Content uploaded by Pierrick Thébault
Author content
All content in this area was uploaded by Pierrick Thébault on Sep 20, 2015
Content may be subject to copyright.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article xx, Publication date: Month YYYY
Embodying Services Into Physical Places :
Toward the Design of a Mobile Environment Browser
PIERRICK THEBAULT, Alcatel-Lucent Bell Labs France, Arts et Métiers ParisTech
DOMINIQUE DECOTTER, Alcatel-Lucent Bell Labs France
MATHIEU BOUSSARD, Alcatel-Lucent Bell Labs France
MONIQUE LU, Alcatel-Lucent Bell Labs France
The tremendous developments in mobile computing and handheld devices have allowed for an increasing
usage of the resources of the world-wide web. People today consume information and services on the go,
through smart phones applications capable of exploiting their location in order to adapt the content
according to the context of use. As location-based services gain traction and reveal their limitations, we
argue there is a need for intelligent systems to be created to better support people’s activities in their
experience of the city, especially regarding their decision-making processes. In this paper, we explore the
opportunity to move closer to the realization of the Ubiquitous Computing vision by turning physical
places into smart environments capable of cooperatively and autonomously collecting, processing and
transporting information about their characteristics (e.g. practical information, presence of people, and
ambience). Following a multidisciplinary approach which leverages psychology, design, and computer
science, we propose to investigate the potential of building communication and interaction spaces, called
information spheres, on top of physical places such as businesses, homes, and institutions. We argue that,
if the latter are exposed on the web, they can act as a platform delivering information and services and
mediating interactions with smart objects without requiring too much effort for the deployment of the
architecture. After presenting the inherent challenges of our vision, we go through the protocol of two
preliminary experiments that aimed to evaluate users’ perception of different types of information (i.e.
reviews, check-in information, video streams, and real-time representations) and their influence on the
decision making process. Results of this study lead us to elaborate the design considerations that must be
taken into account to ensure the intelligibility and user acceptance of information spheres. We finally
describe a research prototype application called Environment Browser (Env-B) and present the underlying
smart space middleware, before evaluating the user experience with our system through quantitative and
qualitative methods.
Categories and Subject Descriptors: [Human-centered computing]: Ubiquitous and mobile computing
General Terms: Ubiquitous computing, Smart Objects, Smart Environments
Additional Key Words and Phrases: Situated Services, Location-Based Services, Mobile Browser,
Interaction Design, User Interface Design, User Study, Web Of Things.
ACM Reference Format:
Pierrick Thébault, Dominique Decotter, Mathieu Boussard, Monique Lu, 2013. Embodying Services into
Physical Places : Towards the Design of a Mobile Environment Browser. ACM Trans. Interact. Intell. Syst.
!9,!4,!Article!39!(March!2010),!6!pages. !
DOI:http://dx.doi.org/10.1145/0000000.0000000
1. INTRODUCTION
For years, researchers of various communities have been pursuing a vision of the
future where technologies disappear into the background and proliferate at the same
Author’s addresses: P. Thébault, LAMPA, Arts et Métiers ParisTech, 2 boulevard du Ronceray, BP 93525,
49035 Angers Cedex 01, France; D. Decotter, M. Boussard, M. Lu, Bell Labs, Alcatel-Lucent, Route de
Villejust, 91620 Nozay, France.
Permission to make digital or hardcopies of part or all of this work for personal or classroom use is granted
without fee provided that copies are not made or distributed for profit or commercial advantage and that
copies show this notice on the first page or initial screen of a display along with the full citation.
Copyrights for components of this work owned by others than ACM must be honored. Abstracting with
credits permitted. To copy otherwise, to republish, to post on servers, to redistribute to lists, or to use any
component of this work in other works requires prior specific permission and/or a fee. Permissions may be
requested from Publications Dept., ACM, Inc., 2 Penn Plaza, Suite 701, New York, NY 10121-0701 USA,
fax +1 (212) 869-0481, or permissions@acm.org.
© 2010 ACM 1539-9087/2010/03-ART39 $15.00
DOI:http://dx.doi.org/10.1145/0000000.0000000
39
39:2 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
time. Since Mark Weiser’s foundational paper [Weiser 1995], many agenda were
framed to realize the concepts of these pervasive [Satyanarayanan 2001], situated
[Hull et al. 1997], sentient [Hopper 2000] or disappearing [Streitz and Nixon 2005]
computers, which leverage new services discovery and delivery mechanisms to
engage people in seamless user experiences [Rogers 2006]. With the ongoing
miniaturization of integrated circuits, communication chipsets and portable sensors,
it has indeed become possible to embed computational capabilities into clothes,
things, appliances or furniture and to move closer to the realization of smart objects’
potentials. Over the last decade, a number of intelligent systems not only capable of
processing information but also of reasoning in a distributed way have been
designed, mainly from a technological perspective. This led researchers to identify
the challenges of implementing such systems [Abowd and Mynatt 2000; Kindberg
and Fox 2002; Edwards and Grinter 2001; Davies and Gellersen 2002], whose
deployment in real-life context requires tremendous efforts, and to consider the need
of evaluating them with users in order to ensure their adoption [Carter and Mankoff
2004]. While there is still significant research to conduct, especially in the way
services permeate objects or environments are made visible and available to users,
Bell and Dourish argued that ubiquitous computing is already here, “in the form of
densely available computational and communication resources” [Bell and Dourish
2007].
If the messy assemblage of heterogeneous technologies that we experience everyday
is far from being the truly ubiquitous and seamless infrastructure foreseen by
Weiser, it offers a widespread access to the resources of the World Wide Web in a
variety of settings or situations. By maintaining the connectivity with the network
and providing a variety of services with context information, mobile phones or tablets
indeed allow for data, information and content to be retrieved, filtered and presented
on a screen to fulfill users’ needs or support their activity. With the proliferation and
broad adoption of mobile telephony, location-based services (LBS) [Rao and
Minakakis 2003] bring people closer to the “when you want, where you want, what
you want” pervasive computing motto. The integration of global positioning system
(GPS) chipsets in handheld devices allowed navigation assistance, local search, geo-
tagged content navigation and location sharing tools to be released and used outside
laboratories. In mobility, such services provide users with a representation of
surrounding physical places (we will also refer to the latter as “places” or “locations”
in the rest of the paper) and communicate potentially relevant information in a
variety of forms.
Despite all the content generated about places on recommendation platforms, social
networks or photo/video-sharing applications, public places such as shops,
restaurants, bars or museums only exist on the Web in the form of a fuzzy collage of
information – while private places such as homes and companies often remain
invisible. As pointed out by Kindberg et Al. in their work on Web presence, “the
physical world and the virtual world would both be richer if they were more closely
linked” [Kindberg et al. 2002]. Instead of fetching data from multiple sources, users
could directly connect to a service representing their immediate or proximate
location. By putting information in context, whether in the form of a webpage or a
portal, we could facilitate information retrieval and provide place-owners with a new
communication and interaction space. This would open the possibility for potential
resources of the place (e.g. smart objects, public displays, local services, sensors
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:3
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
information, etc.) to be made visible to users.
Within the context-aware community, researchers have been running tests-beds of
smart environments (homes [Helal et al. 2008; Skubic et al. 2009; Preuveneers and
Berbers 2012], factories [Lucke et al. 2008; Spiess et al. 2009; James 2012],
supermarkets [Bajo et al. 2006; Black et al. 2009; Kahl and Burckert 2012], museums
[Jbara et al. 2007], etc.) for a long time. Such instrumented environments capable of
monitoring the presence of people, inferring their activities and seamlessly delivering
services on various devices, offer much richer experiences than location-based
services. They reflect the dynamic aspect of information in places and allow in-situ
interactions with a number of intelligent systems (e.g. home automation, follow-me
computing, telepresence, etc.). They nevertheless require heavy infrastructures, are
rarely designed for allowing off-situation access to information or resources and do
not interconnect to form the fabric of the city. In this paper, we propose to work
towards building a vision of the urban environment where information and services
are delivered by places. We claim that creating situated information spheres fitting
the actual boundaries of locations, aggregating content, as well as mediating
interactions with resources of the place would enhance the city experience and lead
to new uses.
To investigate this topic from a user-centered design perspective, we conducted our
research to answer the following questions: 1) what kind of information and services
should places be delivering to users? 2) What representations and tools should be
designed to make these information spheres intelligible and accessible? 3) What
technologies and infrastructures should be leveraged to engage users in the
experience we designed? Our paper reflects this approach and is structured in four
main parts. After illustrating our vision with a use case scenario and discussing the
inherent challenges, we present in section 3 the user studies that led us to validate
some of our hypothesis. The generated insights guided the creation of a generic
mobile browser aiming at navigating on places that we present in section 4. In
section 5, we describe the prototype implementation and the overall infrastructure
that supports our system. In section 6, we detail the methods that led us to evaluate
the user experience of this system and discuss the results of a task-based experiment.
As a team composed of psychologists, designers and computer scientists, we expect to
contribute to the community by providing an example of practical research resulting
from a multi-disciplinary approach.
2. VISION AND CHALLENGES
The way we get information about physical places such as businesses, offices and
homes is changing. While people yesterday had no other option but to consult a
directory and make a phone call or to move to the actual area to know more about a
place, they are now able to leverage a number of resources on the Web such as local
search engines, user reviews platforms, location sharing tools and social networks.
On these services, places exist as specific entities that are often characterized by GPS
coordinates and represented by pages aggregating editorial or user-generated
content. Place-owners usually have the opportunity to “claim” their place so that they
can augment or correct the provided information. This brings users to examine a
distributed set of overlapping information, whose accuracy can be questioned, and
place-owners to deal with multiple platforms. In order to ease the information
retrieval and management, we propose to put information in context by creating
39:4 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
situated information spheres on top of physical places. In this section, we illustrate
with a use case scenario this vision of a city where locations mediate information and
services, and then present the challenges that need to be addressed.
2.1 Vision and motivating scenario
Places have their own characteristics. They can be defined as particular portions of
space with definite boundaries, occupied by or allocated to persons and things. They
deliver services and/or relate to specific activities that can change over time or
unexpected situations. While very practical information (e.g. contact details,
opening/closing hours, amenities) are likely to be gathered and published on the Web,
especially with the help of users, specific aspects including time dependent ones (e.g.
menu, schedule of events, special offers in case of a business) require a continuous
effort of place-owners. If fragments of information are distilled in the content
aggregated by services, users often have to be physically present in the place or
contact the place owner to obtain details and to form a global representation of what
the place has to offer. This can be explained by the lack of time, motivation or
competencies of place owners to communicate more up to date or precise information
on multiple services. To avoid situations where owners would prefer one
communication channel to another (e.g. for example a Facebook page versus a
dedicated web site), leading to the haphazard dispersion of information, we argue
that places should be able to reflect their own characteristics in an autonomous way.
Fig. 1. Illustration of information spheres augmenting places.
To do so, we propose to augment places with information spheres that overlap
locations with respect to their boundaries. Following place-centered techniques, that
link virtual spaces to physical locations [Jones and Grandhi 2005], information is
replaced into context, layered on the real world, and constrained into places, as
people perceive them. We do not want people to think of this information as filtered
and adapted according to their location, as it happens with location-based services,
but more as embodied in the environment (Figure 1). We argue places should be
“smarter” and act as platforms not only delivering context-specific content, but also
exposing services and objects. By creating dedicated hyper local websites, or place
portals, we also want to provide people with new ways to retrieve and consume such
resources: off-situation (remotely) and in-situation (while being present in the
location). Such dimension, which is not yet explored by LBS providers, would allow
enriching the user experience of places. It would allow people to interact with new
situated services or connected objects, bringing us closer to the realization of the
M
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:5
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
Ubiquitous Computing vision. To illustrate the benefits of information spheres and
highlight their potential impact on users’ perception of environments, we created the
following use case scenario:
“Julia and her friends have planned to meet at the Luxembourg train station in order
to spend the afternoon in the surrounding Parisian gardens. Despite sunny forecast,
it starts raining a few minutes after they sat on the outdoor chairs. They hurry up
and leave the park to look for a nice place to spend some time. They walk down the
nearby streets and cannot find anything but crowded cafes and noisy pubs
broadcasting soccer games. After a couple of minutes, Julia’s friend Marco suggests
using his mobile phone to find a proper place. He quickly scans the area and starts
browsing places that can accommodate his group. His mobile application provide him
with information such as users’ reviews (i.e. aggregated from various Web services),
the immediate availability of seats, the type of drinks that are served or the overall
ambience… Marco quickly discovers a cozy coffee shop located on a side street, which
is playing jazz music and exhibits interactive installations. They decide to check it
out and walk to the spot, and finally spend a moment there. Inside the coffee shop,
they enjoy their beverages and pay more attention to the art pieces that consist in
generative graphics displayed on several screens. The place owner and artist work
together to offer a means of interaction through the mobile application Marco
previously used. Customers that are physically present in the coffee shop can post
content and messages that will be converted into graphics. They also get the
opportunity to interact with the hi-fi system by voting for the next tracks or playlists
that will be played.”
2.2 Challenges
By augmenting places with information and interaction spaces, we seek to enhance
the way the physical world is represented in the digital world. The exposition of
places as resources of the World Wide Web indeed opens up the possibility for places
to be considered as the primary source of information and to become the most-
preferred access point for users. We believe such approach will help to reinforce the
relationship users have with places, especially if new delivery mechanisms are
shaped for specific postures or needs. As illustrated in the previous part, there is an
opportunity to make people come and stay in the location to consume, appreciate and
enrich potential new services, as well as letting them interrogate places on-the-go to
help them make decisions on which place to visit. To ensure the user adoption of our
vision, there is therefore a need to carefully investigate the way such information
spheres are defined, communicated and supported. In this part, we propose to
examine the three main challenges that our work encompasses and that guided our
research.
Defining information spheres
By embodying information and services into places, we open up the possibility for
different types of information to be communicated to users. This raises numerous
questions: what contents are likely to augment locations? Which situated services
can be designed? How do they relate with the type of place and make sense for the
users? Who or what generates them? If answers obviously differ according to the type
of places (i.e. public aspect, supported activities, amenities, etc.) and owners’
motivations to publicize their location or business, the challenge is to design
information spheres as relevant and engaging entry points to a service offerings
39:6 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
leveraging up-to-date information. The challenge is therefore to ensure the dynamic
and legitimate aspects of information spheres by investigating the need to combine
“editorialized” information (i.e. administrated by place-owners) with user-generated
content (i.e. aggregated from web services). We can also rely on connected objects or
sensors that are physically present in places to provide users with more information
regarding the place (e.g. presence of people, ambiance, activities, etc.), ultimately in
real-time. To ensure the adoption of our system, we argue that a user study is
required to address these questions. We set up a qualitative study aiming at
capturing users’ needs in terms of information consumption to initiate our research,
as described in section 3.
Making information spheres intelligible
After collecting insights about the way information spheres should be composed and
defining the type of information and services to be delivered, we need to think about
the way they will be presented to users. This brings up new questions related to the
creation of tools allowing people to navigate through places, to reveal their
information spheres and to interact with the provided content and services. In a
mobility context, handheld devices appear to be a relevant medium to develop a user
interface on. However, how to communicate the concept and boundaries of
information spheres on a small-sized display? How to allow people to scan and
connect to these spheres in different situations? How to represent and hierarchize
information to provide them with an overview of the place’s characteristics? It is not
only necessary to make information and services intelligible but also to ensure that
the people will not be overloaded by irrelevant data, while allowing some level of
serendipity at the same time. The challenge consists in reflecting the boundaries of
information spheres and to make them more “tangible” to users. If notifications or
more subtle alerts can be implemented on a smartphone or a tablet, place owners can
also leverage signs (e.g. stickers, badges or posters) or displays to inform people that
their place is augmented. We argue that it is up to designers to build upon the results
of the preliminary user study and to make choices regarding representation modes.
Supporting the deployment of information spheres
Although the question of the supporting infrastructure in Ubiquitous Computing is a
long-standing problem [Fox et al. 2006], it is undeniable that technology has largely
progressed both from an infrastructure side (sensor networking, lower than ever
price of connecting objects to the network, ubiquity of network access…) as well as
from a computer science side (advent of mobile computing, HCI…). However, a
number of questions concerning the technologies and infrastructures that need to be
leveraged to engage users in the experience we designed have to be addressed. They
obviously relate to the enablers that need to be created to embody and represent
places as resources of the Web, but also to the mechanisms can be leveraged to
retrieve the position of users. Thus, how to gather and maintain information from
various sources about a given place? What software components are required to
maintain an updated and filterable list of information spheres? What protocols and
Web standards should be used to create accessible representations of information,
services or objects? What modules need to be implemented to create a mobile client?
Besides technical considerations, the challenge is to reflect the feeling, or create the
perception, that information and services are really embodied into places. The
underlying infrastructure should therefore ensure responsive interfaces and bring
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:7
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
people to believe that it is easier, faster and more secure to interact with information
spheres than querying information on the Web.
3. INVESTIGATING THE INFLUENCE OF INFORMATION ON DECISION MAKING
When we think about information spheres, we imagine places capable of really
reflecting their current states as well as more static characteristics as available on
the Web today. We ideate systems and interfaces that allow users to remotely “sense”
the ambiance of a location, by communicating information about people who were,
are or even plan to spend a moment in the place, about the activities that can be
supported and the ones that are actually performed. We envision interactions with
situated services and smart objects, which can be accessed in a glimpse and whose
access is shared among customers or visitors. If such possibilities help us defining the
type of information and services that are likely to augment places, they are however
really far from what people can experience today with mobile location-based services.
It is therefore really difficult to validate the user relevance of such scenarios, in a
sense that users generally have difficulties to project themselves in a reality that
involves a new paradigm (such as information spheres). In order to conduct a user
research that provides us with useful feedback and insight, we then have to make
sure that people can apprehend the representation of the future that we present
them. In this section, we explore potential users’ informational needs from their
usage of an emergent technology: mobile location-based services. We then describe
the qualitative study that led us to investigate the influence of information provided
by such tools on users’ decision making and discuss our findings.
3.1 Location-based services and decision making
If the possible uses of location-based services are still evolving, such emergent
applications today provide users with most of the available online information about
places. Despite their technical constraints, especially in indoor environments, they
are generally used in a mobility context, whether to check reviews, friends’ positions
and to find special offers or discover new venues. In many cases, such information
aims at helping people to decide where to go [Claus and R. Martin 2004]. By
providing users with information about places that surround them, location-based
services allow them to form a better representation of what to expect and to develop
decision strategies. Behavioral theories in psychology indeed showed that people
generally take into consideration the value, uncertainty and alternatives of their
choices. Information has therefore not only a strong influence on users’ decisions but
is also essential for the decision-making processes to take place.
As has been pointed out by psychology researchers, everyday human decision-making
does not conform to idealized normative models (see, e.g., [Gigerenzer, 2007],
[Jameson 2012]). In everyday life, situations and goals vary across people. This leads
them to make decisions in an opportunistic manner and to develop different
strategies based on the context, the product/service or their personal experiences (cf.
[Solomon, 1996]). While social factors, emotions, preferences, and prior experiences
also have a strong influence on decision tasks, the process mainly depends on
memory and cognitive processes such as information acquisition, storage, retrieval
and use. Decisions indeed rely on long term memory, which stores procedural
information, and on short-term (or working) memory, which acquires perceived
information. This leads users to put available information in perspective with a
39:8 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
comparable past experience or to consider it in an opportunistic way to make a new
choice.
By providing users with information about places, location-based services assist
working memory. They indeed aim at presenting the most relevant information
considering users’ location and preferences (e.g. distance, categories, price, presence
of friends, etc.) in the most intelligible way. This sometimes brings people to face an
amount of information that they do not have the time and the capability to process.
Indeed, limits of memory capacity lead people to limit their cognitive effort and, in
many cases, to adopt a less accurate strategy. It is therefore needed to structure and
prioritize information in order to enhance people’s capability to achieve decision
tasks and to reach their final decision. In the context of our work, we are especially
interested in exploring how users of location-based services rely on information and
how our approach allows introducing new types of information that can better assist
them in their decisions. To do so, we conducted a short user study with a small group
of lead users.
3.2 User study
Envisioning places as information spheres that deliver information and services is a
difficult task for people who are not already familiar with location-based services. In
order to investigate users’ preferences for certain types of place-related information,
we conducted a study based on two experiments (summarized in Table 1) where users
had to complete a decision-making task (i.e. finding a bar) using a prototype
application. In a first step, we especially sought at comparing users’ perception of
information in a planning and in a mobility context while in the second one, we
focused on the evaluation of different types of information in mobility. To do so, we
designed a simulated service aiming at inducing and observing users’ behaviors that
are likely to happen in a real situation. This artifact represented existing types of
information such as address, user reviews and check-ins (i.e. user self declaration of
his current location), but also alternative informational modalities such as a
simulated live video stream or a representation of the place’s ambience (as seen on
Figure 2). Shaped as a widget, this ambience representation combines noise level of
the place and the seats’ availability. We are aware that the video stream and the
ambience representation do not cover the whole scope of our vision, especially in
terms of interactions with objects and services, however they bring in a real-time
aspect that is rarely developed in location-based services. They indeed communicate
to the user the idea that places have been instrumented in some way, either with
connected objects or sensors, and that they are able to generate their own
information in a continuous way. With this approach, we expect to awaken the
curiosity of users and capture feedback about the relevance of different information
types according to the context of use.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:9
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
Table I. Presentation of the experiments
Experiment
A
B
Objective
Evaluate the relevance of real-
time information according to
the context of use.
Evaluate the type of
information in a mobility
context.
Conditions
2
6
1) Mobility context.
2) Planning context.
1-5) All bars display the same
type of information.
6) Each bar communicates a
different type of information.
Task
Use the application to find nearby bars, check the information
provided by each of them and make a choice.
Number of
bars
3
5
Participants
Twelve participants (mean age=29.3; standard deviation=7.4) having an extensive
use of location-based services were recruited through an online survey. The panel
was composed of nine men and three women living and working in Paris. All of them
owned either an Android phone or iPhone, which they often use to obtain information
about places and to leverage location-based services to assist them in their decisions.
In order to create a heterogeneous panel, we gathered active users of Foursquare,
Yelp, Qype, DisMoiOù, PagesJaunes, Google Maps or Allociné, which are among the
most famous applications in France. Each participant was using at least two of these
applications on a regular basis to fulfill his or her information needs. We therefore
consider their level of proficiency as advanced. They indeed demonstrated the skills
to understand a variety of navigation techniques, to perform searches and filtering
operations, to handle complex functions such as location declaration and sharing as
well was to achieve simultaneous tasks that involve different applications. As
location-based services are still considered as emerging in France, we also identify
our participants as lead users. According to Von Hippel’s theory [E. Hippel 2006; E.
V. Hippel 1986], lead users are aware of their needs before products’ releases and are
more likely to propose and evaluate solutions to the issues they face everyday. All of
them volunteered for the study and were granted a 20 euros voucher for their
participation.
Mockup
A mobile mockup application was created for the study, as depicted in Figure 3,
showing a map where the position of the user is simulated with surrounding bars.
Each bar shown on the map was placed equidistantly from users’ simulated location,
and labeled with simple letters (i.e. from A to E) to prevent users from choosing a
place from any familiar name. Mockup A for our first experiment presented three
bars while mockup B for the second showed five bars. The selection of a place
triggered the sliding of another page presenting a specific type of information. Users
could go back to the main screen at any time. Eight versions of the mockup
application were created to cover all the conditions of both experiments. All of them
were running on an iPhone 4.
39:10 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
Fig. 2. The “ambience” widget and a video stream.
Fig. 3. User interface of the mobile mockup application provided to users.
Procedure
Participants were asked to use the prototype to decide which bar they would select,
according to specific situations and scenarios. They had to navigate through the
different places, consider available provided information and communicate their
choice regarding the place they would select. After each task, the participants were
invited to answer a short questionnaire based on a six-point Likert scale to evaluate
the degree of perceived usefulness regarding different types of information that were
presented to them in order to take their final decision. A video-recorded interview
was conducted after the two experiments to collect user feedback and discuss their
perception of the ambience widget.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:11
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
Conditions and scenarios
In experiment A, participants were asked to pick a bar according to two experimental
conditions as shown in Table I. In the first condition, the user is in a mobility context,
looking for a bar in order to have a drink with a friend who will join her 30 minutes
later. In the second condition, a planning scenario brings users to imagine being at
home and looking for a place where they could hang out with some friends in the
forthcoming weekend. For each condition, three bars were shown, each having one
type of information proposed to users, as described in Table II. In experiment B,
users went through six different versions of the prototype displaying five different
types of information, as summarized in Table III. For each condition, they were put
again in the situation where they are waiting for a friend. They had to spend their
time finding a place to go with him.
Table II. Experiment A – type of
information presented to users
Condition
Bar
Type of
information
Real-
time
aspect
1 –
Mobility
A
Address
No
B
Weekly
ambiencea
No
C
Ambience
widget
Yes
2 –
Planning
A
Weekly
ambiencea
No
B
Ambience
widget
Yes
C
Address
No
a The ambience widget has been adapted to
show the average sound level for each day of
the previous week.
Table III. Experiment B – type of
information presented to users
Cdt
Bar
Type of
information
Granularity
1
A-E
User reviews
(Yelp-like)
Excellent, good,
neutral, negative,
none.
2
A-E
User check-ins
(Foursquare-
like)
Many, few, none.
3
A-E
Video stream
Moving or fixed
point of view.
4
A-E
Ambiance widget
Calm, crowded and
noisy.
5
A-E
Address
Name of the bar
and address.
6
A-E
Each bar shows one type of
information (from 1 to 5)
3.3 Results and discussion
In experiment A, participants evaluated the relevance of real-time information
according to the situation of use. In order to determine if people are likely to prefer
real-time (i.e. the ambience widget) or non-real time information (i.e. weekly
ambiance), we conducted a paired-sample t-test on the results of the questionnaires.
Summarized in Figure 4, the test shows that there are significant differences
between the use of non real-time information (t(11)= -2.72, p=0.02) and real-time
information in a mobility (M=3.33, SD=0.89) and planning (M=4.42, SD=0.90)
contexts. In the same way, real-time information is not used in the same way
(t(11)=3.34, p= 0.007) in a mobility (M=5.17, SD= 0.84) and in a planning situation
(M=3.33, SD=1.435). Unsurprisingly, this confirms that users prefer to use non real-
time information when they want or need to plan a forthcoming event and are willing
to rely on real-time information when they need to make opportunistic or immediate
decisions.
39:12 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
Fig. 4. Results of experiment A: information relevance according to the context of use.
Note: The difference between the "mobility" and "planning" contexts is significant by a 2-way
paired-sample t-test with p < .02 for each of the variables "weekly ambience" and "ambience
widget".
In experiment B, the task of participants was to evaluate the usefulness of several
types of information in a mobility context. According to the results of the
questionnaires (Figure 5), user reviews are considered as the most useful type of
information (M=5,17) when people are looking for information about places. Although
webcams are today rarely installed in bars, the video streams that we made available
to users was perceived as useful (M=4.42), but do not have an influence on all
participants’ decisions (SD=1.24). User check-ins are also considered as relevant
(M=3.50), but again opinions strongly vary among users (SD= 1.38). The ambiance
widget we introduced has finally been positively perceived. Its mean is slightly
higher than the average score (M=3.33, SD= 0.89).
Fig. 5. Results of Experiment B: Users’ perception of usefulness of 5 types of information. (Error bars show
the standard error of the mean.)
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:13
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
The main objective of the interviews we conducted at the end of the experiments was
to gather feedback about the perception of the ambience widget. As the latter was
presenting information about the sound level of the place, as well as its occupancy,
we tried to measure the interest people might have for these two dimensions. Ten out
the twelve participants (83.3%) considered the metrics about the seat availability as
relevant, while only four (33.3%) mentioned the noise measurement. A user response
especially reflects the perception of the widget in this study: “I do not really care
about the noise level, but knowing the availability of seats is huge! It helps not
ending up in crowded places.” We also questioned the participants about their
general perception of the widget and their potential use. For three participants
(25%), the information delivered by the widget is perceived as really helpful and
useful, while seven participants (58.7%) would rather use it in conjunction with other
types of information (e.g. user reviews, check-ins, video stream, etc.). Especially, one
declared that “it would be nice to have a mix of video, availability of seats and
reviews” and another mentioned that he could “decide whether to go or not (to a
place) with a video, seats and reviews”. Two participants (16.7%) finally stated that
the widget is not helpful in their decision-making process.
To conclude, we observed that people did not have difficulties to apprehend the
concept of real-time and non real-time information offered by places. As one can
expect, in a planning context, people tend to sum up the maximum amount of
information to make their own mental representation of what a place has to offer. We
nevertheless did not predict that our participants would also use this strategy while
they are on the go. While results of experiment A indicate that real-time information
are likely to be used in mobility, those of experiment B also showed that non real-
time information are considered. We explain this contradiction by the fact that
location-based services today do not provide users with dynamic representations of
places but rather foster the use of reviews in every situation. The very positive
feedback we observed about the ambience widget and the video stream we introduced
nevertheless demonstrates the added value of real-time information in the decision
making process. As such type of representation is still novel and disruptive, we
believe people are likely to more actively rely on real-time information, as soon as
they are introduced in well-known location-based services. Moreover, we noticed that
few participants did not fully understand how to use the ambience widget at the first
glance and asked questions about it. This especially highlights the need to design
intelligible representations of information spheres.
4. TOWARDS THE DESIGN OF AN ENVIRONMENT BROWSER
By confronting a small group of lead users to a limited but practical fragment of our
vision, we expected them to project in a reality where places aggregate different types
of information that can be accessed in a variety of situations. The insights we
collected reflect the need for up-to-date information to be communicated in an
effective manner to support the decision-making process. During interviews,
participants especially mentioned the frustrations caused by the lack of information
about places, whether they directly relate to a disappointing experience such as
showing up in a bar at a time when they unexpectedly stopped serving alcoholic
beverages (“I would have wanted to know that before going, instead of wasting my
time.”) or to daily situations such as gathering information about today’s special
meals (“This is especially the type of information that is missing! Information that
often changes is interesting.”). We therefore believe in the opportunity of working
39:14 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
towards the creation of information spheres that augment users’ awareness and
allow the “missing” information to be effectively communicated. Based on the results
of our study, we propose in this section to define the different types of information
that will be delivered by places. We then describe the design choices that we made
regarding the way information spheres are made intelligible to users and finally
illustrate these choices by going through the user interface of an environment
browser that allows users to connect to places and reveal their potentialities.
4.1 Defining information spheres
With information spheres, we aim at creating a virtual communication and
interaction space that mirrors the information already physically available in a place
and aggregates the content that has been shared on the Web about it. Fitting the
boundaries of places, information spheres support users with the possibility to
leverage the most up-to-date information from a single entry point and provide place-
owners with a communication channel that relieves them from the effort of using
multiple platforms. When augmented with information spheres, places are therefore
turned into open platforms that can interoperate with existing services to retrieve
user-generated content but also serve as new sources of information. We especially
identified four types of information that could be collected, generated or designed to
support users’ needs or practices. Their origin, their updatability and the degree of
interaction they offer differ from one type to another, as described in the following.
— Static information relates to any content provided and communicated by place
owners. They can be compared to freely formatted micro-messages or blog posts
aiming at informing users of the place activities (e.g. menu, program, special offers,
arrivals or meals, future events, etc.).
— Dynamic information is made of data generated by all sensors and smarts objects
that are physically present in the place. Such data is processed in real-time and
presented to users in an intelligible way to reflect the most up-to-date characteristics
of places (e.g. noise level, presence of people, type of music broadcasted, occupancy of
the place, etc.).
— User-generated information encompasses all check-in data, reviews, comments,
pictures or videos that are produced by people. They can be retrieved from existing
location-based services and social network platforms, or directly hosted on
information spheres (i.e. assuming that a tool allowing publishing content will be
made available to users).
— Situated services describe the informational interactive systems that are
embedded into information spheres. They rely either on smart objects (e.g. a service
offering to vote for the next song to be played on the music player or the next video
clip to be displayed on the television) or on Web services (e.g. a ticket or table booking
service, a digital music library, games, etc.).
Depending on the technical infrastructure that is deployed into places, and the level
of geolocation precision it offers, we also defined a range of parameters that could be
attributed to the different type of information. A spatial parameter would for
instance allow restricting the delivery of services to the boundaries of the place and
encourage people to consume them in-situation. Such mechanism could also be
leveraged to filter the information that are mainly relevant during the decision-
making process and avoid redundant information to be communicated to users that
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:15
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
are physically present in the place. Another parameter dealing with users’
preferences, expectations or familiarity with the place would also enable a dynamic
adaptation of the content communicated by information spheres. Reward
mechanisms could then be implemented by place-owners (e.g. new customers would
be welcomed in a special way whereas regular ones would be offered special offers).
Finally, a temporal parameter could trigger the disappearance of information after a
certain period of time, avoiding information overload and irrelevance. This could
boost the participation of users who are concerned about the trace they leave (e.g.
content would be shared among friends or deleted after a while).
4.2 Making information spheres intelligible
After defining the different types of information delivered by places, it is necessary to
address the issue of their representation. Static, dynamic, user-generated
information and situated services therefore need to be organized and prioritized to
ensure the consistency of information spheres among places and to avoid information
overload. In this part, we especially describe the choices that we made regarding the
way information spheres are being communicated to users in the form of portals
aggregating widgets. Basic information design guidelines that can be applied to
desktop, tablet or mobile-based user interfaces are proposed in the following.
— Information sphere representation. We chose to design information spheres as
dedicated hyper local websites, whose layout follow a common template that can be
personalized or branded by place-owners. Called “place portals”, they differ from
traditional websites in the way they are structured and reflected according to the
situation. If all portals can be browsed in-situation and off-situation, some
information might, for instance, not be accessible from a remote location.
— Place characteristics representation. Information spheres share common
characteristics such as their ambience, the presence of people and the content that
has been generated by users about them. We therefore propose to consider these
characteristics as “parameters” of the place, which will be presented in the same way
among place portals. Each of them would potentially retrieve user-generated
information, dynamic information provided by smart objects or sensors, as well as
static information posted by users through the place portal. This would depend upon
place-owners’ choices and places’ available infrastructure.
— Information and services representation. All other types of information, related to
place-owners’ generated content (i.e. static information), places’ instrumentation with
smart objects (i.e. dynamic information) and situated services are meant to be
displayed as scrollable graphical widgets. These widgets provide users with a
comprehensive representation of the information, as well as a means of interaction
with specific service or smart objects. In the latter case, a dedicated interface
reflecting the capabilities of the object will be designed.
— Information spheres links representation. Another type of information that we did
not describe as part of the information sphere is the similarity that exists between
two of them. If the aim is not to replace search-based services, we propose to link
places together in order to enable a more serendipitous place-to-place browsing. Such
links could be based on user-generated groups or lists as they already exists in
location-based services, on semantic analysis of information provided by places, or
more complex generation of tags representing the type of activities, ambience or
people that relates to a place.
39:16 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
4.3 Designing the user interface of an environment browser
If information spheres should ultimately be accessible from a variety of terminals, we
chose to focus on the design of a smart phone application that allows scanning places,
connecting to them and revealing their information or services in mobility. Following
the examples of existing location-based services, we designed an environment
browser called Env-B. In the following, we illustrate its user interface and describe
the different views, menus and tools that allow interacting with information spheres.
Fig. 6. List of places (a), map (b) and place portal (c).
Navigation view
When starting the application, users are presented with a tabbed-based user
interface allowing them to quickly display the surrounding and favorites places. The
“places” tab (Figure 6a) reveals a vertically scrollable list of places that can be
filtered and sorted according to several parameters (e.g. category of place, name,
distance, number of people, noise level, number of shared content items). On each list
item, an icon allows to switch to the “map” tab (Figure 6b) and automatically
graphically display the location of the place. Besides the geographical representation,
this mode also offers users with a small description and a way to quickly slide from
one place to another by touching the “previous” and “next” icons. On every tab,
additional tools are provided on the top-bar of the interface to refresh the content (i.e.
based on users’ location) or to “teleport” to another location (i.e. by typing another
address). A sliding search panel also allows finding a specific place that is not easily
accessible from few taps on the screen.
Place portal view
After selecting a place from a specific tab, a portal-based interface representing the
information sphere is presented to users. Figures 7a and 7b depict the place portal of
a coffee shop containing five widgets. Each of them appears as cards that can be
scrolled with a horizontal “slide in” gesture. Small icons highlight the active widget
and the total number of widgets, while a vertical sliding panel allows revealing a list
of all of them. Called the “place index” (Figure 7c), it allows to have a clear view of all
available widgets and to know which ones are only accessible in-situation. A bottom
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:17
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
menu bar triggering the display of a popup window allows users to access the
presence, ambience and user activity representations of the place (Figure 8a, 8b, 8c),
to browse existing lists that encompass the place, and to share information such as
check-ins, reviews or pictures (Figure 6c). By touching the top right corner icon, users
can finally bookmark the place portal or add it to a list, while the left one serve as a
back button.
Fig. 7. Two widgets (a, b) and the place index (c).
Fig. 8. Presence (a), ambience (b) and user activity (c) popups.
39:18 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
5. PROTOTYPE IMPLEMENTATION
A number of research projects aiming at creating mobile user interfaces for
discovering and interacting with smart objects, separately or along with the
middleware enabling the creation of smart environments or more generally of an
Internet of Things, have been conducted over the past years [Newman et al. 2002;
Roduner 2010; Kortuem et al. 2010; Garcia Macias et al. 2011]. These efforts vary
significantly in focus, introducing new interaction paradigms, user interfaces or
technical components that support novel user experiences or new applications. Our
work encompasses these three aspects, to provide users with a global application
allowing them to consult information spheres of surrounding or remote places, and to
interact with their resources (i.e. static or dynamic information, situated services,
smart objects, etc.). To prototype the environment browser, we chose to rely as much
as possible on Web principles and technologies, abiding to the “Web of Things”
approach [Guinard et al. 2011; Zeng et al. 2011; Pfisterer et al. 2011]. This approach,
which refers to the more general corpus of Internet of Things research, leverages
technical standards of the Web (e.g. in our case HTML and Javascript for
presentation, RESTful Web Services and APIs for resource access and exposition, and
OWL or RDF-a for semantic descriptions) to propose Web-savvy developers with
simple means of creating Internet of Things, and more generally Ubiquitous
Computing, applications. We indeed argue that exposing the resources of a place
facilitates the way they can be used or reused to create services, and that Web
technologies are today the de facto standard for many application developers. In this
section, we present the different components of the system we developed, before
describing the implementation of the Env-B client on the Android mobile platform.
Fig. 9. Overview of the Place Enabler framework architecture
Note: 1 Retrieving relevant information spheres for the given
context. 2 Retrieving place portal from a given place enabler.
3 Interacting with place resources through widgets.
Place
enabler
Place resolver
Env.
Browser
1
3
2Place
enabler
Place resources
Smart objects
Situated services
People & ambiance
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:19
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
5.1 Exposing resources of the environment to users and applications via the Place Enabler
Framework
Our solution to represent resources of a given environment to users and applications
relies on the concept of Web-based Virtual Objects (VO), their containers or Place
Enablers and the Place Resolver component, which together form the Place Enabler
Framework (Fig 7) [Boussard et al. 2011]. VOs are software agents representing the
corresponding environment resource to users, through a digital representation, and
to other software components, through a processable description and a set of RESTful
programming interfaces. For example, a connected lamp (Figure 10) is represented
through a Lamp VO exposing 1) a user representation for visualization and
interaction (described in HTML) purposes, 2) RESTful APIs that advertise
functionalities (e.g. change on/off state of the object) and retrieve the HTML
representation and 3) an OWL-based semantic description that supports the
reasoning for the composition or mash-up of this lamp with other VOs or services (not
discussed in this paper, please refer to [Boussard et al. 2011]). In the case of Virtual
Objects representing a smart object, this approach allows to visualize, interact and
compose this resource with others in new services, hiding the inherent complexity of
how this object is connected and interfaced below the VO Web API. In particular it
allows abstracting from the variety of technical realizations of smart objects and
supporting networking mechanisms, by providing a general view to other components
of the system. Virtual Objects can also be used to represent data resources (e.g. an
information feed attached to a given environment) or a more elaborated situated
service using the same general framework.
Fig. 10. Facets of a Web-based Virtual Object.
Virtual Objects live in a software component called the Place Enabler, which
typically represents a given place (e.g. home, a restaurant, etc). This software
component aggregates the VOs of the corresponding physical place, and typically can
be placed in the network topology next to a network aggregation point (access
point/network technology gateway) to enable the connection and the exposition of
local physical resources of varying access network technologies (e.g. a LAN public
display or a Zigbee sensor). Place Enablers register themselves at the Place Resolver
component, to allow for their discovery. Figure 11 provides a simplified UML class
diagram of the overall framework showing the main methods and relationships
between the components of the Place Enabler Framework.
API
(REST)
Semantic
description
(OWL)
User representation
and display (HTML)
39:20 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
Fig. 11. UML Class Diagram of the Place Enabler Framework.
Note: All these attributes are implemented as REST attributes. Specific attributes may be added and used
for VOs (e.g. state) PEs (e.g; address) or both (e.g. iconURL). Only the most important attributes are
represented in the figure. Private methods, specific method signatures as well as additional support
systems (e.g. user support & database, etc.) were left out for readability.
Virtual Object
+Description: VO_Description
+VO_URL
+getVODescription()
+getAttribute(attr)
+setAttribute(value, attr)
+deleteAttribute(attr)
Place Enabler
+Description: PE_Description
+PE_URL
+getVOList()
+ping()
+getDescription()
+ getAm biance()
+ getPres ence()
+ getMessages( )
+ getGrou ps()
#addVO()
#removeVO()
isHostedby
10..*
Place Resolver
+getPlaceEnablers(user, location, locationFilter, radius)
#registerPE(PE: Place Enabler)
#unregisterPE(PE: Place Enabler)
isRegisteredAt
1
0..*
+lterPE s()
hasFilter
1..*
1
FemtoCellFilte r GeoLocFilter AttributeFilter
VO_Description
+id
+name
+type
+version
+htmltemplate
PE_Description
+name
+id
+location
+category
1
1
1
1
Resolver Filter
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:21
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
The Place Enabler is implemented using the OSGi1 framework together with Jetty2
and Jersey3 (to provide VO and Place Enabler interfaces as RESTful web services).
This allows to take advantage of the resulting high degree of modularity: VOs are
embodied as OSGi bundles, which simplifies their provisioning (including at runtime
when a new object is brought in the environment), and other bundles provide utility
functions such as persistency, access control or eventing (publish-subscribe pattern).
Place Enablers typically provide public interfaces to retrieve the list of VOs for the
considered environment or to retrieve general information. VOs can respond to
“discovery requests” returning an HTML, XML or JSON representation (depending
on the HTTP header of request), or to commands on specific URL endpoints (e.g. an
HTTP PUT on http://placeenablerURL/lamp/state to turn on a lamp). Besides
providing access to VOs of the relevant place, Place Enablers can provide static
information about the related place such as location, ownership, or any other type
defined by the place owner as attribute. Finally, Place Enablers have been adapted
from the original VO-Gateway Framework depicted in [Boussard et al. 2011] to
provide dedicated methods for supporting dynamic and user generated information
(ambiance, presence, messages) and the list of groups the place is part of. It is worth
noticing that these different methods implementation are specific to the relevant
Place Enabler and may rely on a separate VO – this allows for example a Webcam
VO to be directly bound to the ambiance view of the relevant Place Portal. Similarly,
a dedicated VO may be in charge of aggregating check-ins or user generated content
on a variety of existing location-based web services (e.g. Yelp, Foursquare, etc) to be
displayed in the presence or user activity views.
The last component of the Place Enabler Framework is the Place Resolver, which is
responsible for maintaining an updated list of living Place Enablers, and returning
requesting Env-B clients with a list of relevant places, depending on specified filters
such as the reported user location and a geographical radius. To enable this, a
number of possible geolocation techniques are usable, from self-declarative
approaches (GPS coordinates-based check-in retrieved from the user terminal or via
a widely deployed service like Foursquare4), to infrastructure-derived ones such as
femto-cell geolocation. Different Resolver Filters thus allow accommodating the
different location mechanisms, or even provide additional filtering behaviour (e.g.
based on the value of a specific attribute value of PEs), as described in Fig 9. The
Place Resolver is implemented using the OSGi framework, with dedicated bundles
for each available filter, and to manage persistency and events.
As our goal is to make Env-B a mobile browser capable of identifying if it is being
used within the viewed place or from a remote location, it is therefore needed that 1)
the system distinguishes between the two situations and 2) depending on this
situation, Place Enablers may present differently or grant different access to
resources of the place, leading to different user experience of the corresponding place
portals. Different solutions are possible to allow this. For example, it is possible to
rely on self-declarative methods, i.e. having the Env-B client publishing its location
to PEs or even to a centralized component that PEs would interrogate (which can be
1 OSGi, http://www.osgi.org
2 Jetty, http://jetty.codehaus.org
3 Jersey, http://jersey.java.net/
4 http://www.foursquare.com
39:22 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
an existing services like e.g. Foursquare). It is also possible to use infrastructure-
derived location. We have typically experimented with a femto-cell based location
mechanism, where location of users is gathered and published by femto-cells (e.g.
spread on a building floor) to a central repository. Upon reception of a request from
the Env-B client (e.g. getVOList() or getVODescription()), the Place Enabler can
decide which content to reply (e.g. to adapt the list of accessible VOs to the user
profile, or to present a different HTML representation of a given VO depending on
the user profile or actual location).
5.2 Environment Browser client and interactions
The mobile Env-B client is implemented following the design choices depicted in
section 4 as an Android 2.3 application (Figures 6, 7 and 8 show actual screenshots of
the application), and allows three activities:
— Visualize on a map (possibly centered on the user location) the surrounding
information spheres. This is done using the MapView UI component of Android SDK,
and uses a list of relevant information spheres based on a request to the Place
Resolver component containing the user identity and location. Such interaction is
mediated by an internal component called Place-agent: by using location information
and user preference information, this module communicates with the Place Resolver
to get the relevant information spheres around the selected location. Alternatively, a
list view can be used (Figures 6 and 6b).
— Display and interact with a selected place portal. After selecting an information
sphere from the map or list, the client offers to access resources depending on their
visibility (as depicted on Figure 7). The Android SDK WebView component is
leveraged to render the HTML representation of a Virtual Object within the defined
boundaries of the associated portal widget, offering visualization and interaction with
the VO. All the information aggregated in the place portal view is finally retrieved
from the Place Enabler and associated Virtual Objects by the Resources-agent
component.
— Managing users’ preferences. The application maintains a user profile (user
interests and identifier) as well as a list of bookmarked place portals. This
information can be used to filter the widgets that are displayed on portals and to
leverage social network mechanisms (e.g. sharing lists of places).
Figure 12 shows an overview of the internal architecture of the Env-B client,
separated in the Communication Component layer (responsible for communication
with the Place Enabler/VO Framework); Core Service Modules (internal
representation/handling of information) and UI views.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:23
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
Fig. 12. Components of the Env-B mobile client.
The following sequence chart diagram highlights the main interactions between the
Env-B client and the Place Enabler Framework components for a set of basic
interactions.
Place
Agent
VO
Agent Rendering User
Profile
Map View
UI Views
Core Services Modules
Communication Component
List View Bookmark
View
Place
Portal View
Bookmark
VO Framework Access Proxy (Rest APIS + JSON/XML/HTML)
VO Framework
39:24 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
Fig. 13. Sequence chart of the exchanges between Env-B client and Place Enabler Framework
components
Note: Example of interactions when 1) requesting a list of relevant places 2) selecting and displaying a
specific place portal from list and 3) selecting the ambiance view.
In the following section, we describe and discuss an evaluation of the implemented
prototype in a controlled environment, describing the technical setup of the
components described above together with the evaluation process and results.
: EnvB Client : Place Resolver Home : Place Enabler Webcam : Virtual O bj ect
1 : displayPEList()
2 : getPlaceEnablers()
3 : List of PEs
4 : displayPlacePortal()
5 : getVOList()
6 : List of VOs
7 : getVODescription()
8 : description
9 : getAttribute()
10 : htmlView
11 : displayPEAmbiance()
12 : getAmbiance()
13 : ambianceInfo
User launches EnvB client on his mobile
User selects the Home Place Portal
User requests the Ambiance view for Home
getAmbiance() method may
in turn invoke specific methods
on e.g. the Webcam VO
User may activate a control on the Webcam VO htmlview
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:25
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
6. EVALUATION AND DISCUSSION
Evaluating a Ubiquitous Computing system is acknowledged by the research
community as a challenging issue, due to the tremendous efforts that are required to
deploy and maintain technical assets in a real life context [Carter and Mankoff 2004].
Furthermore, such applications are often designed to support specific tasks and to be
integrated in specific contexts, which makes even in-lab evaluation very difficult to
conduct. While working towards realizing the vision of information spheres, we
created an infrastructure and a mobile client that open up the opportunity for people
to retrieve and interact with resources of places. We described a vision of a future
that is not too far from what people already experience with existing services, but
requires a spread of connected objects and sensors and the creation of situated
services. This makes the appreciation of our production rather difficult to make and
brings us to consider two essential questions: 1) at what scale should people
experiment our system? 2) Which metrics should be used to evaluate its success? As
many Ubiquitous Computing applications, the value of our project is tied to the wide
scale availability of both technical systems and information. It promises a new user
experience, in which Env-B takes a central role, and which would ultimately bring
people to adopt our technology and embrace our vision. To evaluate this experience,
we therefore need to work towards re-creating an ecosystem of places as close as
possible of our vision, and investigate the usefulness, usability and desirability of our
system. As a first attempt to evaluate our work, taking into account that a
longitudinal study would be necessary to obtain more robust results, we conducted a
task-based test in our lab. In this section, we describe the set up we built to conduct
our experiment, the methods we chose to use and the results we gathered, before
discussing the limits of our approach.
6.1 Creating an ecosystem to evaluate the user experience
User experience is shaped by all design choices that allow people to make a
representation, or a model, of what a system can do. It is a subtle balance between
the relevancy of functionalities, the clarity they are communicated with, and the
attractiveness they generate. Usefulness is of course at the foundation of user
experience: a good product or tool should help people to accomplish their goals, to
support their activities or serve their needs. Usability somehow “exposes” or “reveals”
the usefulness of a system, by making sure people will be able to achieve their goals.
Desirability refers to more emotional aspects, such as user engagement,
empowerment or trust, as well as to the ease of use. In our work, from the user
perspective, Env-B reflects the potential and the possibilities of the system. Its user
interface presents organized information, communicates the functionalities, and is
designed to be easy to use and attractive. All the design choices that we made have a
direct impact on the way people will perceive and use the system. They shape the
user experience. To investigate the usability, usefulness and desirability of our
system, it is therefore necessary that people manipulate, test and interact with the
system through the mobile client we implemented in a real-life context.
As deploying our system in the city would have been very consuming in terms of time
and resources, we tried to realize the vision of information spheres within our lab. In
order to create a compelling and convincing ecosystem, we identified surrounding
places at our premises that could benefit from an information sphere, whether to
expose information, services or objects. Except for our smart home testbed, we did not
recreate places, but relied on existing ones like the coffee spot, the canteen, a meeting
39:26 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
room or a bus stop. By targeting our colleagues as potential testers and real users of
our system, it was necessary to consider their everyday activities to decide which
objects to connect, sensors to instrument or services to deliver. We tried to find a
balance between the technical effort required and the perceived usefulness of the
information we could deliver through our system.
We therefore deployed a Place Resolver and five Place Enablers on an open network
in our premises. For each place, we designed a place portal that aggregates widgets
and represents the presence, ambience and user activity, as defined in section 4. A
number of widgets, ranging from simple representations of a status (e.g. availability
of a room or a coffee machine) to interactive widgets (e.g. allowing to turn on/off a
lamp, book a meeting room) were implemented. Depending on the nature the place,
the ambience was communicated in different ways, whether with a real time/average
representation of the noise level or a video stream. Table IV summarizes these
specific variations and describes the different resources exposed by place portals.
Table IV. Creating an ecosystem to evaluate the user experience.
Places
Widgets
Presence
Ambience
User
activity
Lists
Noise
level
Video
stream
Home
Mailbox
Lamp
X
-
-
X
3
Meeting Room
Room availability
Today’s schedule
Tomorrow’s schedule
X
Average
X
3
Canteen
Today’s specials
Charge your account
X
Real
time
X
X
3
Coffee spot
Coffee machine
Bell Labs screen
Post a message on Bell
Labs screen
X
Real
time
-
X
2
Bus stop
Information display
Schedule
X
-
X
X
1
Most widgets were relying on a live and running infrastructure. The mailbox and
lamp are part of the “smart home” test bed that has been deployed in our labs. We
also built upon one of our instrumented meeting rooms, which connects to our
framework, to expose its availability in real time. In the same way, we created
widgets that draw connections to the “Bell Labs screen” service (i.e. a small network
of information screens and a social networking platform) and expose its content
through the Coffee Spot place. Insofar as we could not access the data of the coffee
machine and the real bus stop information display, we created a wizard-of-oz system
that allows changing the status or information of such devices. Due to the lack of
flexible APIs, we finally chose to create simulated widgets instead of interoperating
with the existing meeting room booking system that is available on our corporate
intranet and the canteen services that are provided by the catering company. For the
presence, ambience, user activity and lists views, we recorded video footage of the
real places, created animated representation of the noise level and wrote dedicated
content for each place. We put the effort into making the messages plausible and
familiar (e.g. using names of colleagues or highlighting common issues or topics), as
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:27
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
we imagine people would use it, so participants would be able to reflect upon it and
generate feedback.
6.2 User evaluation
Based on the ecosystem we created, we imagined a number of tasks that would bring
people to navigate on the different place portals and experience the possibilities of
information spheres. These tasks steered the whole evaluation process, they can be
considered as fragments of experience, which, when combined, would allow people to
get a better idea of what the system offers and provide us with a situation to
evaluate. Presented to participants as questions, depicted on Table V, they cover all
aspects of Env-B (i.e. from widgets to presence, ambience, user activity and list
views) and vary in their level of complexity (i.e. retrieving an explicit information,
interpreting information to make a personal decision and interacting with an object
or service).
Usability
A task-based questionnaire allowed us to measure the effectiveness (i.e. the ability of
users to complete the tasks) and the efficiency (i.e. the level of resource consumed in
performing tasks) of our system. By calculating the number of tasks achieved by
participants and the required amount of time for each of them, we were therefore
able evaluate the usability in a quantitative way. We chose to combine this method
with a widely used survey scale, the System Usability Scale (SUS)5, to obtain more
general results. Composed of 10 statements that are scored on a 5-point scale of
strength of agreement, the SUS provides a single score (out of 100) that can be used
for relative judgment (e.g. comparing alternatives or iterations) [Brooke 1996]. It is
considered as a highly robust and versatile tool. It has proven to be technology
agnostic and very easy to use by both participants and testers.
Table V. Task-based questionnaire for usability evaluation.
Questions
Place
Resource
Type of
questiona
1
Can you get a caffe latte from the coffee
machine at the 2nd floor?
Coffee spot
Connect
object
Reading
2
Who posted the last tweet on the Bell Labs
screen of the 2nd floor?
Coffee spot
Connected
object
Reading
3
Would you go for a cup of coffee if you were
looking for a very quiet break?
Coffee spot
Ambience
Choice
4
Who is looking for
a colleague to play tennis with?
Coffee spot
User activity
Reading
5
What would you like to eat at the “Villarceaux
Canteen” today?
Canteen
Situated
service
Choice
6
Are many people having lunch at the
“Villarceaux Canteen” right now?
Canteen
Ambience
Interpretation
7
Has Dominique D. already left the building to
have lunch?
Canteen
Presence
Reading
8
Book a time slot of the “2E50” meeting room for
tomorrow.
Meeting
room
Situated
service
Action
5 Developed by John Brooke in 1996, as part of the usability-engineering program in integrated office
systems development at Digital Equipment Co Ltd., Reading, United Kingdom.
39:28 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
9
Is the “2E50” meeting room usually quiet on
Fridays?
Meeting
room
Ambience
Reading
10
How many videoconference rooms are available
in Villarceaux?
Meeting
room
List
Reading
11
Did you forget to turn off your lamp before
leaving home?
Home
Connected
object
Reading
12
Did David A. attend your party 2 days ago?
Home
Presence or
user activity
Reading
13
When is the next bus heading to Massy arriving
at the bus stop?
Bus stop
Connected
object
Reading
14
Are there many people waiting at the bus stop
right now?
Bus stop
Ambience
Interpretation
15
Is Cedric traveling with Pierrick in the bus?
Bus stop
Presence
Reading
a Depending on the questions, participant had to read and report a specific information, make a choice based on
the available information or interpret a situation by leveraging different types of information.
Usefulness and desirability
Measuring the usefulness and the desirability of our system is a more complicated
task, especially during such a short period of test. Such dimensions are very
subjective and their appreciation can change over time depending on users’ needs
and social context. If a longitudinal study would be necessary to assess our insights,
we tried to address these aspects with qualitative methods. We created a satisfaction
questionnaire based on a five-point Likert scale (questions are detailed on Table VI)
that aims at capturing the most-interesting features or information types, and the
global attractiveness Env-B. This questionnaire also served as an introduction to a
short interview where we gathered feedback and discussed the way such tool would
fit in the daily life of participants.
Participants
Twenty participants (mean age=31.3; standard deviation=7.28) were recruited in our
lab. The panel was composed of sixteen men and four women working as researcher,
Ph.D. candidate or intern in various research domains (i.e. multimedia, security,
human-computer interaction). All of them were sharing the same facilities and
taking their lunch break at the same place. 80% of them owned a smart phone, most
often running on Android (63%) or iOS (25%), while the other 20% used a feature
phone that does not allow to access to an application market. We consider their level
of proficiency as intermediate. They indeed demonstrated the skills to interact with
touch screens through pinch or pan gestures, to navigate on map-based, list-based or
tab-based representations and to trigger buttons or type text. All of them were
familiar with location-based services but did not use them on a regular basis (i.e. only
30% declared to used them “occasionally”, others not to use them at all). Unlike the
previous panel (section 3.2), these participants therefore cannot be considered as lead
users having an extensive practice of mobile applications, but rather as regular users
with an extra background on computer science (i.e. they are aware of new trends or
uses, but are not early adopters). Their level of competency is therefore sufficient to
accomplish the test.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:29
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
Procedure
Participants were introduced to the research project, drawing upon a link with
existing location-based services in order to explain how we designed Env-B as a
medium to represent information spheres. We then gave them an overview on the
different parts of the application by navigating one of the place (i.e. the canteen) and
briefly explained the icons or gestures that allow browsing the content of a place
portal. Participants were then asked to use Env-B to answer the questionnaire
detailed on Table V. In order to avoid biases in the measure of efficiency (i.e.
participants would take less time to perform actions after completing a similar task),
we created four versions of the questionnaire where blocks of questions are ordered
in different manners. Participants were free to navigate on the different place portals
to find the right information. An answer needed to be written down to facilitate the
evaluation of success. Each task was timed (i.e. from the moment the participant
finished reading the question to the moment she wrote the answer on the
questionnaire). Misunderstandings or remarkable events were observed and noted.
After the test, participants filled out the SUS and satisfaction questionnaires and we
finally opened up the discussion to collect their feedback or insights on the overall
user experience.
Result analysis
To evaluate the effectiveness of our system, we considered that a task is achieved
when a participant correctly retrieved the information we wanted him to look for,
interpreted it and reported the expected answer in the questionnaire. We distinguish
two kinds of non-achieved tasks: failures, where users were not able to provide any
answer at all (i.e. leading the participant to skip the question after a certain period of
time), and mistakes, where users provided an answer that we do not consider as
correct or complete (i.e. due to a misunderstanding or misinterpretation of the
retrieved information or the question). This difference is important to make in the
measure of efficiency. Indeed, in cases of failures, we could not use the amount of
time that was observed without biasing the results. We therefore decided to balance
them by calculating an average value that is used for all failures (i.e. we multiplied
the standard deviation per two and added it to the mean value). Mistakes were
considered in the same way as achieved tasks, since we were able to track the time
needed by participant to provide an answer, whether it is consider as correct or not.
6.3 Results
Results of the experiment are presented through two different parts: the evaluation
of usability, which is based on the task-based questionnaire and the SUS, and the
evaluation of the usefulness and desirability, which relates to the satisfaction
questionnaire and the interviews we conducted after the test.
Usability
On average, participants achieved 87% of tasks (SD=8.5) in a cumulated time of
approximately six minutes (M=360.9 seconds, SD=124.52). Out of the fifteen
questions, participants were therefore unable to complete an average of two tasks
(M=1.95, SD=1.27), which can be considered as a good performance. In 25.6% of
cases, participants failed to provide an answer, while in the other 74.4% they made a
mistake. To investigate these results, we created graphs representing effectiveness
and efficiency both per task (Figure 14) and types of task (Figure 15). While the first
one allows identifying the questions that causes problems, the second one provides us
39:30 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
with a more general view that can be used to evaluate specific parts of Env-B user
interface.
Fig. 14. Effectiveness and efficiency of performance for each task.
Fig. 15. Effectiveness and efficiency of performance for each type of task.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:31
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
Tasks dealing with objects representation are the most effective (96.66% of
achievements) and efficient (M=15.6, SD= 7.71). Those relating with situated services
also show a very good success rate (98.33%) but are more time-consuming (M=27.06,
SD=18). This can be explained by the fact that more information is presented to the
users (i.e. augmenting the cognitive charge) and that interactions were required in
some tasks (e.g. when booking a room, users had to type text on the smartphone
virtual keyboard). While only 15% of participants failed to complete the single task
related with the lists of places, it took a while for most of them to answer the
question (M=24.45, SD=22.29). Tasks covering the presence and user activity
representation were easily completed (92.5%) in a reasonable amount of time
(M=25.22, SD=19,80 and M=24.45, SD=22.29). Surprisingly, only 66.25% of
participants achieved the tasks related with the ambience representation. This
highlights an interesting issue that we will discuss in the following part. Apart from
this specific issue, which relates with questions 6 and 14, results of the evaluation
can be considered as good: on average, participant successfully performed each task
in less than 25 seconds (M=24.06, SD=16.43).
Results of the System Usability Scale also assess the usability of Env-B. With an
average SUS score of 74.5 (SD=9.5) out of 100, we argue that our system was well
perceived by participants. Research conducted on usability judgment brought
professionals to consider scores above 70 as “good” and above 80 as “excellent”
[Bangor et al. 2008]. During our test, 80% of participants scored the system above 70.
30% often gave a score above 80, as depicted on Figure 16.
Fig. 16. SUS score distribution per participant.
Usefulness and desirability
Results of the satisfaction questionnaire (Table VI) give us a better understanding of
the way participants perceived the experience. Figure 17 shows that 60% of questions
scored above the scale level 4, which signifies agreement with a statement. The
average score, calculated from all statements, is also very positive (M=4.03,
SD=0.43). On average, users agreed to say that the system would support (M=3.9,
SD=0.71) their activities. According to them, Env-B is very well tailored for
interactions with connected objects (M=4.45, SD=0.60) and information retrieval
(M=4.3, SD=0.73). Navigation principles were also well perceived (M=4.2, SD=0.76)
and participants expressed the desire to extend the ecosystem to others places
(M=4.5 SD=0.82). Ambience representations were also well received, but we observed
high variations (SD>1), while presence, user activity and list representations left
people more skeptical. We managed to get more feedback about this during the short
interviews.
39:32 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
Table VI. Satisfaction questionnaire.
Statements
1
I liked the way I can connect to different places
through portals.
2
The application provided me with information about
the place that is difficult to retrieve at a distance.
3
The comments or contents posted by people bring
value to the place portals.
4
Displaying the noise level of the place is useful for
certain types of place.
5
Streaming a video of the environment is useful for
certain types of place.
6
I found interesting to aggregate foursquare check-
ins.
7
I would like to discover new places by browsing the
lists created by people.
8
The application allows me to interact with remote
objects in an easy way.
9
This application would help me to support my
decisions in everyday life.
10
I would like more places to be connected and
accessible on the application.
Fig. 17. Mean levels of agreement with the items of the satisfaction questionnaire (Table VI). “5” =
“Strongly agree”, “4” = “Agree, “3” = “Neither agree nor disagree”, “2” = “Disagree”, “1” = Strongly disagree”.
(Error bars show the standard error of the mean.)
During the discussion we had with participants at the end of the test, all of them
mentioned the relevancy of the information provided by place portals. Participant 9
for instance told us that he “expected to find this kind of information in these places”,
while participant 8 stated: “They are all places we visit everyday, we need this
information everyday”. It reveals that we created an ecosystem that “makes sense”
(participant 12) for the testers. Among all places, the canteen and the bus stop
generated most feedback. People are indeed interested in retrieving information that
usually cannot be retrieved at a distance or retrieving it in a more efficient way.
Participant 3 for example declared that he “can find the same information
elsewhere”, but with Env-B he can access it “directly”. Participant 4 also stresses out
the fact that information is centralized: “we often navigate from one application to
another without finding the information we need. It is interesting to put all this
information in one place”. Resources of place portals help to “make a decision” or
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:33
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
“decide when to go” (participants 1, 2 and 11) and would allow avoiding situations
where they “waste time” (participant 18).
As stressed out by the results of the satisfaction questionnaire, testers had mixed
feelings about the presence, ambience and user activity representations. Especially,
non location-based service users considered foursquare check-ins and messages as
“too limited” (participant 9), “gadget” (participant 4) or even “dull” (participant 12).
The ambience representations raised more enthusiasm among the panel, but in many
cases people preferred one type (i.e. the video stream or the noise level graph) to
another. For instance, participant 16 really liked the fact that “you see the place with
your own eyes”, judging that “it is easier to capture the ambience with a video” while
participant 5 highlighted the practical issue of placing the camera and stressed out
that “you do not always see if there many people or not, the sound level is just fine”.
In the same way, participant 9 declared that he liked the real time and average
variations of the noise level graphs because “you immediately see the difference”,
while participant 3 said this feature was “his least favorite” because it was “too
complicated and could be replaced by a 3-state representation (calm, ambient, loud)”.
Participants 1 and 14 finally raised concerns regarding the ethical and privacy
aspects of streaming a video of a public place.
All participants gave very positive feedback about Env-B and showed a strong
interest in the system. The mobile client is considered as “easy to use” and
“intuitive”. Participant 8 mentioned that he “did not have to look for the different
buttons for a long time” and “always knew on what place he was navigating”, while
participant 11 declared he “quickly understood how it works” and that it “does not
take a long time to get used to the application”. Except participant 12, who would
prefer to “consume these information on a public display or desktop rather than on a
mobile phone”, all others declared that they were willing to use Env-B not only at the
office but also in their personal life. Most of them had ideas of other place portals
they would like to use (e.g. subway stations, specific roads, restaurants, bars, theater,
medical offices, gym or golf clubs, supermarkets) and that they could make a habit of
using Env-B, even if “the value reveals itself in the practice” (participant 3).
6.4 Discussion
Results of the experiment showed that the overall user experience of our system is
satisfying. Participants managed to browse places portals with Env-B, were able to
retrieve or interact with the resources of places, and highlighted the qualities of the
user interface design. However, people sometimes misunderstood the ambience
representations. Before concluding this paper, we propose to examine why people did
not use these representations as much as we expected and to discuss general
considerations deploying information spheres on a larger scale.
Getting used to real time information
When analyzing the results of our test, we were surprised by the repartition of non-
achieved tasks. Indeed, 69.23% of them occurred while asking questions about the
ambience representation (especially during tasks 9 and 14). Statistics showed that in
88.88% of cases, participants did not fail to complete the task, but made a mistake.
Instead of looking for the noise level graph and the video stream to answer the
question “are there many people?”, 75% of them relied on the presence information to
make their opinion and often reported the number of Foursquare check-ins. We
39:34 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
therefore considered these answers as mistakes. When facing the same type of task
for the second time, 40% of them had learned to access the ambience information (i.e.
at some point they ended up exploring the ambience view) and relied on the latter to
answer the questionnaire, leading us to report their task as achieved. The others
kept making their decision based on the presence information. The way we mixed the
tasks through the different versions of the questionnaire explain why this
phenomenon is not reflected by Figure 14.
When, later during the interview, we asked participants why they did not use the
ambience representations (at first or at all), participant replied that they “forgot”
such information was available in Env-B. We assume that the conditions of our
experiment did not allow participants to apprehend such a new type of information
as quickly as we imagined. They instead leveraged a well-known type of information,
which was also communicated in a more direct way (i.e. the total number of check-ins
is displayed in the tab, close to the presence icon). Participant 4 confirmed this
hypothesis by declaring, “I did not pay attention to the webcam, I could have used it
to check if the place was crowded though… I do not have the habit of using it, I am
not yet familiar with the application!”. Participant 9 also said he “did not see if the
video was in real time or not”, and “thought it was a picture”, referring to the video
stream of the uncrowded bus stop. This reveals a more general issue: participants
were not used to deal with such type representations. Participants 1, 2 and 11
“thought” the latter were provided in real time and up-to-date, but still raised the
question during the interviews. Participant 7 was more confused. He said he “did not
have the feeling it was in real time because some places display the average noise
level”. Such insights therefore stresses out the need to integrate a short tutorial or
some tool tips in the user interface, as Facebook or Foursquare did, to help users
discover and leverage all the features of Env-B.
Remaining issues
A number of technical issues remain to be addressed to make the whole information
spheres concept alive in the real world. But besides technical centered
considerations, the whole user acceptance of such an approach is bound to the ease of
1) setting up an information sphere for place owners, and 2) creating a service
ecosystem where developers can create new services that make use of information
spheres resources. Providing tools and mechanisms enabling place owners to easily
build information spheres for their place can ease the set up phase. They should
ideally find and connect smart objects to the information sphere and maintain the
system up-to-date. The challenges here clearly lie in the heterogeneity of the
considered field (i.e. various network connectivity, different object providers, etc.).
Depending on the openness of the system we create, developers will be able to create
situated services that can instantiated into information spheres. By relying largely
on Web principles, we aim at facilitating this kind of broad adoption by communities
of developers. Engaging people to create a service ecosystem is a difficult task, since
it is also bound to the availability of such systems. Finally, it is interesting to note
that our approach also enables to scout new types of situated services that can be
brought in by users, instead of being pre-defined by the place owner, assuming the
latter will allow such behavior. We are currently investigating these hypotheses
leveraging the semantic descriptions associated with Virtual Objects.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:35
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
7. CONCLUSION
In this paper, we presented a multi-disciplinary approach that led psychologists,
designers and computer scientists to explore a vision of the city where information
and services are virtually situated, localized or embodied into physical places.
Bridging concepts from location-based services and smart environments, this
practical research especially aimed at investigating the potential of shifting from a
paradigm where users query a search-based application, to another where they
directly interrogate a place. From a user point of view, the perspective of creating
hyper local Web fragments that fit the boundaries of a location is considered as
relevant. The study we conducted with a small group of location-based services lead
users indeed revealed their difficulties to gather up-to-date information about places
and confirmed the importance of leveraging different types content in the decision-
making process. They especially highlighted the need to confront static information
such as user-generated reviews with more dynamic ones, such as the ambience
widget or the video stream that we designed to reflect the real-time dimension of
places.
Drawing upon these insights, we defined information spheres as communication and
interaction spaces that allow 1) exposing places’ characteristics on the Web to
facilitate information retrieval and communication, 2) leveraging sensors or smart
objects to automatically generate real-time information about the place, 3)
instantiating a variety of services that can be experienced in or off-situation by users.
Places therefore serve as open platforms for information, objects and services, which
can be accessed by users but also by other Web services. Such an approach opens up
the possibility for developers to release or sell applications that can be instantiated
into places by places owners, depending on the way places have been instrumented
(i.e. some services may rely on specific kinds of smart objects or sensors); and for
place-owners to promote their business or location. To illustrate this potential, we
designed and implemented a mobile environment browser called Env-B that makes
information spheres intelligible and accessible to users. People can make use of this
application to reveal and interact with the different resources (information, objects,
services) of a place, which are exposed through a Web of Things framework, in a
planning or mobility context. The evaluation of our system, through a task-based
questionnaire that led users to experience an ecosystem of places, allowed us to
assess the usability of Env-B and the perceived usefulness of information spheres.
Results of the experiment showed that people are willing to use our system in their
daily life despite the difficulties some of them had to apprehend real time
information. This highlights the novelty of representations such as the ambience
view and stresses out the needs to provide users with knowledge or training in the
interaction with interactive and intelligent systems.
With this work, we hope to contribute to the Ubiquitous Computing community by
illustrating the role of smart objects in the city, and by illustrating the importance of
co-creating the technical systems and tools with users. If people lack a reference
point for judging usefulness and usability of emerging technologies such as location-
based services [Junglas 2007], we argue that they can provide researchers with a
number of insights that will feed the design process. Evaluating the potential of
smarts objects and environments, especially in the urban context, is a difficult task,
since not many products have hit the market. Longitudinal studies aiming to
experiment our system and tool in a real-life context would therefore be necessary to
39:36 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
ensure the adoption of information spheres. Another research opportunity would be
to explore the vision from the place-owner perspective. If communication experts
showed the potential of location-based services in the success of marketing strategies
of small businesses [Salt 2011], we certainly can benefit from owners’ point of view to
improve our system and provide them with administration tools. Finally, we believe
there is an opportunity to integrate the notion of activity to create more intelligent
systems. Users are indeed likely to want to consume information or services that are
relevant with the type of activity they perform in a place and its frequency [Jones
and Grandhi 2005]. This therefore opens up new challenges for inferring, monitoring
and communicating place-supported activities, not only from sensors but also with
smart objects.
8. ACKNOWLEDGEMENT
Authors would like to thank Olivier Le Berre for his contribution to this work, Fahim
Kawsar, Arnaud Gonguet and Florentin Rodio for their support, and all users who
helped shaping and evaluating the user experience of Env-B.
9. REFERENCES
ABOWD, G.D. AND MYNATT, E.D., 2000. Charting past, present, and future
research in ubiquitous computing. ACM Transactions on Computer-Human
Interaction, 7(1), pp.29–58.
BAJO, J., DE PAZ, Y., DE PAZ, J., MARTIN, Q. AND CORCHADO, J., 2006. SMas: A
Shopping Mall Multiagent Systems. In E. CORCHADO, H. YIN, V. BOTTI, &
C. FYFE, eds. Intelligent Data Engineering and Automated Learning –
IDEAL 2006. Lecture Notes in Computer Science. Springer Berlin /
Heidelberg, pp. 1166–1173.
BANGOR, A., KORTUM, P.T. AND MILLER, J.T., 2008. An Empirical Evaluation of
the System Usability Scale. International Journal of Human-Computer
Interaction, 24(6), pp.574–594.
BELL, G. AND DOURISH, P., 2007. Yesterday’s tomorrows: notes on ubiquitous
computing’s dominant vision. Personal and Ubiquitous Computing, 11(2),
pp.133–143.
BLACK, D., CLEMMENSEN, N.J. AND SKOV, M.B., 2009. Supporting the
supermarket shopping experience through a context-aware shopping trolley.
In Proceedings of the 21st Annual Conference of the Australian Computer-
Human Interaction Special Interest Group: Design: Open 24/7. ACM, pp. 33–
40.
BOUSSARD, M., CHRISTOPHE, B., LE BERRE, O. AND TOUBIANA, V., 2011.
Providing user support in Web-of-Things enabled smart spaces. In
Proceedings of the 2nd International Workshop on Web of Things. ACM, pp.
11:1–11:6.
BROOKE, J., 1996. SUS - A quick and dirty usability scale. In P. W. JORDAN, B.
THOMAS, B. A. WEERDMEESTER, & I. L. MCCLELLAND, eds. Usability
evaluation in industry. Taylor & Francis, pp. 189-194.
CARTER, S.A. AND MANKOFF, J., 2004. Challenges for ubicomp evaluation,
Computer Science Division, University of California.
CLAUS, R. AND MARTIN, R., 2004. Personalized multi-criteria decision strategies in
location-based decision support. Annals of GIS, 10(2), pp.149–156.
DAVIES, N. AND GELLERSEN, H.W., 2002. Beyond prototypes: Challenges in
deploying ubiquitous systems. Pervasive Computing, IEEE, 1(1), pp.26–35.
Embodying Services into Physical Places: Towards the Design of a Mobile Environment Browser 39:37
ACM Transactions on xxxxxxxx, Vol. xx, No. xx, Article xx, Publication date: Month YYYY
EDWARDS, W. AND GRINTER, R., 2001. At home with ubiquitous computing: Seven
challenges. In Proceedings of the 1rst International Conference on Ubiquitous
Computing. pp. 256–272.
FOX, A., DAVIES, N., DE LARA, E., SPASOJEVIC, M. AND GRISWOLD, W., 2006.
Real-World Ubicomp Deployments: Lessons Learned. Pervasive Computing,
IEEE, 5(3), pp.21 – 23.
GARCIA MACIAS, J.A., ALVAREZ-LOZANO, J., ESTRADA, P. AND AVILES
LOPEZ, E., 2011. Browsing the Internet of Things with Sentient Visors.
Computer, 44(5), pp.46–52.
GIGERENZER, G. (2007). Gut feelings: The intelligence of the unconscious. Penguin.
GUINARD, D., TRIFA, V., MATTERN, F. AND WILDE, E., 2011. From the Internet
of Things to the Web of Things: Resource Oriented Architecture and Best
Practices. In D. UCKELMANN, M. HARRISON, & F. MICHAHELLES, eds.
Architecting the Internet of Things. Springer, pp. 97–129.
HELAL, A., KING, J., ZABADANI, H. AND KADDOURAH, Y., 2008. The Gator Tech
Smart House: An Assistive Environment for Successful Aging. Advanced
Intelligent Environments, H. HAGRASS, Editor, Springer Verlag.
HIPPEL, E., 2006. Democratizing innovation, MIT Press.
HIPPEL, E.V., 1986. Lead Users: A Source of Novel Product Concepts. Management
Science, 32(7), pp.pp. 791–805.
HOPPER, A., 2000. The Clifford Paterson Lecture, 1999. Sentient Computing.
Philosophical Transactions of the Royal Society of London. Series A:
Mathematical, Physical and Engineering Sciences, 358(1773), pp.2349–2358.
HULL, R., NEAVES, P. AND BEDFORD-ROBERTS, J., 1997. Towards situated
computing. In Digest of Papers of the 1rst International Symposium on
Wearable Computers, 1997. pp. 146–153.
JAMES, T., 2012. Smart factories. Engineering Technology, 7(6), pp.64 –67.
JAMESON, A. (2012). Choices and decisions of computer users. In J. A. JACKO,
Editor. The human-computer interaction handbook: Fundamentals, evolving
technologies and emerging applications (3rd ed.). CRC Press.
JBARA, S., KUFLIK, T., SOFFER, P. AND STOCK, O., 2007. Context Aware
Communication Services in “Active Museums.” In Proceedings of the
International Conference on Software-Science, Technology Engineering. pp.
127 –135.
JONES, Q. AND GRANDHI, S.A., 2005. P3 systems: putting the place back into
social networks. IEEE Internet Computing, 9(5), pp.38– 46.
JUNGLAS, I., 2007. On the usefulness and ease of use of location-based services:
insights into the information system innovator's dilemma. International
Journal of Mobile Communications, Inderscience, 5(4), pp.389–408.
KAHL, G. AND BURCKERT, C., 2012. Architecture to Enable Dual Reality for Smart
Environments. In Proceedings of the 8th International Conference on
Intelligent Environments. pp. 42 –49.
KINDBERG, T. ET AL., 2002. People, places, things: Web presence for the real world.
Mobile Networks and Applications, 7(5), pp.365–376.
KINDBERG, T. AND FOX, A., 2002. System software for ubiquitous computing.
Pervasive Computing, IEEE, 1(1), pp.70–81.
KORTUEM, G., KAWSAR, F., FITTON, D. AND SUNDRAMOORTHY, V., 2010.
Smart objects as building blocks for the Internet of things. Internet
Computing, IEEE, 14(1), pp.44 –51.
39:38 P. Thébault et al.
ACM Transactions on xxxxxxxx, Vol. xx, No. x, Article x, Publication date: Month YYYY
LUCKE, D., CONSTANTINESCU, C. AND WESTKÄMPER, E., 2008. Smart Factory
- A Step towards the Next Generation of Manufacturing. In M. MITSUISHI,
K. UEDA, & F. KIMURA, eds. Manufacturing Systems and Technologies for
the New Frontier. Springer London, pp. 115–118.
NEWMAN, M.W. ET AL., 2002. Designing for serendipity: supporting end-user
configuration of ubiquitous computing environments. In Proceedings of the
4th Conference on Designing Interactive Systems: Processes, Practices,
Methods, and Techniques. pp. 147–156.
PFISTERER, D. ET AL., 2011. SPITFIRE: toward a semantic web of things.
Communications Magazine, IEEE, 49(11), pp.40 –48.
PREUVENEERS, D. AND BERBERS, Y., 2012. Intelligent Widgets for Intuitive
Interaction and Coordination in Smart Home Environments. In Proceedings
of the 8th International Conference on Intelligent Environments (IE). pp. 157
–164.
RAO, B. AND MINAKAKIS, L., 2003. Evolution of mobile location-based services.
Communications of the ACM, 46(12), pp.61–65.
RODUNER, C., 2010. BIT–A Browser for the Internet of Things. In Proceedings of the
CIoT Workshop 2010 at the 8th International Conference on Pervasive
Computing. pp. 4–12.
ROGERS, Y., 2006. Moving on from weiser’s vision of calm computing: Engaging
ubicomp experiences. In Proceedings of the 8th International Conference on
Ubiquitous Computing. pp.404–421.
SALT, S., 2011. Social Location Marketing: Outshining Your Competitors on
Foursquare, Gowalla, Yelp & Other Location Sharing Sites, Pearson
Education.
SATYANARAYANAN, M., 2001. Pervasive computing: Vision and challenges. IEEE
Personal communications, 8(4), pp.10–17.
SKUBIC, M., ALEXANDER, G., POPESCU, M., RANTZ, M. AND KELLER, J., 2009.
A smart home application to eldercare: Current status and lessons learned.
Technology and Health Care, 17(3), pp.183–201.
SOLOMON, M.R., 1996. Consumer behavior: buying, having, and being, Prentice-
Hall.
SPIESS, P. ET AL., 2009. SOA-Based Integration of the Internet of Things in
Enterprise Services. In Proceedings of the IEEE International Conference on
Web Services, 2009. pp. 968 –975.
STREITZ, N. AND NIXON, P., 2005. The disappearing computer. Communications of
the ACM, 48(3), pp.32–35.
WEISER, M., 1995. The computer for the 21st century. Scientific American, 272(3),
pp.78–89.
ZENG, D., GUO, S. AND CHENG, Z., 2011. The Web of Things: A Survey (Invited
Paper). Journal of Communications, 6(6).
