=Paper=
{{Paper
|id=Vol-175/paper-20
|storemode=property
|title=Context as a Foundation for a Semantic Desktop
|pdfUrl=https://ceur-ws.org/Vol-175/23_heathmottadzbor_context_final.pdf
|volume=Vol-175
|dblpUrl=https://dblp.org/rec/conf/semweb/HeathMD05
}}
==Context as a Foundation for a Semantic Desktop==
https://ceur-ws.org/Vol-175/23_heathmottadzbor_context_final.pdf
Context as Foundation for a Semantic Desktop
Tom Heath, Enrico Motta, Martin Dzbor
Knowledge Media Institute, The Open University,
Walton Hall, Milton Keynes, MK7 6AA, United Kingdom
{t.heath, e.motta, m.dzbor}@open.ac.uk
Abstract. Adoption of semantic web technologies and principles presents an
opportunity to change the conceptual model of desktop computing. Moving
from a traditional position where the desktop is largely tied to a specific
computational device, a semantic desktop could exist as a broad, networked
space defined relative to the user. In this position paper we argue that personal,
computing, and knowledge contexts are the appropriate means by which to
define and shape the desktop space, and that collectively they provide the
foundation for novel functionality in a semantic desktop.
1 Introduction
The traditional conceptual model of desktop computing is no longer applicable. In a
heavily networked world, the distinction between a desktop as realised by an
operating system running on local hardware, and the web as something separate and
external, is false. What makes the desktop unique is not the location of execution of
its underlying code, but its role as a space that is largely under the control of one
individual; a personal domain that in some way reflects or represents their world.
Consequently, developers of a semantic desktop face a choice between simply
applying semantic web principles to a traditional desktop environment whilst
retaining the same conceptual model, or using the power of these technologies to
enable a more ecologically valid interaction paradigm that emphasizes the person, the
tasks they perform, and the context in which they do so, irrespective of where the
code is executed. Whilst both approaches may bring benefits, we advocate the second.
In this position paper we will discuss how that context may be defined, represented,
and used in pursuit of this aim.
Ubiquitous computing literature has largely defined context in terms of computing
resources, the user’s location, and the identity of people around them [1]. Our
conceptualization of a semantic desktop requires a broader view to be taken if this
greater ecological validity is to be achieved. In the following sections we will identify
three facets of context (personal, computing, and knowledge) and the interactions
between them, whilst also examining how each may be represented and utilised in a
semantic desktop, and what novel features this may provide.
2 Personal context
Reminiscent of a phenomenological view of context [2], personal context
encompasses properties of the individual as they exist in the world, which may
influence a task being performed but are not specific or unique to it. In [3] the authors
identify several such properties, including a person’s social context (social networks
they are part of), their preferences (values or opinions held), resources they have
available for performing a task (such as time or attention), and their location.
Inevitably these factors interact, such as one’s location in a busy office reducing the
attention available for performing a task.
In terms of semantic representation, social networks and nodes within them can be
represented with vocabularies such as FOAF [4] and SWAP Personal Information
Markup [5]. Representing the resources available for performing a task may be
informed by work in the field of instant messaging, regarding how to describe
presence and availability (e.g. online, low attention, do not disturb). Personal
preferences might be described using multiple domain-specific vocabularies (such as
the Vegetarian Ontology [6]) or through a generic ontology of preference which could
be applied to any domain. Vocabularies such as Basic Geo [7] can be used to express
latitude and longitude locations in a machine-readable way. However, to
meaningfully represent the individual’s location this will need to be complemented by
ontologies of place and space that describe locations in terms of their function,
significance, and interrelations.
At present these personal context factors, if represented or used at all within a
desktop environment, are done so on a per-application basis, adding unnecessary
cognitive overhead to the performance of tasks. For example, a user may have to
maintain separate contacts lists in their email and instant messaging applications, or
may have to state their location when searching the web for local restaurants and
again when booking a flight.
The ability to create personal context widgets to represent and manage this context
information and make it available across tasks is a key benefit of a semantic desktop
over a conventional one. We hypothesise that this provides a more realistic cognitive
model to the user, where the factors and entities that characterise their world are
captured in one place rather than distributed piecemeal or not represented at all. A
trend towards integrating applications that share a social component can be seen in
Chandler [8], and this should be welcomed as long as the focus remains on integrating
contextual information and providing services on top of this.
3 Computing context
In the ubiquitous computing literature, resources such as network connectivity [9],
applications available, and characteristics of the device being used are seen as key
aspects of context. These factors should be represented and utilised in a semantic
desktop as they may impact on how a task is performed, and they do reflect aspects of
the user’s world. This may be achieved using CC/PP profiles [10], which provide a
means to describe the capability of a device and preferences about how it is used, and
may be extensible to describe factors such as the network connectivity available at
any one time.
An ontology of application types that describes their capabilities in functional
terms (e.g. ImageManipulator, MessageHandler) could be populated with those
applications available to any particular computing environment (whether they are
implemented at the local machine or the webtop level).
However, our view emphasises the desktop as a reflection of the user’s world.
Consequently, the objects that populate that world and the tasks performed within it
are key, with the resources and applications available playing only a secondary, more
abstract role. Influenced by aspects of the Xerox ‘Star’ computing environment [11]
we believe that a semantic desktop can enable an interaction paradigm centred on
digital objects, their contexts, and associated tasks, in a way that a conventional
desktop cannot. By object context we mean factors such as the people or objects
depicted in a photograph or the occasion on which it was taken, the person who sent a
message, or the organisation who published a document.
Whilst some actions users may wish to perform would be generic to most objects
(such as view, share, edit), others may be determined by the object type (such as
cropping or resizing an image), or its associated contexts. Awareness of how context
manifests itself for different types of object would allow true context menus to be
implemented, providing access to functionality or services tailored to that specific
object. For example, such a menu for an audio file could provide access to other
tracks by the same artist, a discography, or current tour dates. Similarly, a context
menu for a message object might provide access to contact details for the sender, or
inferences about how best to contact them in response, based on their current
availability.
Ontological descriptions of the types of objects available in a semantic desktop, the
actions that may be performed on them, their associated contexts, and the capabilities
of applications available to the desktop could enable a semantic registry for the
desktop, with reasoning able to determine the best code to execute to perform a
particular action or task, rather than requiring an explicit decision from the user.
4 Knowledge context
Whilst a traditional filesystem is concerned with managing files that exist on a local
machine, the same limitations do not apply to a semantic desktop. Instead, a semantic
filesystem should focus on the management and application of knowledge that
supports user tasks wherever it resides, enabling a knowledge-oriented computing
environment that adheres to our model of the desktop as a representation one person’s
world.
Here semantic web technologies can bring great benefits over conventional
desktops, firstly through improved knowledge representation, secondly through the
ability to reference any resource, irrespective of whether it is a digital object itself or
simply a reference to an entity in the real world, and thirdly irrespective of whether it
is located locally or remotely.
However, if these abilities are to be maximally exploited in the support of tasks,
then we must be able to assess the provenance and validity of knowledge in the
system; the knowledge context. Just as conventional filesystems have metadata
recording when a file was created and last modified etc., so a semantic filesystem
should have knowledge metadata indicating the source of knowledge in the system, its
age, whether it has been validated or not, and inferences about its likely
trustworthiness. An ontology of knowledge characteristics could define the exact
nature of this metadata, populated over time as knowledge is added to the system and
evidence is accumulated to qualify, validate or contradict the assertions.
Awareness of knowledge context brings some novel features to a semantic desktop
which could support the tasks outlined in [3]. If locating a certain piece of knowledge
within a semantic desktop, the ability to prioritise results based on inferred
trustworthiness could help reduce cognitive overhead. Similarly the user may only
wish to monitor knowledge within the environment that is recent whilst ignoring older
stable items, and knowledge context can enable this. It also provides a foundation for
the evaluating task, which consists of “determining whether a particular piece of
information is true, or assessing a number of alternative options”.
5 Interactions between facets of context
Whilst the context facets discussed here have been treated separately there are
inevitable interactions between them, and a semantic desktop must facilitate these to
provide maximum benefit. For example, how well someone is known to us (personal
context, social factor) is likely to effect how much we trust knowledge they share with
us (knowledge context). Similarly, in a situation where someone has limited resources
for performing a task they may accept knowledge as a solution even if it is from a less
trusted source, simply to have reached some solution. These interactions can enable
novel features, as shown by the example above where an objects context menu could
provide access to the author’s contact details and, where permission exists, to their
personal context, such as current availability and inferences about how best to contact
them.
The extent of these interactions can be illustrated by the task of locating recipes for
a dinner party. In this case a semantic desktop could take into account the personal
context of the user by excluding certain cuisines they have indicated they don’t like,
and by prioritising results from the same sources as recipes that have a trusted
knowledge context. An object context could be provided by creating links between
the semantically annotated recipes and the necessary ingredients as listed in an online
shop.
6 Conclusions
A semantic desktop has the potential to introduce a new style of interaction in
personal computing that is not feasible with conventional technologies. In this
knowledge-based environment, the desktop is defined in relation to the user, not the
hardware, operating system, application, or protocol being used. However, parameters
are required to shape this semantic desktop to the individual, and in this paper we
have argued that context is an appropriate and powerful basis on which to do so.
Whilst each facet of context we have discussed enables different functionality,
interactions between them are central to the performance of tasks by the user. A
semantic desktop based on the foundation of these facets and their interactions
represents a novel and powerful interaction paradigm.
Acknowledgements
This research was partially supported by the Advanced Knowledge Technologies (AKT)
project. AKT is an Interdisciplinary Research Collaboration (IRC), which is sponsored by the
UK Engineering and Physical Sciences Research Council under grant number GR/N15764/01.
The AKT IRC comprises the Universities of Aberdeen, Edinburgh, Sheffield, Southampton and
the Open University.
The authors would also like to thank John Domingue and Marian Petre, conversations with
whom inspired some of the ideas in this paper; and two anonymous reviewers, whose
comments led to significant improvements.
References
1. Dey, A. K., Abowd, G. D., Salber, D.: A Conceptual Framework and a Toolkit for
Supporting Prototyping of Context-Aware Applications. Human-Computer Interaction 16
(2001) 97-166
2. Svanaes, D.: Context-Aware Technology: A Phenomenological Perspective. Human-
Computer Interaction 16 (2001) 379-400
3. Heath, T., Dzbor, M., Motta, E.: Supporting User Tasks and Context: Challenges for
Semantic Web Research. In: Proc. ESWC2005 Workshop on End-User Aspects of the
Semantic Web (UserSWeb) (2005)
4. Brickley, D., Miller, L.: FOAF Vocabulary Specification. http://xmlns.com/foaf/0.1/ (2005)
5. SWAP Personal Information Markup. http://www.w3.org/2000/10/swap/pim/doc (2003)
6. Golbeck, J., Parsia, B., Hendler, J.: Trust Networks on the Semantic Web. In: Proc.
Cooperative Intelligent Agents (2003)
7. Basic Geo (WGS84 lat/long) Vocabulary. http://www.w3.org/2003/01/geo/ (2003)
8. O. S. A. F.: What's Compelling About Chandler: A Current Perspective.
http://www.osafoundation.org/Chandler_Compelling_Vision.htm (2005)
9. Khedr, M., Karmouch, A.: Negotiating Context Information in Context-Aware Systems.
IEEE Data Engineering Bulletin (2004) 21-29
10. Composite Capability/Preference Profiles (CC/PP): Structure and Vocabularies 1.0.
http://www.w3.org/TR/2004/REC-CCPP-struct-vocab-20040115/ (2004)
11. Johnson, J., Roberts, T. L., Verplank, W., Smith, D. C., Irby, C., Beard, M., Mackey, K.:
The Xerox Star: A Retrospective. Computer 22 (1989) 11-26, 28-29