Fernando Ribeiro and José Metrôlho
Informatics Department, Polytechnic Institute of Castelo Branco, Castelo Branco, Portugal
Keywords: Adaptive systems, Ambient intelligence, Context-aware applications, Public displays, Pervasive advertising,
Public recommender systems, Socially-aware computing, Ubiquitous computing.
Abstract: Services relevance is strictly connected to the environment in which they are provided and one service that
is highly relevant in a particular environment can be a completely inappropriate service in a different
context or in a different environment. To provide relevant and appropriate services, the systems should be
aware of their environment, the needs and interests of users and should adapt their behaviour to the specific
needs of each particular place and each particular set of users’ current interests. The combination of users’
personal interactions to build the social context of a public place represents an important support to allow
services adaptiveness and thus adjust systems behaviour to best fit users’ needs and space characteristics.
This paper provides an overview of some aspects that involve the combination of personal interactions to
build socially-aware systems. It also describes two experiments, where social context was used to improve
services usefulness. Results indicate that this combination can represent an important aspect to be
considered in the way to provide users with novel, relevant and suitable services.
Ubiquitous computing vision foresees a world where
computing solutions and communication services are
available everywhere and at all time. Achieving this,
entails a new generation of computing systems,
where the environment is continuously sensed and
systems adapt their behaviour to the continuously
changing characteristics of the environment. Only
then it is possible to provide services that fit the
users’ needs and that can help them in their day-to-
day tasks without being too obstructive. In the
broadcasting of content to people in public spaces,
e.g. ads, advices or simply some informative
content, the selection of suitable ways to deliver the
content is central to improve the efficiency of the
system. This is why many advertisers and content
publishers are constantly struggling to find the best
advertising strategies to promote their services. They
need to be able to reach the target population that is
potentially interested in their products or services
and they need to deliver the content properly in
order to capture users' attention. On the other hand,
the dissemination of large displays and mobile
computing devices has created new opportunities for
the joined use of these devices, which allows us to
foresee a range of new applications that can go far
beyond what is supported today, that is essentially in
using these displays for cyclically presenting
advertising or broadcasting information of local
interest. These devices can enrich public spaces
providing users with relevant information and giving
advertisers new opportunities to promote their
products or services. However, promoting ads in
these scenarios requires different strategies from the
ones that are used in traditional media like
newspapers, radio, TV, e-mail, etc., or even from
online ads or traditional notification services where
users subscribe some type of services or drive their
search for content they want. In public spaces the
content is mainly pushed to the users by the
promoters, hoping that it will fit the users’ interests.
In this paper we provide an overview of some
challenges that involve the selection of appropriate
ads in public spaces and we describe two
experiments where personal interactions are used to
support socially-aware content selection and to
improve services usefulness. Early results indicate
that this can be an interesting approach to the
problem of providing users with novel, relevant and
suitable services in public spaces.
Ribeiro F. and Metrôlho J..
DOI: 10.5220/0003825102540257
In Proceedings of the 2nd International Conference on Pervasive Embedded Computing and Communication Systems (PECCS-2012), pages 254-257
ISBN: 978-989-8565-00-6
2012 SCITEPRESS (Science and Technology Publications, Lda.)
There are many challenges when developing
systems for advertising in ubiquitous computing
environments that are not usually found when
developing advertising systems for traditional media
or online advertising. The particularities of these
environments make them very rich in interaction
mechanisms and create new opportunities for several
innovative applications but, at the same time, many
other additional difficulties should be considered.
Some of them are pointed out by (Muller et al.,
2009); (Partridge and Begole 2009) namely: the way
advertisers specify and bid on ads; the extra
complexity of processing contextual information and
how to use this information in an appropriate way;
serendipitous advertisings, privacy and security.
Others could be added: the difficulty in obtaining
information on the interests of the target audience or
how to get user feedback on each presented ad.
Globally, these challenges are not related with the
lack of content but with two central questions: how
to deliver the right content to the right people? and
how to deliver the right content in the best context.
2.1 Delivering the Right Content to the
Right People
How to deliver ads to people who are potentially
interested? How to capture users’ attention? It is not
possible to answer these questions without knowing
something about the users’ interests or activities.
Only being aware of the users’ needs the system will
be able to respond accurately and according to their
expectations. However, interactions in public spaces
should be conducted effortless and potentially
demand little or no prior training. Additionally, in
public spaces, it is important to consider that the
content is targeted to a large audience instead of a
person and thus rather than having a person profile
we need a profile that combines the preferences of
the multiple users that may see the content.
2.2 The Best Context
It is well known that the relevance of a particular
content is strictly connected with the target
population and with the context where it will be
delivered. An ad can be highly relevant in a business
place but it is totally misplaced and without
relevance if it is delivered in a school. An ad can be
very relevant in a time period before lunch but lost
its entire relevance after lunch. However, delivering
the content at the right time is not only about hours
or minutes. The main question is how to figure the
best possible time and the best possible context to
deliver the ad. The set of contextual information that
can be used to improve content selection accuracy in
public spaces is vast. Information concerning the
social environment, the physical and social space,
available devices to deliver the content, time, and
also the views history can play an important role in
the selections of the next content to be presented. A
better use of contextual information can allow
pervasive advertising to be more effective and can
also better fit to the interests of the target population
but also the interests of the advertisers/publishers.
Several advertisements systems that adopt pervasive
computing technologies have been presented and
discussed in recent years. A significant part of them
are proposed to deliver online ads (Edelman et al.,
2007); (Googl inc. 2011) but some of them have
been built to deliver context-aware ads in public
places. (Partridge and Begole 2009); (Orsi, 2011)
use mobile phones as the main device to display ads
and other. Other systems explore public devices, like
large displays, in public spaces combined with
personal mobile device to understand users’ interests
and activities and thus to deliver ads for groups of
people instead of a person. (Partridge and Begole
2009) explore how can histories of contextual data
help target advertising more effectively. The
BlueScreen (Payne et al., 2006) uses Bluetooth-
enable devices for identifying users and it explores
history information of past users’ exposure to certain
sets of adverts. (Holleis et al., 2010) combine public
screens and mobile devices and explores personal
user profiles and explicit input from users in order to
provide appropriate ads. They use the users’ mobile
phones to store information about their owners, track
their activities, and let them interact directly with the
display. There are many proposed systems that try to
provide personalized ads to individuals or for
recommending content to people in public spaces
but few of them base their decisions in the social
context that represents the place visitors’ interests.
When developing systems to delivering ads in public
spaces, there are two viewpoints that should be taken
into consideration: the advertiser viewpoint and the
consumer viewpoint. While the viewpoint of the
advertiser is easily to include in the system
behaviour using relatively simple contextual rules,
including the consumer viewpoint could be a
complex task.
4.1 Social Interests of the Audience
We build a social characterization of the public
space using information from personal interactions.
Each user is associated to a bag of tags that represent
its past interactions. Tags that are used more recently
are associated to higher weights. The system also
distinguishes between users that are currently in the
space on users that already leave the space giving
higher weights to present users. All these
information is combined into a dynamic social
model of the place. The social model is combined
with a set of contextual rules that characterize each
ad to decide what and when to present a specific ad.
4.2 Ads
When create an ad, there are essentially two aspects
to be considered. The content and its appearance that
should be chosen to attract users’ attention and the
preferable contextual conditions in which each ad
should be selected. The first aspect requires special
attention from the design viewpoint and it is not the
focus of this work. The second aspect is a central
question in the selection process. It represents the
advertiser viewpoint of the preferable conditions
under which the ad gets more added value. In this
case the advertiser may define a set of contextual
rules per each which include rules about: time,
weather, category, a set of associated keywords,
location and the number of times to be presented.
Both systems described in next subsections were
developed with the main goal of delivering
appropriate information for people in public spaces.
Both use information acquired from user mobile
devices interactions in the form of tags (Bluetooth
and NFC) to build socially-aware systems and to
support their decisions about what is the best ad to
be delivered at each time.
5.1 Public Ads with Bluetooth
In this system we explore Bluetooth capabilities as a
medium to express users’ interests. Users express
their interests defining a profile in the system or
spontaneously interacting with the system using
keywords in their Bluetooth device name to specify
their interests. Using Bluetooth scanners with
different ranges, the system differentiates two types
of interaction zones: a mobile zone and a display
zone. This is used to identify the user location and
thus to select the most appropriate device to deliver
the information. When he is detected within the
display zone his profile and his interactions are
processed conjunctly with profiles and interactions
of other users in the space and they are processed to
support decisions on the next content to be presented
in the public display.
5.1.1 Selecting Content
The first task is to decide if every ad can be
presented in the public display or if the ad is not
classified as public and thus it should be delivered
only to the users’ personal mobile device. In the
mobile zone the system delivers new pending ads for
registered users according to their profiles. In the
display zone, if there are no Bluetooth devices the
display presents cyclically available public ads
according to the social context. If there are
registered devices the display presents ads related to
the profiles and general content. If there are explicit
interactions (using keywords in the Bluetooth device
name) the system searches for related ads (related to
the topic of the keyword).
5.1.2 Evaluation and Results
After 3-weeks of experiment, the system detected
103 distinct mobile devices and 23 of them are
registered in the system. The ratio of detected
devices that received messages was 22.3%. During
this period the system delivered 62 personal
messages, corresponding to 14 distinct messages, to
23 distinct students. After the experimental period,
we’ve asked the users to fill a questionnaire where
they were able to express their opinion. More than
73% of them refer that the system presents some
benefits over common online software to support
students/teachers interaction.
5.2 Public Ads with NFC
Using NFC as interaction mechanism, users are able
to explicitly express their interests in a particular
moment. This allows the user to directly access to
the desired information using his mobile device but,
at the same time, the combination of the multiple
interactions of the users allow characterizing the
PECCS 2012 - International Conference on Pervasive and Embedded Computing and Communication Systems
space interests, in the form of a social profile, and
thus presenting in a public display ads in which
other people might be interested in. This allows
delivering to people in the place the ads that are
representative of the interests of people who attend
the same space and thus possibly with similar
Figure 1: Public Ads with NFC.
5.2.1 Selecting Ads
The system architecture can be seen in Figure 1.
Each NFC tag is associated to a category of ads
(food, clothes, shoes, etc.). By reading the NFC tag
the user could access to a web page with a list of ads
with a short summary (the user may select each
specific ad and read more information) that are
related to the category in the NFC tag. To select the
next ad to be presented in the public display we
realize a matching between the ad, contextual rules
and the real context (social context), and we use the
presentation history to avoid the presentation of the
same ad too often.
5.2.2 Evaluation and Results
We have run a preliminary evaluation with four
users to provide a better insight into the system,
which may identify new issues/problems to consider,
and to obtain initial feedback on users’ acceptance
and perception of the system main features. The
prototype includes a smart poster with three distinct
NFC tags and a display that presents content related
to the place interests. The system includes 5 ads
within each category (NFC tag) which totals 15 ads.
All users referred that the usability of the system
as a strength. However, three of them referred that
they did not have associated the behaviour of the
display to their interactions. This may be related to
the time period in which the users stay in front of the
display (because the display does not respond
immediately to the users’ interactions in order not to
denounce their actions and preserve their privacy)
but can also mean that some adjustments may be
needed in the behaviour of the display.
We propose an approach that combines public
displays and personal mobile devices for delivering
relevant content in public spaces. It combines tags
from personal interactions to build a social context
of the public space and uses this information to
deliver appropriate content to people in public
spaces. We run two experiments based on those
principles. Initial evaluation results suggest that the
combination of public displays and tag-based
interactions from personal mobile devices represent
an interesting combination to provide relevant and
personalized information for users in public spaces.
Currently we are working in different algorithms
for representing the social context and for improving
the content selection in the public display. We are
analysing: include more information on the
algorithm; contextualize users’ interactions
(location, organizational context) giving a
contextualized sense to the users interactions and
defining an economic model that appropriately
represents the added value of intentional interactions
and social context.
Edelman, B. B., M. Ostrovsky, et al., (2007). “Internet
Advertising and the Generalized Second-Price
Auction: Selling Billions of Dollars Worth of
Keywords.” American Economic Review 97: 242-259.
Googl inc., (2011). Online Advertising by Google, Online:, Accessed date: Jan. 2011.
Holleis, P., G. Broll, et al., (2010). Advertising with NFC
Workshop on Pervasive Advertising and Shopping
(PERVASIVE 2010), Helsinki, Finland.
Muller, J., P. Holleis, et al., (2009). Pervasive Advertising.
2nd Workshop on Pervasive Advertising in conjuction
with Informatik, Lübeck.
Orsi, L. C. a. G., (2011). A New Perspective in Pervasive
Advertising, Oxford University: 21.
Partridge, K. and B. Begole, (2009). Activity-based
Advertising: Techniques and Challenges. Workshop
on Pervasive Advertising. Nara, Japan.
Payne, T., E. David, et al., (2006). Auction Mechanisms
for Efficient Advertisement Selection on Public
Displays. European Conf. on Artificial Intelligence,
Riva del Garda.