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.
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..
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.
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
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.
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
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
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.
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
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.
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
Contextual overlap
Level of expertise
Minimal context
Minimal expertise
Minimal combined
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.
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
Task Similarity
Contextual Overlap and Levels of Expertise to Connect Knowledge Workers
and their perceived value for the participants.
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
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
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
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
Contribute more when seen as
applies fully
does not
does not
apply at all
N = 18
Respond more often if requests
match context
applies fully
does not
does not
apply at all
N = 18
Formulation easier if receiver in
same context
applies fully
does not
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
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.
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
Formulation easier if requester
in same context
applies fully
does not
does not
apply at all
N = 18
Contextual match with receiver
is important
applies fully
does not
does not
apply at all
N = 18
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.
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
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.
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:
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–
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,
Nahapiet, J.; Ghoshal, S. (1998): Social capital,
intellectual capital, and the organizational advantage.
In Academy of management review 23 (2), pp. 242–
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.
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.
Zimbardo, P. G. (1983):
Psychologie. Berlin: Springer.
KMIS 2011 - International Conference on Knowledge Management and Information Sharing