=Paper=
{{Paper
|id=None
|storemode=property
|title=Context as a Tool for Organizing and Sharing Knowledge
|pdfUrl=https://ceur-ws.org/Vol-626/regular1.pdf
|volume=Vol-626
|dblpUrl=https://dblp.org/rec/conf/ekaw/WarrenGRDTD10
}}
==Context as a Tool for Organizing and Sharing Knowledge==
https://ceur-ws.org/Vol-626/regular1.pdf
Context as a tool for organizing and sharing knowledge
Paul Warren1, Ian Thurlow2, John Davies2, Igor Dolinsek3, Carlos Ruiz4, Jose-
Manuel Gomez-Perez4
1
Eurescom GmbH Wieblinger Weg 19/4, D-69121 Heidelberg, Germany
BT Innovation & Development, Adastral Park, Martlesham Heath, Ipswich IP5 3RE, U.K.
2
3
Hermes SoftLab d.o.o., Litijska 51, 1000 Ljubljana, Slovenia
4
iSOCO, Pedro de Valdivia, 10, Madrid, Spain
warren@eurescom.eu, {ian.thurlow, john.nj.davies}@bt.com, igor.dolinsek@comtrade.com,
{cruiz,jmgomez}@isoco.com
Abstract. This paper describes how tools and technologies have been
developed within the ACTIVE project to enable user context to be employed to
organize and share knowledge. The approach taken is a combination of top-
down, where the user explicitly creates and manages context on the system, and
bottom-up, where machine intelligence techniques support the user. The work
of ACTIVE is being validated in three case studies, and the paper describes in
particular how the approach to context is being validated in a case study in BT.
Keywords: context, information overload, knowledge sharing, machine
learning, context detection, context discovery
1 Introduction
Information overload is a significant problem for all who perform knowledge work
and the use of context is seen as an important technique for information delivery [1].
This paper describes work within the ACTIVE project (http://www.project-active.eu)
which is using context both to combat information overload and to facilitate
knowledge sharing. The paper describes the tools being developed and how they are
being validated in a case study in BT.
ACTIVE is a European integrating project. Fundamental to the project are three
interrelated research themes:
Facilitating knowledge sharing through the synergy of a Web 2.0 approach and a
more formal ontological approach.
The provision of tools and technologies for describing, sharing, learning and
optimizing informal, working processes.
The use of context for combating information overload and for knowledge
sharing.
This paper concentrates on the last of these themes. However, before discussing
ACTIVE’s work with context in detail, section 2 provides some background
information about the project. Section 3 then talks specifically about the work in the
project relating to context, whilst section 4 describes the validation of this work in the
BT case study. Section 5 discusses related work and section 6 summarizes some
conclusions and discusses next steps.
2 The ACTIVE project
2.1 The ACTIVE research themes
Knowledge sharing – a synergy of formal and informal approaches
The first of these themes involves two complementary approaches. The Semantic
MediaWiki (http://semantic-mediawiki.org/wiki/Semantic_MediaWiki) [2] is being
extended on the basis of requirements elucidated from the project’s case studies. The
Semantic MediaWiki uses an informal approach to knowledge representation. For
example, users are free to create their own relations rather than select from a pre-
defined ontology. Where different relations are used to express the same semantics,
these can be equated. The result is an approach to ontology and knowledge creation
which offers less of a barrier to entry than a more formal approach. Whilst restricted
in power it offers most or all of what is required in a knowledge management
application. The second approach to knowledge sharing also emphasizes ease of use
through the use of tagging to describe information objects. Here again, the simplicity
of the approach arises from the fact that the user is not restricted to a pre-defined
vocabulary. The user can choose whatever tags he or she likes. In addition, we are
using machine intelligence techniques to suggest tags; the suggestions can be
accepted or rejected. The suggestions are based on an analysis of file contents and
also on tags used for similar files. Similarity is computed using a technique based on
cosine similarity [3].
Informal processes – describing, sharing, learning and optimizing
Work on informal processes is motivated by the insight that, besides a relatively
small number of well-defined business processes which every organization has, we all
create our own informal processes to achieve our everyday working goals [4].
Because these processes are rarely written down they are not easily shared between
colleagues. Hence they are continually being reinvented and not subject to the
informal peer-review which would lead to their improvement. In ACTIVE we
combine top-down and bottom-up approaches to the management of informal
processes. A top-down approach means that users are given tools to describe these
informal processes. In the bottom-up approach, technology is being developed to
monitor user actions and learn the frequently-used processes. Once learned, they can
be shared and compared with what is thought to be the ideal. Technology is also
being developed to optimize these processes.
Context – for combating information overload and for knowledge sharing
Our work on context also combines a top-down and a bottom-up approach. On the
one hand, users are free to define their contexts and to specify their current context,
that is the context in which they are currently working. On the other hand, machine
intelligence techniques are available to discover the user’s contexts from his or her
activities and to detect the user’s current context. The development of this approach,
and its validation in the BT case study, is the theme of this paper and more
information is provided in sections 3 and 4.
Work on each of these themes does not take place in isolation. ACTIVE is an
integrating project, and the three themes all interact. This is achieved through the
ACTIVE knowledge workspace (AKWS). The AKWS integrates the tools developed
in ACTIVE, and ensures interoperability with an extensively enhanced Semantic
MediaWiki and extended Microsoft Office tools.
2.2 ACTIVE architecture
ACTIVE uses a service-oriented approach. ACTIVE services perform user,
information, context and process management as well as process mining and metadata
recommendation. Each organization deploys the ACTIVE services on a set of their
integrated servers or places them in a cloud. The complete functionality of ACTIVE
services is accessible through a rich set of web service interfaces and through the
ACTIVE event bus. End-users access the ACTIVE services through the ACTIVE
web portal; through a dedicated ACTIVE taskbar; and through ACTIVE-extended
Microsoft Office tools, and Windows File Explorer.
2.3 ACTIVE case studies
Three case studies are being used within ACTIVE to validate the research and
development activity. In two of these, with Accenture and BT, we are using
ACTIVE technology to support generic knowledge work. In the other, with Cadence
Design Systems, the technology is being used to support the specific activity of
electronic circuit design.
Accenture (http://www.accenture.com) is a global management consulting,
technology services and outsourcing company, with more than 186,000 people
serving clients in over 120 countries. Accenture consultants heavily rely on
knowledge sharing and reuse to accomplish various tasks, for example the creation of
customer proposals. This case study is making use of context, e.g. to present and
prioritize search results.
The BT case study is similar to that of Accenture in being focused on people who
work very much with customers and perform tasks such as customer proposal writing.
However, whereas in the Accenture case study, ACTIVE technology is being used to
enhance a pre-existing knowledge management system, in BT ACTIVE is deploying
tools and user interfaces developed within the project. More detail about the BT case
study is provided in section 4.
In Cadence, as with Accenture, ACTIVE technology is being used to enhance an
existing suite of tools. As already noted, the goal is to support the electronic design
process. The complexity of the design processes is such that it is often not obvious to
an inexperienced designer what is the next appropriate step to take. This will often
depend, e.g. on the type of integrated circuit being designed, and hence this kind of
knowledge develops with experience. A goal of ACTIVE is to capture this tacit
knowledge so that it can be shared with less experienced designers. The complexity
of the process is also such that it is difficult to understand a designer’s current
position in a design process, i.e. how far through the current process he or she has got.
This makes management of the process difficult. ACTIVE’s process learning
technology is being used to identify the designer’s current position in the overall
process and thereby make such design processes more manageable.
2.4 Validating the benefits
ACTIVE has a number of activities with the goal of validating the project benefits.
Usability evaluation is being performed to ensure that the tools developed are right for
their users. For the BT case study, this is discussed in more detail in section 4.
The business benefits of our approach are being evaluated within each of the case
studies. This is being done by particular reference to the key goals of each case study.
For example, in the Accenture and BT case studies a key goal is to develop higher
quality customer proposals. To gauge the impact of our approach, customer proposals
can be evaluated along a number of dimensions, e.g. the extent to which they address
the customer’s stated requirements, and internal consistency. Work is also being
undertaken to understand the costs and benefits of creating and using lightweight
ontologies, as in the Semantic MediaWiki, and folksonomies.
3 Context in ACTIVE
3.1 Using context
In ACTIVE we are concerned with using context to combat information overload and
to facilitate knowledge sharing. At any given time, a user needs information relating
to the context of his or her current work. A key, therefore, to combating information
overload is to prioritize that information over information which relates to other
contexts.
To take some concrete examples, we imagine that for a customer-facing person the
set of contexts will include all the customers, or proposals, with which he is currently
working, plus perhaps a few additional contexts such as ‘admin’ and ‘personal’. For a
lawyer, his or her contexts are likely to be the particular cases currently being worked
on. However, we are not obliging our users to adopt any particular set of contexts.
We are providing them with a toolset which they can adopt to their particular way of
working.
What ‘prioritized’ means will depend on the particular application. For search it
means that search results related to the user’s current context appear higher up the
search list than unrelated results. When opening a file within an application such as
word-processing, it means that instead of seeing the most recently opened files, the
user can opt to see the most recently opened files related to the current context. This
is illustrated in figure 1, where the ‘open from context’ feature is being used, as
shown highlighted.
We also believe that the use of shared contexts helps the sharing of information, as
for any given context, a user can see what information his or her colleagues have
available.
Note that ACTIVE enhances the user’s existing applications, rather than replacing
them. In the BT case study described in section 4, the applications used are Microsoft
Outlook, Word, Excel, PowerPoint, Internet Explorer and Windows File Explorer.
Moreover, ACTIVE adds to existing functionality, it does not detract from it. So, in
the word-processing example, if the user wants to see all his recently opened files,
rather than just the ones in context, he is free to do so. Referring to figure 1, the user
would simply click on the ‘open’ command further up the menu.
Fig. 1. A user in Word opting to select a file from those associated with his current context
In ACTIVE the relationship between a specific process and a specific context is for
the user to determine. Processes and contexts can be independent. For example, in
the BT case study a typical user might create a number of contexts corresponding to
the customer with whom he or she regularly deals. A customer proposal process
might then be created and be applicable to each of these customer contexts. In
addition, a process can be assigned to a context, just like any other information object.
So a user might have a context concerned with the financial outgoings incurred in his
work, and a process for creating a travel expenses claim.
3.2 Top-down – creating and using context explicitly
Our starting point was to provide users with the ability to:
create context
assign themselves to contexts, and de-assign from contexts
associate information objects with contexts
set the current context
Note that the need for the user to assign him or herself to a context, rather than the
context being automatically assigned when it is created, arises for two reasons.
Firstly, contexts can be shared. So after one user has created a context, a second user
may wish to assign him or herself to that context. Moreover, a user may cease to use
a context for a period of time, but not want to delete it completely, because the
context may be useful in the future. If the user de-assigns the context, he is free to re-
assign it later.
Figure 2 shows the user creating a new context, called ‘Search’. Contexts can be
hierarchical; in this case the user is specifying that ‘Search’ is a sub-context of ‘BT
projects’. In the menu on the left there are a number of additional command options,
including the option for assigning contexts to other users. The user can also display
the currently assigned contexts, as show in Figure 3. As can be seen, a user can also
be de-assigned from a context. In another option, a user can view all the contexts in
the workspace.
Fig. 2. Creating a context
Fig. 3. Displaying assigned contexts
Associating an information object with a context is done via the ‘associate’ feature
which is implemented in all the relevant applications. When the user clicks on the
associate button, then the opened file, webpage or email is associated with the current
context. Figure 4 illustrates this for Microsoft PowerPoint; in the figure the associate
button is at the top-right, within the ‘ACTIVE toolbar’. Figure 5 illustrates a number
of actions which can be applied to arbitrary files and directories in the extended
Windows File Explorer. Files which are inserted into the workspace are marked with
a small ACTIVE icon and the file manipulation menu is extended with ACTIVE
actions, e.g. context association. Note that in figures 4 and 5, as well as an associate
button, there is also a tag button to implement tagging. Clicking on ‘tag’ gives the
user the opportunity to create his own tags or to accept system-suggested tags, as
mentioned in section 2.1.
Users can also configure their profiles such that, whenever an information object is
opened, then an association is automatically made between that object and the current
context.
Fig. 4. Illustrating ‘associate’ in Microsoft PowerPoint
Fig. 5. Illustrating actions on a file in Windows File Explorer
The user switches between contexts by using the ACTIVE taskbar, shown at the
top of Figure 6. On the left of the bar the current context (BT) is shown highlighted.
Other contexts are shown in the drop-down menu; by clicking on one of these the user
switches between contexts. The drop-down menu shows a flat list of contexts. As
already noted in section 3.2, it is also possible for a user to create hierarchical
contexts, so that one context is a sub-context of another. The figure illustrates a
number of other taskbar functions. For example, the icon displaying two faces is used
to show all team members assigned to the user’s current context, whilst the briefcase
icon is used to display all information objects associated with that context. In
addition, the Context Visualizer, as described in section 3.4, can be invoked from the
ACTIVE taskbar.
Fig. 6. The ACTIVE taskbar, showing the current context and alternative contexts
In summary, our top-down approach to context management assumes that users
will create a set of contexts appropriate to their work; that they will associate
information objects and informal processes with those contexts; that they will assign
themselves not just to contexts they have created themselves, but to contexts created
by others, thus forming shared contexts; and that they will explicitly set their current
context in the taskbar. We believe that some users will be happy to follow this
approach and will find it beneficial. The extent to which this hypothesis is true is a
matter for validation, see section 4 below. We also take the view that many more
users will benefit from a context-based approach to information management if there
is a degree of automation, e.g. the automatic provision of suggested contexts. This is
the subject of section 3.3.
3.3 Bottom-up – letting machine intelligence help
In ACTIVE we use machine intelligence:
to detect the current context, i.e. based on the user’s activities to determine what
his current work focus is;
to discover new contexts, i.e. to cluster the user’s information objects in a way
which reflects the contexts in which the user works.
As the user is working, the ACTIVE system analyses the documents, spreadsheets,
emails, web-pages etc. which the user accesses and, based on this analysis, estimates
what is the user’s current working context. When a change in the user’s context is
detected with a sufficient degree of confidence, then one of two things happens.
Depending on the user’s preference settings, either the system switches context
automatically or the user is asked to approve a switch in context.
Context discovery can be used to augment top-down user created contexts, or it can
be used as the sole means of creating contexts. As with context detection, user
behaviour and information objects are continually analysed and potential new
contexts are identified. When the system has a certain degree of confidence that a
new context has been discovered, then the user is alerted. In general, of course, this
would happen much less frequently than the detection of a change of context. The
user is provided with a list of the information objects associated with this new
context, and is free to name the context and adopt it as one of his set of contexts, or to
simply reject it.
The approach taken for context discovery is described in another paper in this
workshop [5].
3.4 The context visualizer
A key goal of ACTIVE is to help knowledge workers manage and understand the
context of their daily collaborative processes. The articulation of context using
visualization tools supports this goal by helping understand the relationships within
collaborative processes.
Context visualization tools need to explicitly articulate knowledge structures and
their relationships for a particular given context. They also need to reduce the
inherent complexity, facilitating the understanding of complex relationships within a
context. They need to combine multiple modalities, e.g. graphs and text, and to
provide filter options.
The ACTIVE Context Visualizer is a graphical visualization tool that explicitly
articulates knowledge structures and their relationships for a particular given context.
This allows the user to explore and browse the context and, for example, see how it is
comprised of resources and how these resources are related and influenced by each
other. Moreover, it offers other utility features such as zoom and diverse types of
filtering. Figure 7 shows a screenshot of the tool.
The Context Visualizer contains two main parts. On the right, the Diagram Pane
contains the graphical representation of the context. On the left, the Information Pane
contains information about the context being viewed and the file, user or task
currently selected.
Figure 7 The context visualizer
The Diagram Pane shows the context graph and has three levels: the first level is
the actual context, shown in the centre of figure 7; the second level comprises the
categories of objects involved in the context, i.e. processes, users and files; the third
and final level includes the particular instances of processes, users, and files. Each
node is labelled with its name to allow easier exploration, and each node image is
resized using a weight attribute depending on its importance within the context so the
user can more easily locate the most relevant content. The currently selected file, user
or task is shown in bold (‘Proposal Summary’ in the figure). The currently selected
object is also shown interconnected, in red, with other resources which have a related
use. Thus, in the figure, ‘Proposal Summary’ is interconnected with ‘Brandon L
Harvey’ by red lines, implying the file has been created or accessed by Brandon L
Harvey.
The Diagram Pane also contains the Filter Pane, which is used to filter files by type
or nodes by their weight. In contexts with several hundred nodes, the filter helps the
user to focus on the most important information by filtering by weight. As we expect
the real-life contexts to have a large number of files, it will also be relevant to be able
to filter these by their type, furthermore reducing the complexity of the visualized
context and facilitating easier browsing and location of interesting information.
The Information Pane, on the left, is comprised of a combo box “Context
selection” to select the context to browse and two information grids: “Context
information” and “Selected node information”. The former displays information
about the context being viewed. The latter has detailed information on the file, user
or task currently selected in the context graph.
The components of the Context Visualizer architecture are:
• The Context Mapper receives the incoming context from other AKWS
services and maps it to the internally used Context Representation Model.
• The Context Representation Model is the underlying model used to represent
the context notion in ACTIVE [6]. It is based on PIMO [7] (Personal Information
Model Ontology) which uses RDF and NRL (Nepomuk Representational Language)
to express personal information resources
• The Context Presentation Engine defines a graph-based model for
representing contexts based on JPowerGraph [8]. Such an engine enables flexible
configuration.
4 Validation in the BT case study
The BT case study is validating a number of the tools being developed in ACTIVE;
including the use of context to combat information overload and share knowledge.
The trial is taking place within BT Business, a division of BT which works with small
and medium-sized businesses across the U.K. The trialists themselves are ICT
specialists, technical consultants, solutions consultants and sales managers. Whilst
they represent a variety of roles, they are all customer-facing and our expectation was
that their particular contexts would relate to the customers with whom they are
dealing.
During the first year of the project an exercise was undertaken to understand their
problems in handling information and how ACTIVE could help. This took a number
of forms. Some potential trialists were interviewed. For a few we undertook job-
shadowing, i.e. we observed the user working. To save travelling time and costs, and
also to minimize disturbance to the potential trialists, this was done remotely.
Microsoft LiveMeeting was used to view the subject’s screen, and the voice channel
of an instant messaging service was used to monitor conversations.
In parallel with this exercise we were developing the key ACTIVE technologies
and understanding how those technologies could be applied. This enabled us to create
a number of ‘mock-ups’ of ACTIVE functionality which were used at the beginning
of the second year of ACTIVE to get more detailed feedback from potential users.
Feedback received during this phase suggested that the relevancy of context to a
person’s work is dependent on their role. For some participants the ability to switch
between different contexts was particularly relevant. This was true, for example, of
the ICT specialists who were regularly changing their focus of work. For others, in
particular where work on a particular activity tended to be less fragmented, the
context features of ACTIVE were not considered as relevant.
The exercise identified a number of requirements, some of which related to the
theme of context:
The effect of context switching on the behaviour of applications needs to be
capable of being overwritten, i.e. one can always revert to the original form of
information presentation. This had already been considered. For example, as
explained in section 3.1, when using Word, the user can look at the most recently
opened files in the particular context, or simply the most recently opened files in
the usual way.
An information item should be able to be associated with more than one context.
This, also, is possible in the implemented system.
Where users work in many contexts, there is a need for automatic context
discovery and detection, as described in section 3.3.
On context switch, applications which were previously open in the new context
but had subsequently been closed, should be reopened. Conversely, applications
which were previously closed and subsequently opened, should be closed. This
feature can be implemented from the technical perspective, but it would require
substantial development efforts. As noted in section 5, the ACTIVE approach is
information-centric. This requirements corresponds to an alternative, activity-
centric, approach.
As already explained, context is also seen as a way of enabling and encouraging
knowledge sharing. When asked about knowledge sharing, the issue of trust was
raised by participants. The majority of participants stated that they would be most
likely to share information with members of their immediate team. Conversely, they
would be most likely to make use of information which originated within their
immediate team. In practice, an ACTIVE knowledge workspace could be limited to a
small team, or extended to a department or even the whole organization, according to
the needs of the business environment.
At the time of writing the ACTIVE system is being deployed to members of the
BT Business trial community. During the closing months of the project an extensive
validation exercise will take place, starting with around 20 users and ramping up to
potentially 100. The overall goal will be to understand, and hence through feedback
to improve, the users’ perception of the system. All trialists are volunteers and are
aware of the nature of the trial. Steps are taken to ensure confidentiality. In
particular, the text of documents is not automatically stored on the ACTIVE server;
for the purposes of context detection and discovery, only the 'bag of words'
representation of the documents is required. Furthermore, the bag of words is
tokenised such that each word is represented by a token whose meaning is held in a
separate file. Moreover, users are free to specify that https web pages should be
excluded from the analysis. Users will be questioned about their experiences with the
system after a period of time using it. In addition, more immediate feedback on
ACTIVE functionality is being obtained. Use of ACTIVE features will be monitored,
and periodically the user will be requested to provide feedback relating to particular
system interactions. For example, when the system detects a potential context switch,
the user could from time to time be queried as to the appropriateness of the suggested
switch.
5 Related work
There has been a great deal of research in recent years in the area of context.
However, a glance at the proceedings of a conference devoted to the subject of
context, e.g. [9] will illustrate the breadth of topics covered. Discussions of context
include location-aware services, physical context and, closer to the interests of this
paper, the use of context to resolve ambiguities. For a brief overview of these topics,
and a list of some key recent papers, see [10]. The NeOn project (http://www.neon-
project.org) is also looking at an aspect of context. The goal of NeOn is to improve
“the capability to handle multiple networked ontologies that exist in a particular
context, are created collaboratively, and might be highly dynamic and constantly
evolving”. NeOn recognizes that information is processed differently by different
people and applications according to their context. Domain ontologies do not always
fully specify knowledge because that knowledge is implicit in the context in which
the ontologies are used. If the ontology is to be used in a different context, then this
knowledge does need to be made explicit. This requires that the ontology be
contextualized, i.e. that the relations be found with other ontologies which express
this context [11]. More closely related to ACTIVE is work within the APOSDLE
project (http://www.aposdle.tugraz.at/). APOSDLE provides support for e-learning
within the “the user’s immediate work environment and context”. To achieve this,
and to support user collaboration, it needs an understanding of user context. For more
information on the APOSDLE use of context see, e.g., [12] and [13]. Another
approach to context, also addressing the need for context switching, is described in
[14]. Here, the user is able to switch between tasks and reinstate a group of windows
associated with each task. This approach is activity-centric, in that it reinstates the
user’s total environment. Our approach is information-centric; more limited in that it
simply reinstates the information prioritization associated with a context; more
ambitious in the way it handles information objects and in the use of machine
intelligence. Finally, parallels can be seen between our concept of context and the
web application Google Wave (http://wave.google.com/about.html). In ACTIVE,
context can contain a range of different kinds of information objects and supports the
sharing of those objects. For Google Wave, each wave can be seen as analogous to a
context, also comprising different kinds of information objects.
6 Conclusions and next steps
We have described a system which permits both the top-down creation and
management of user contexts, and also the bottom-up discovery and detection of such
contexts. Our system extends the standard Microsoft applications so that individuals
can use their familiar applications whilst benefitting from the enhanced functionality.
We believe that, whilst the top-down approach is on its own beneficial to some users,
others will find real benefit in combining this with a bottom-up approach based on the
use of machine intelligence. This has been confirmed in discussions with potential
trialists.
As we have already indicated in section 4, during the final six months of the
project, there will be an extensive validation exercise in the BT case study. This will
include a detailed analysis of users’ reactions to our contextual approach to
knowledge management. Similar validation exercises will be undertaken in the other
two case studies. In the Accenture case study, context is also being used to enhance
knowledge worker productivity. The same underlying ACTIVE technology is being
used as in the BT case, but implemented through a different toolset. Comparison of
user reaction in the BT and Accenture case will further enhance our understanding of
how context can best be used to help knowledge workers achieve their goals.
Acknowledgements
The work described in this paper has been funded as part of the IST-2007-215040 EU
project, ACTIVE.
References
1. Driver, E., Moore, C., and Markham R., Context is king in the new world of work,
Forrester Research (2008)
2. Krötzsch, M, Vrandecic, D., and Völkel, M., Semantic MediaWiki, Proceedings of
ISWC2006, LNCS4273, edited Cruz et al., pp. 935-942, Springer-Verlag (2006)
3. Mitchell T M: ‘Machine Learning’, The McGraw-Hill Companies Inc
(1997).
4. Hill, C., Yates, R., Jones, C., and Kogan, S., Beyond predictable workflows:
enhancing productivity in an artful business process, IBM Systems Journal, vol. 45,
no.4, pp. 663-682 (2006)
5. Stajner, T., Mladenic, D., Groblenik, M. Exploring contexts and actions in
knowledge processes, Workshop on Context, Information and Ontologies
(CIAO2010) held in conjunction with EKAW 2010 (2010)
6. Mladenić, D., Getoor, L., Ferlež, J., Grobelnik, M., D2.3.1 Context definition and
representation within enterprise environments, ACTIVE project (IST-2007-215040)
http://www.active-
project.eu/fileadmin/public_documents/D231_context_definition_and_representation.
pdf (2009)
7. PIMO: Personal Information Model.
http://sourceforge.net/apps/trac/oscaf/wiki/PIMO
8. JPowerGraph. http://sourceforge.net/projects/jpowergraph /
9. Context 07, Sixth International and Interdisciplinary Conference on Modeling and
Using Context Roskilde University http://context-07.ruc.dk/ (2007)
10. Mehra, P., Context-Aware Computing: Beyond Search and Location-Based
Services, computing now, IEEE, August 2010,
http://www.computer.org/portal/web/computingnow/archive/august2010 (2010)
11. Le Duc, C., Matching ontologies for context: The NeOn Alignment plug-in, NeOn
deliverable D3.3.2, ftp://ftp.inrialpes.fr/pub/exmo/reports/neon-332.pdf (2006)
12. Musielak M., Hambach S., Christl C. APOSDLE Contextualized Cooperation. In:
ACM SIGCHI: ACM Conference on Computer Supported Cooperative Work
Electronic Proceedings: CSCW 08 [CD-ROM]. New York: ACM Press (2008)
13. Lokaiczyk, R.; Godehardt, E., Faatz, A., Goertz, M., Kienle A., Wessner, M.,
Ulbrich, A. Exploiting Context Information for Identification of Relevant Experts in
Collaborative Workplace-Embedded E-Learning Environments, EC-TEL 2007, Crete,
Greece (2007)
14. Robertson, G., Smith, G., Meyers, B., Baudisch, P., Czerwinski, M., Horvitz, E.,
Robbins, D., and Tan, D. Explorations in Task Management on the Desktop, in
Beyond the Desktop Metaphor, ed. Kaptelinin V. and Czerwinski, M., p.p. 101-138,
the MIT Press (2007)