=Paper=
{{Paper
|id=Vol-197/paper-2
|storemode=property
|title=MyWorkPlace: Personalised information about a Ubiquitous Computing enabled building
|pdfUrl=https://ceur-ws.org/Vol-197/Paper2.pdf
|volume=Vol-197
}}
==MyWorkPlace: Personalised information about a Ubiquitous Computing enabled building==
https://ceur-ws.org/Vol-197/Paper2.pdf
ubiPCMM06:2nd International Workshop on Personalized Context Modeling and Management for UbiComp Applications
MyWorkPlace: Personalised information about a
Ubiquitous Computing enabled building
David Carmichael, Judy Kay, Bob Kummerfeld and William Niu
University of Sydney
School of Information Technology
NSW 2006 Australia
{dcarmich, judy, bob, niu}@it.usyd.edu.au
ABSTRACT pants would have hundreds or thousands of items and clearly
overwhelm the user.
This paper describes MyWorkPlace, which uses personalisa-
MyWorkPlace solves this problem by modelling users, places,
tion of automatically generated ontologies to provide users
devices, sensors, services and objects to provide personalised
with personalised information about new and invisible items
views of the items within a user's environment. An impor-
within a ubiquitous computing environment.
tant part of the system is the use of an automatically gener-
A key component of this system is the automatic gener-
ated ontology to assist with the selection of items to display
ation of the ontologies, and models used to drive it. This
for the user.
data is being gathered from a number of sources including:
The ability for ontologies to facilitate human-machine and
building maps, the build manual, sta directory, student
machines-machine communications has gained wide recogni-
timetables, the departmental calendar and room bookings.
tion in the development of the UbiComp. It has been used
We describe planned evaluation of the system in a deploy-
in middleware to facilitate context management and reason-
ment to a new building and its new inhabitants.
ing [3, 4, 5], and user modelling [6].
A novel aspect of our work is the use of ontological data
Keywords generated using dierent sources which have dierent levels
User modelling, Invisibility Problem, Automatically Gener- of reliability to personalise the information given to a user
ated Ontologies based on their context.
The eort in creating a comprehensive ontology is sub-
stantial. Partial or completely automated generation has
1. INTRODUCTION the possibility to greatly reduce this eort. Depending on
Ubiquitous computing aims to embed our everyday envi- the degree and type of automation, the reliability of the on-
ronment with devices, sensors and services in such a way tology can vary greatly. To cope with this, we are examining
that they are as unobtrusive as possible, to the point of be- multiple levels of ontologies.
coming invisible to common awareness [1]. When achieved Our automatically generated ontology in being built from
this invisibility creates its own problems. Users may be un- a number of sources including: building maps, the build
able to discover what services are available to them, what manual, sta directory, student timetables, the departmen-
sensors are detecting them or why the system has reacted tal calendars and room bookings.
in a particular way. We call this the Invisibility Problem [2] The rest of this paper is organised as follows: We rst
To motivate this paper we consider a real life example of describe some related work in Section 2 to set the scene.
the invisibility problem, a University Department moving Section 3 describes MyWorkPlace when used in the scenario
to a new building instrumented with a number of ubiqui- described of Fred. The methods we use for automatically
tous computing features. We examine the interactions and generating the ontologies are described in section 4. We
information needs of Fred, an academic. conclude with a discussion of our proposed evaluation and
Initially all facilities (ubiquitous computing and other- future work in section 5.
wise) within the building are unknown to Fred. He may
have an idea of some types of facilities which are available,
but is unlikely to know the full extent of the facilities avail-
2. RELATED WORK
Weiser predicted that ubiquitous computing would be-
able and details about them. A list of all the information
come a technology which disappeared and became invisi-
about the building, its sensors, services, devices and occu-
ble [1]. Others such as Heer and Khooshabeh have exam-
ined the nature of this invisibility [7]. They note that an
invisible interface does not imply literal physical invisibility.
Permission to make digital or hard copies of all or part of this work for Edwards notes some of the problems associated with Invis-
personal or classroom use is granted without fee provided that copies are ibility while examining the challenges of putting ubiquitous
not made or distributed for profit or commercial advantage and that copies computing into the home [8].
bear this notice and the full citation on the first page. To copy otherwise, to There has been some work which addresses the issue of in-
republish, to post on servers or to redistribute to lists, requires prior specific forming users of ubiquitous computing systems what devices
permission and/or a fee.
and services are available to them. The AFAIK system is
ubi-PCMM at UbiComp’06, September 2006, Orange County, CA, USA.
. a multimodal help system for an intelligent room [9]. Help
ubiPCMM06:2nd International Workshop on Personalized Context Modeling and Management for UbiComp Applications
content must be entered in XML, but the help system is those of the students he supervises. He knows approximately
not personalised to the user or their context. The Digis- where the the front counter is, but has not been there so is
cope [10] is a system for viewing attributes of objects within unaware of what facilities are available. He knows nothing of
an intelligent environment. It consists a large semitrans- the seminar room, sta common room, pervasive computing
parent display mounted on an movable arm. Information is laboratory, or undergraduate computer laboratories.
retrieved from a database about objects which are identi ed Figure 1 shows a screenshot from MyWorkPlace person-
using RFID and visual tagging. The NearMe system [11] alised for Fred, as it would be shown on a PDA while he is
provides users with a list of nearby devices by examining standing in the Foyer of the building.
the signatures of nearby wi access points, and making a
request to a server of all known nearby devices. This is dif-
ferent from our work as it does not seek to deliver the same
level of detail and is not personalised.
The CONON system [12] is an OWL encoded context on-
tology (CONON) for modelling context in pervasive com-
puting environments. Its context model is split into into an
upper ontology and other more speci c ontologies. The up-
per ontology describes high-level features of basic contextual
entities, of which, the most fundamental ones are location,
person, activity and computational entity. Then each sub-
domain has a more speci c ontology with additional details.
It is implemented using Jena2 Semantic Web Toolkit and
OWL-Lite. Reasoning is either: ontology reasoning using
description logic or user-de ned reasoning using rst-order
logic. COBRA-ONT [3] is an ontology used in the Context
Broker Architecture (CoBrA) to facilitate knowledge shar-
ing and context reasoning in ubiquitous computing. The
system tries to determine location and status of agents (hu-
man or software) with in it. COBRA-ONT is expressed in
OWL and models places, agents, and events. The ontology is
categorised into 4 themes: 1) physical places, 2) agents (hu- Figure 1: The view Fred is presented by MyWork-
mans and software agents), 3) location text of the agents, Place, as he is standing in the foyer (room 100), after
and 4) activity context of the agents.
Outside of Ubiquitous Computing there has been work on
inhabiting his new o ce building for a week.
extracting ontologies from existing text sources. To do this
The Status bar at the top tells Fred what the system be-
both concepts and relationships between them need to be
lieves his location and status is. In this case, his location
learned. ConceptNet [13] is a massive ontology of common-
is believed to be the Foyer, because the Mac address of his
sense knowledge. The concepts and relationships are ex-
Bluetooth mobile phone has been detected there. There is
tracted by processing the 70000 sentences of the Open Mind
a details button to allow him to scrutinise and correct the
Common Sense Project. The sentences are elicited from the
reasoning used for his location and status.
user in a semi-structured way in order to make the informa-
The content panel of the main screen consists of ve ex-
tion easier to extract. Khan and Luo [14] focus on concept
pandable headings. The headings are Devices at this loca-
learning from a text corpus. Concepts are placed in a
tion, Nearby Devices, Nearby Places, Services /Events, and
hierarchy. However, the type of relation between them is
People. Clicking a heading shows or hides the contents. The
ignored; that is, it is only possible to tell two concepts are
[Show all items] button displays all the items the user is
related, but not how they are related. Some projects, such
allowed to use. It also allows the user to see why an item
as MindNet [15], Mecureo [16] and Janninik and Wieder-
was included or excluded by MyWorkPlace.
hold's approach [17], focus on extracting relations between
The system must determine which of the myriad of de-
terms from dictionaries.
vices, sensors, places, services, events to display to Fred.
For each heading, it examines the evidence and places each
3. SYSTEM DESCRIPTION item into one of a number of relevance categories :
The MyWorkPlace system provides users with person-
alised views of places, sensors, devices, services, objects and
• Already knows - The user is believed to already know
about this, based on either user feedback, or observa-
people in their environment. This advances on our earlier
tions such as the use of a device, or being detected in
work, MyPlace, described in [2]. The key dierence from
a location.
earlier work is the inclusion of an automatically generated
ontology to assist with the selection of items to show the
• Needs to learn - The information is thought to be use-
user.
ful to the user and the user is believed not to already
We now return to the scenario from the Introduction for
know it. Whether information is useful to a user is
interactions of Fred, a member of sta interacting with My-
determined based on manually entered stereotypes, or
WorkPlace shortly after moving into the newly constructed
the generated ontologies.
School of IT smart building.
Fred is an academic who does not know very much about • Needs to know now - This is a special case of Needs
the building. He knows the location of his own o ce, and to learn where the item is believed to be important
ubiPCMM06:2nd International Workshop on Personalized Context Modeling and Management for UbiComp Applications
based some aspect of the users' current context. For each item is shown if the user hovers the mouse over it. An
example: If the user has stated they are on their way example explanation might be You are teaching Algorithms
to a seminar in a room they are believed not to know 101, this room is used for Algorithms 101, and you have not
the location of, then location of the the seminar room yet been detected there.
is very important.
• Not relevant (Neutral) - The information is about some-
4. ONTOLOGY GENERATION
thing for which there is no information suggesting that The data in our ontology is being built from a number
it is useful to the user. of sources, such as building plans, sta directory, the build-
ing manual, student timetables, the departmental calendar,
• Doesn't want to know - The user has indicated that room bookings, and a relatively small, handcrafted base on-
they do not wish to be informed about this, or a very tology. The degree of automation used and level of user
similar item. input required in generating ontological information from
these sources varies considerably. The reliability of each
The main screen shows items in the Needs to learn and source also varies for a number of reasons, such as input
Needs to learn now categories. The user can override this errors and frequency of maintenance. MyWorkPlace takes
personalised selection of information to see all items and account for the inaccuracy problem with its evidence accre-
their relevance categories by choosing the [Show all items] tion and delayed resolution approach [2]. This means that it
button. can apply simple, explainable reasoning processes for deal-
The Nearby Places category in Figure 2 lists a number ing with con icting and noisy information.
of the places which Fred does not know about. There are Our initial source for location relationships were the build-
many others which he is not informed about as the system ing plans for each oor of the building. Features on the plans
does not believe they are relevant to him. For example with are grouped in the relevant layers. For example all the room
suitable information in the system in the system it might be number labels are in one layer, the room description texts
able to omit details of the Undergraduate Laboratories, as are in another, another layer holds all the doors, one layer
it is semester break so he is not currently teaching classes. holds all the solid walls while another holds all the glass.
Figure 2 shows the view when Fred returns to his o ce. There are over 100 dierent layers in total.
Here are more devices he may wish to learn about. Clicking
on an items brings up more information about the object.
It includes usage instructions and troubleshooting informa-
tion. In addition to this, it includes links to related items,
as suggested from the automatically generated ontologies.
Each time Fred clicks a link requesting more information, a
piece of evidence is added to his user model suggesting he
knows about it.
Figure 3: A section of the building plans with only
selected layers displayed.
Analysis of the lines and text data in each layer allows us
to use relatively simple reasoning to determine rooms and
relationships between them (e.g. distance). Figure 3 shows
a section of the plans (left-hand side sub- gure), and further
extracts showing: a) only wall and door layers (top right-
hand sub- gure), and b) only room number and label layers
(bottom right-hand sub- gure). The majority of the data
Figure 2: The view Fred is presented by MyWork- generated from these plans is assumed to be very reliable as
Place, when he returns to his o ce (room 324). the building was built according to them.
The departmental sta directory is also an important source
for automatic population of an ontology and user models.
It is important for the user to be able nd out why a cer- The sta directory yields a list of all academics, administra-
tain item has been displayed to them, and others have not. tive sta and postgraduate students. It also gives relation-
We call this scrutability. When a user clicks the [Show all ships between people and their research groups (e.g. un-
items] button the full list of items is colour coded accord- dergraduate coordinator, chair of a research group), o ce
ing to which of the ve relevance categories de ned above it or workspace locations, and contact information. A num-
belongs. An explanation of the reasoning used to categorise ber of the student-supervisor links are also available in this
ubiPCMM06:2nd International Workshop on Personalized Context Modeling and Management for UbiComp Applications
database. This data can be extracted with minimal human [3] Chen, H., Finin, T., Joshi, A.: An ontology for
18
interaction. However, it suers from some inaccuracies as context-aware pervasive computing environments.
the phone list is not always kept completely up to date, Knowledge Engineering Review (3) (2004) 197207
for example, when people move o ces or change research [4] Christopoulou, E., Goumopoulos, C., Kameas, A.: An
groups. ontology-based context management and reasoning
Another source of ontological data is the building man- process for ubicomp applications. In: sOc-EUSAI '05:
ual. This documents various elements about the building Proceedings of the 2005 joint conference on Smart
in a glossary and human readable format. Example entries objects and ambient intelligence, New York, NY,
in the manual are an entry for fax that describes policies USA, ACM Press (2005) 265270
for using a fax machine, and an entry for reception that [5] Gu, T., Wang, X.H., Pung, H.K., Zhang, D.Q.: An
gives the opening hours and the location of the reception ontology-based context model in intelligent
desk. We believe we will be able to use Sago, a descen- environments. In: Proceedings of Communication
dant of Mecureo [16], to extract implicit relationships from Networks and Distributed Systems Modeling and
this manual to populate the base ontology. Mecureo is an Simulation Conference 2004. (2004)
ontology learning tool that takes a glossary and mines rela- [6] Heckmann, D., Krueger, A.: A user modeling markup
tionships between the concepts, or glossary terms. language (userml) for ubiquitous computing. In
We propose to use other sources, such as: the university Brusilovsky, Corbett, d.R.e., ed.: LNAI 2702: Proc. of
timetable that lists times and venues for every class in the the Ninth Int. Conf. on User Modeling (UM'03),
university; the departmental calendar and room bookings Springer (2003) 393397
that give scheduled times and locations for certain activi-
[7] Heer, J., Khooshabeh, P.: Seeing the invisible. In:
ties. Both sources may help generate more evidence that
Workshop on Invisible and Transparent Interfaces,
assists the MyWorkPlace system to deduce, say, current lo-
part of Advanced Visual Interfaces (2004). (2004)
cation and activity for a person, and suitable activities for
[8] Edwards, W.K., Grinter, R.E.: At home with
certain venues. In future we may link databases of particu-
ubiquitous computing: Seven challenges. In:
lar class enrollments to generate views of MyWorkPlace for
Proceedings of the Conference on Ubiquitous
undergraduate students. However, this information might
Computing (Ubicomp 2001). (2001)
not be as reliable as some of the other sources.
[9] Chang, A.: Afaik: A help system for the intelligent
The nal source is a small, handcrafted base ontology.
room. Master's thesis, Department of Electrical
This serves as a base for other sources to build on.
Engineering and Computer Science, Massachusetts
Institute of Technology (2001)
5. DISCUSSION [10] Rerscha, A., Keller, M.: Digiscope: An invisible
At the time of writing the new building for which the
worlds window. In: Video in 5th International
system has been implemented has only just been occupied.
Conference on Ubiquitous Computing. (2003)
We plan to evaluate this system by two methods: rstly
[11] Krumm, J., Hinckley, K.: The nearme wireless
we will ask a number of users to perform a task requiring
proximity server. In Davies, N., Mynatt, E.D., Siio, I.,
knowledge of the building, and compare the performance
eds.: UbiComp 2004: Ubiquitous Computing: 6th
of a group of people given access to MyWorkPlace against
International Conference, Nottingham, UK, September
those who are only given access to the data sources used
7-10, 2004. Proceedings. Volume 3205 of Lecture
to populate the ontology (probably building plans, building
Notes in Computer Science., Springer (2004) 283300
manual, and phone list).
[12] Wang, X.H., Zhang, D.Q., Gu, T., Pung, H.K.:
The other form of evaluation we wish to perform is to
Ontology based context modeling and reasoning using
make MyWorkPlace available to volunteer academic sta
owl. In: Proc. of Workshop on Context Modelling and
and postgraduate students, and record their use of the sys-
Reasoning, IEEE PerCom '04. (2004)
tem. In addition to this we will request feedback from a
[13] Liu, H., Singh, P.: Conceptnet - a practical
sample of users.
22
commonsense reasoning tool-kit. BT Technology
One aspect we are particularly interested in is how any in-
Journal (4) (2004) 211226
accuracies in the generated ontological data aect the user
[14] Khan, L., Luo, F.: Ontology construction for
experience. As mentioned previously some aspects of this
information selection. In: ICTAI '02: Proceedings of
data are known to be very accurate, while others are ex-
the 14th IEEE International Conference on Tools with
pected to contain some out of date or incorrect items.
Arti cial Intelligence (ICTAI'02), Washington, DC,
Another piece of future work is to investigate the use of
USA, IEEE Computer Society (2002) 122
a suggested template and categories for entries in the build-
ing manual to better facilitate information extraction. We [15] Richardson, S.D., Dolan, W.B., Vanderwende, L.:
would also like to include data extracted from existing, ma- Mindnet: Acquiring and structuring semantic
turer ontologies. information from text. In: Proc. of COLING '98.
(1998)
6. REFERENCES [16] Apted, T., Kay, J.: MECUREO: Ontology and
modelling tools. WBES of the International Journal of
265
[1] Weiser, M.: The computer for the 21st century.
14
Scienti c American (3) (1991) 94104 Continuing Engineering Education and Lifelong
[2] Carmichael, D., Kay, J., Kummerfeld, R.: Consistent Learning (3) (2004) 191211
modeling of users, devices and environments in a [17] Jannink, J., Wiederhold, G.: Ontology maintenance
with an algebraic methodology: A case study. In:
15
ubiquitous computing environment. User Modeling
and User-Adapted Interaction (2005) 197234 AAAI Workshop on Ontology Management. (1999)