008): Fostering knowledge

exchange in online communities: a social capital

building approach. In Proceedings of ICIS. Paris,

France, 14.-17.12.2008, pp. Paper 173.

Lin, N. (2001): Building a network theory of social

capital. In Social capital: Theory and research, pp. 3–

29.

Makolm, J.; Weiß, S.; Reisinger, D. (2007): Proactive

knowledge management: the DYONIPOS research

and use-case project. In Proceedings of 1st

international conference on theory and practice of

elecontric governance. ICEGOV. New York, NY,

USA: ACM.

Nahapiet, J.; Ghoshal, S. (1998): Social capital,

intellectual capital, and the organizational advantage.

In Academy of management review 23 (2), pp. 242–

266.

Newell, A.; Simon, H.A. (1972): Human problem solving:

Prentice-Hall Englewood Cliffs, NJ (104).

Nonaka, I. (1994): A dynamic theory of organizational

knowledge creation. In Organization science 5 (1), pp.

14–37.

Olivera, F.; Goodman, P.S; Tan, S.S.L (2008):

Contribution behaviors in distributed environments. In

MIS Quarterly 32 (1), pp. 23–42.

Quinn, James Brian (1992): Intelligent enterprise. A

knowledge and service based paradigm for industry.

New York: Free Press.

Qureshi, Sajda; Briggs, Robert; Hlupic, Vlatka (2006):

Value Creation from Intellectual Capital: Convergence

of Knowledge Management and Collaboration in the

Intellectual Bandwidth Model. In Group Decision and

Negotiation 15 (3), pp. 197–220.

Seid, Dawit Yimam; Kobsa, Alfred (2003): Expert finding

systems for organizations: Problem and domain

analysis and the DEMOIR approach. In Journal of

Organizational Computing and Electronic Commerce.

Serdyukov, P. (2009): Search for expertise. Going beyond

direct evidence. SIKS dissertation series.

Tajfel, H. T (1986): J. C. (1986). The social identity

theory of intergroup behavior. In Psychology of

intergroup relations, pp. 7–24.

Teigland, R. (2003): Knowledge networking: Structure

and performance in networks of practice: Stockholm

School of Economics.

Turner, J. C.; Hogg, M. A.; Oakes, P. J.; Reicher, S. D.;

Wetherell, M. S (1987): Rediscovering the social

group: A self-categorization theory: Basil Blackwell.

Wasko, M. M.; Faraj, S. (2005): Why should I share?

Examining social capital and knowledge contribution

in electronic networks of practice. In MIS Quarterly

29 (1), pp. 35–57.

Wasko, M. M; Faraj, S. (2000): "It is what one does": why

people participate and help others in electronic

communities of practice. In Journal of Strategic

Information Systems 9 (2-3), pp. 155–173.

Wegner, D. M.; Giuliano, T.; Hertel, P. (1985): Cognitive

interdependence in close relationships. In William

John Ickes (Ed.): Compatible and incompatible

relationships. New York, NY, pp. 253–276.

Ye, Yunwen; Nakakoji, Kumiyo; Yamamoto, Yasuhiro

(2008): Understanding and Improving Collective

Attention Economy for Expertise Sharing. In Zohra

Bellahsène, Michel Léonard (Eds.): Proceedings of 20

th CAISE,. Berlin: Springer. pp. 167–181.

Ziegler, Cai-Nicolas; Golbeck, Jennifer (2007):

Investigating Interactions of Trust and Interest

Similarity. In Decision Support Systems 43 (2), pp.

460-475.

Zimbardo, P. G. (1983):

Psychologie. Berlin: Springer.

KMIS 2011 - International Conference on Knowledge Management and Information Sharing