Salil Pradhan, Glenn Steiner
Internet and Mobile Systems Laboratory
1501 Page Mill Road
Palo Alto, CA 94304, USA
web-services often use location information for service customization. Most such
services operate on the user's physical/geographical location. Unfortunately,
such common forms of location carry little context information. In the CoolTown
project at HP labs, we have addressed this deficiency by defined an orthogonal
form of location - Semantic location. Use of this form of location can also help
lower the risk of invasion of privacy.
Semantic location, semantic web, Pervasive Computing, location awareness,
privacy, Ubiquitous Computing.
Location information is an important service customization parameter for
mobile web-services. The most common forms of representing location information
are physical and geographical. In particular, absolute physical location is the
form of choice for most positioning devices such as GPS receivers. However, it
is not easy for humans to express and comprehend1
a <latitude, longitude> pair. Even with an accurate absolute physical location
of a passenger traveling on a public transport bus, a service may not be able to
conclusively determine the bus's route number. As a result, the service cannot
offer richer services like estimating the time-of-arrival at the passenger's
The city of Palo Alto will be recognized by those interested in technology as
an important city in the San Francisco bay area. However, a person who has never
heard of that city won't know where it is. City and other common location
descriptions such as zip codes and postal addresses convey physical location as
well as certain implicit semantics. Such geographical location representations
are useful for services requiring explicit user input, as they are easier for
people to remember and communicate. However, this representation format can be
ambiguous, difficult to sense with devices and too coarse for many applications.
Geographic location representation carries more semantic information than
physical but there is still not enough information in either representation to
determine the nature or purpose of a place.
To address these issues, we have defined an orthogonal form of location -
2 Semantic location
Physical locations specify the position of places 
or points based on a global coordinate system. Geographical locations are
organized hierarchically. E.g., Palo-Alto is contained in California, which in
turn is part of the United States. Among other things, they signify an area or
boundary usually governed by a common law. However, it's not easy to extend or
define a new location, especially if it is just for e-commerce.
Figure 1(from left to right) Physical
location - grid based, Geographical - hierarchical, Semantic - web like
Semantic location is an orthogonal form of location representation, in which
a place is represented by a URI .
Besides a URI, links to a place can have other attributes, perhaps even the
physical/geographical location of the place. The relationship between semantic
locations is best described as "web-like".
locations are unambiguous (URIs are by definition unique), user definable,
extendable, and machine as well as human readable.
This approach of associating semantic attributes to a link is similar to the
approach taken by the Resource Description Framework 
and the Semantic Web 
projects at the WWW Consortium. According to Tim Berners-Lee, most information
on the Web is designed for human consumption. The Semantic Web approach is to
develop languages for expressing information on the web in a machine processable
Annotating places with semantics
In their paper, Scalable and Flexible Location-Based Services for Ubiquitous
Information Access ,
Jose and Davies from University of Minho and Lancaster University respectively
present a generic architecture for building location-based applications. The
emphasis of this work is on translating a physical location into semantic
information. Using their location-based service infrastructure Jose and Davies
have built a tourist guide application.
While building similar applications like the CoolTown WebBus, 
we realized that this approach of mapping semantics to a physical location is
particularly suitable for open areas that are covered by cellular, GPS and other
physical positioning technologies. GPS usually does not work indoor and besides
the positioning accuracy may not be sufficient for many applications2.
Proximity sensing technologies are of great value indoors but may be impractical
in a large open area.
In the CoolTown project, our approach is to label places with URLs 
and to use standard web infrastructure to link and access them. This also allows
us to use a diverse set of technologies to annotate places with URLs and sense
them. Some of the technologies we have explored include:
Beacons: A CoolTown beacon broadcasts a semantic location. Typically,
the range of the broadcast is determined by the Beacon's physical network
technology. General, this range should be short enough to confine the overlap
between places. Short-range wireless technologies such as Infrared or Bluetooth
are suitable physical network technologies for implementing Beacons. CoolTown
beacons are configured to transmit an XML string to any sensor that comes within
range. The XML string includes the relevant semantic attributes and a URL
representing the associated place.
Bar Codes: We 
are also exploring the notion of encoding URLs in variable length barcodes. Such
URL could be extracted by directly scanning the barcode. Of course, existing
barcodes such as UPC will need to be scanned and then resolved in the network
before a URL can be obtained.
Mapping from Physical location: This approach is similar to .
A trusted mapping service translates a nomadic user's physical location into a
URL. This service will be described in a separate paper.
While on-board a public transport bus, John uses his PDA to receive a URL
from a beacon and browses the site. This
URL includes the bus route and a snapshot of the bus' current physical location.
With this information, the service offers richer services like estimating the
time-of-arrival at San Jose bus terminal.
3 Privacy and Safety
With popular service models, a mobile user's physical location and unique
identifier are criterion used to provide personalized service. While at a bus
stop, if a commuter equipped with a WAP enabled cellular phone wanted to search
for the nearest starbucks, she would have to provide her current physical
location (positioned using cellular triangulation or GPS) to a location based
search engine. Along with privacy concerns, this also raises safety concerns.
Unscrupulous services may sell not only her online-profile but also her physical
whereabouts tracked through the day. With the CoolTown approach, the commuter's
phone could receive a URL emitted by a bus stop beacon and browse the site. The
web service at that URL is aware of the physical location of the bus stop and
can assist the user in her quest for the nearest starbucks without asking for
her current physical location. The web service cannot conclusively pinpoint her
physical location; using the web site does not imply that she is at the bus
It is not always possible to populate all places with beacons and therefore
mapping from a physical location to a URL may be required. Again, the CoolTown
approach is useful; a trusted vendor can provide the mapping while the resulting
semantic location could be handled by other services. Since a] a reverse mapping
from semantic location to physical location is not always guaranteed to result
in a unique result and b] use of a semantic location does not imply presence at
a particular physical location, potentially unscrupulous web services cannot
conclusively track their user's physical movements.
A user's location is an important service customization criterion. The most
common forms of location representation are physical and geographical. In
CoolTown, we have explored another form, semantic location.
Semantic locations are globally uniform, unambiguous (URIs are by definition
unique) and links to them can carry as much semantic information as required.
They are represented by URLs, and are linked and accessed using standard web
infrastructure. This is a highly scalable approach as there is no central
control point. Depending on the application, the place, and the desired
accuracy, one of the many URL sensing technologies can be selected. Even at a
single place, a heterogeneous set of technologies can be deployed.
Visiting a semantic location does not imply physical presence at the
associated place. This impedes the ability of mobile e-services to conclusively
track the physical movements of nomadic users, without restricting their ability
to provide localized service.
The CoolTown project includes a number of people at HP Labs. The people
working on location based services include Salil Pradhan, Venky Krishnan, Gene
Becker, Debbie Caswell, Philippe Debaty, Gita Gopal, Glenn Steiner, Tim Kindberg,
John Barton, Steve Loughran, Jeff Morgan, Wei-Ying Ma, Miranda Mowbray, Mirjana
Spasojevic, Mike Spratt, Ed MacDonald & Steve Wright.
 The US postal service doesn't deliver letters marked with
just a latitude and longitude.
 For Example, sensing that a visitor is viewing a particular exhibit in the
 Venky Krishnan et al. Location Awareness in HP's CoolTown.
 Debbie Caswell. Creating a web representation for Places.
 Jose, R. and N. Davies. Scalable and Flexible Location-Based Service for
Ubiquitous Information Access, Handheld and Ubiquitous Computing; Proceeding
from the First International Symposium, HUC'99, Karlsruhe, Germany, September
 CoolTown home page.
 WWW Names and Addresses: URIs, URLs, ...
 Tim Kindberg and John Barton. Private Communication. HP Laboratories.
 Bluetooth home page.
 Resource Description Framework.
http://www.w3.org/RDF/  Tim Berners-Lee. Semantic Web.