Content uploaded by Schubert Foo
Author content
All content in this area was uploaded by Schubert Foo
Content may be subject to copyright.
TILES: classifying contextual
information for mobile tourism
applications
Esther Meng-Yoke Tan, Schubert Foo, Dion Hoe-Lian Goh and
Yin-Leng Theng
Wee Kim Wee School of Communication and Information,
Nanyang Technological University, Singapore
Abstract
Purpose – The design of context-aware mobile applications can be improved through a clear and
in-depth understanding of context and how it can be used to meet users’ requirements. Using tourism
as a case application, this paper aims to address the lack of understanding of context and tourists’
goals.
Design/methodology/approach – This is achieved through a literature review of existing research
and focus groups to gather information needs for tasks commonly executed by tourists.
Findings – This paper proposes the TILES (temporal, identity, location, environmental and social)
model to define and classify five main contextual types, and properties associated with each type for
tourism-related applications. The TILES model (with 32 factors) derived from the analysis of the
literature review is refined through inputs from two focus groups to incorporate an additional ten
factors.
Research implications/limitations – The TILES model can be generalised to support domains
other than tourism, such as medical and edutainment.
Originality/value of paper – The model will help to achieve a better understanding of context,
users’ information needs and their goals. In addition, this work extends findings in the field of
context-aware computing and information retrieval on mobile devices. Solution providers will also be
able to adopt TILES as a framework for guiding the design of their context-aware mobile applications.
Keywords Mobile communication systems, Information retrieval, Tourism
Paper type Research paper
1. Introduction
In the past few years, the adoption of mobile devices has grown tremendously (Kawash
et al., 2007), and their characteristics of mobility and connectivity support on-demand
services that are tailored to users and their specific situations, any time, anywhere.
CRUMPET (Poslad et al., 2001) and GUIDE (Cheverst et al., 2002), for example, are
mobile tourism applications designed to be aware of the tourist’s location and interests.
They are described as context-aware applications because they are sensitive to a user’s
context.
Saracevic (1996a, b) argued that context should be used to consider the relevance of
information. Setten et al. (2004) have demonstrated that context can be used to measure
the relevance of information such that only appropriate information is presented. They
are supported by Wilson (1973) and Mizzaro (1997) as well as Borlund and Ingwersen
(1997). This is also confirmed by Albers and Kim (2002), who have also highlighted
that “delivering the right content for the right context is crucial” (p. 194). In addition,
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/0001-253X.htm
Classifying
contextual
information
565
Received 12 April 2009
Revised 19 June 2009
Accepted 18 August 2009
Aslib Proceedings: New Information
Perspectives
Vol. 61 No. 6, 2009
pp. 565-586
qEmerald Group Publishing Limited
0001-253X
DOI 10.1108/00012530911005526
Spink et al. (1998) have cautioned that the usefulness of the content is largely
dependent on the users’ judgment.
For these reasons, many mobile applications in different fields are designed with
context-awareness features. These include those in the medical field (Kjeldskov and
Skov, 2004), office automation (Schilit et al., 1994) and tourism (Setten et al., 2004).
As will be discussed, while context-awareness has advantages, there are many
challenges in incorporating such techniques into mobile applications. This research
aims to analyse limitations in context-aware application design using mobile tourism
as a basis. This is because there are many good examples of how the context can be
used in context-aware applications for tourists. Existing research has also explored
and identified some outstanding problems faced by tourists. These findings form the
foundation of this research when analysing the weakness of such applications as
discussed in the following section.
1.1 Challenges in context-awareness in mobile tourism
Contextual information is important when adapting information to meet tourists’
needs. A review of mobile tourism applications has uncovered major issues such as a
lack of consensus in the definition of context, and a lack of understanding of tourists’
goals.
1.1.1 Lack of consensus in the definition of context. Researchers have proposed
different context types in their definitions of context. For example, Dey and Abowd
(2000) defined context to include location, temporal elements, identity and activity.
Schilit et al. (1994) also included descriptions on identity and environment. Korkea-aho
(2000) extended the scope to include the social aspect, proximity, device and
physiology. Context-aware mobile tourism applications are also designed to support
different context types. GUIDE supports identity, social and environmental contexts
(Cheverst et al., 2002). COMPASS supports location, temporal, identity and
environmental contexts (Setten et al., 2004), and CRUMPET supports location,
identity, network and device contexts (Poslad et al., 2001).
Designers of context-aware applications have also classified contextual properties
differently. For example, GUIDE supports current time and date under the
environmental context (Cheverst et al., 2002), but m-ToGuide (Kamar, 2003) and
CATIS (Pashtan et al., 2003) include time and date under the temporal context.
Some context types, such as the identity context, are supported by rich contextual
properties. They include the user’s name, age, preference in food, lodgings, price range,
information-seeking trend, shopping lists and travel agenda (Pashtan et al., 2003;
Poslad et al., 2001; Setten et al., 2004). On the other hand, the social context is less
well-defined in general, and merely includes information on tour companions, people
nearby and others’ comments (Schilit et al., 1994; Cheverst et al., 2002).
With different definitions of context, mobile tourism applications have proposed
different repository designs for storing and indexing of the contextual information
(Feng et al., 2004; Hinze and Buchanan, 2005). Therefore, information providers, such
as places of interest and restaurants, will have to structure and format their
information differently to suit these different applications. The user interface designs
of these mobile tourism applications are also tightly associated with their definition of
context. When using different mobile tourism applications during their tours, tourists
AP
61,6
566
will have to adapt to different definitions of context and properties, making it difficult
for them to learn and use the applications.
1.1.2 Lack of understanding of tourists’ goals. Current work in context awareness
typically employs logic that determines tourists’ contexts, such as interest and current
location. They then adapt their information automatically based on these contexts.
However, as Cheverst et al. (2002) have highlighted, there is a concern that there could
be a mismatch between tourists’ goals and the adaptation due to a lack of
understanding fo tourists’ goals with respect to context. In the case of GUIDE, the
locations of attractions pertaining to tourists’ interests will not be revealed if they are
closed, but their study has revealed that tourists may still want to visit a place despite
it being unavailable. For example, tourists could still visit and view the Eiffel Tower
outside its operating hours. Hinze and Buchanan (2005) have also highlighted that
users change their roles and situations very frequently. For instance, while walking on
the street towards their destination, they may meet with a friend who has access to a
car, hence having a new set of preferences. The former planned route, estimated
travelling time and destination may no longer be valid and there would be a need for
re-adaptation. This observation shows that despite the intelligence embedded in the
information adaptation process, such information may not meet tourists’ requirements,
especially when situations change frequently. Thus, there is a need to investigate how
contextual information should be gathered and used to meet tourists’ changing goals.
1.2 Research objectives
Using mobile tourism applications as a basis, this paper proposes and evaluates the
collection of contextual information relevant to this domain with sub-objectives that
include:
.reviewing the contextual information that applies to mobile tourism with the aim
of developing a comprehensive typology of information attributes that relate to
the context (herein referred to as typology of contextual information); and
.evaluating the comprehensiveness and usefulness of the typology of contextual
information.
Section 2 describes existing contextual information models supporting context-aware
mobile tourism application, while Section 3 gives a brief description of the research
methodologies used to triangulate the results. Section 4 reports the findings of the
research, while Section 5 presents our proposed contextual information model.
2. Related work
A study of the related literature has shown that contextual information is organised
differently to support each context-aware mobile application. The following sections
include reviews of organisational methods used for contextual designs in existing
work. The goal is to understand possible ways of organising and presenting contextual
information. These context categorisation frameworks unanimously present
contextual information in a hierarchical method.
Feng et al. (2004) organised contextual information into two major groups:
(1) user-centric; and
(2) environmental.
Classifying
contextual
information
567
The user-centric context refers to any contextual information related to the user, such
as behaviour, and physiological and emotional state. The environmental context refers
to contextual information related to the physical, social and computational
environment. The physical environment refers to elements such as time and
location. The social environment refers to people nearby, and the computational
environment refers to devices nearby.
Hinze and Buchanan (2005), however, have grouped contextual information into
three major categories:
(1) network context;
(2) device context; and
(3) application context.
The application context is then further categorised into user and site context. The user
context can be described as user static information such as user interests and
background and user-fluent information such as time, location and direction of the
user. This grouping of user context into static and fluent is supported by O’Grady et al.
(2007). They also assumed that the user’s profile will remain the same for a certain
length of time and that the user’s current location and activity will change
dynamically. Hinze and Buchanan (2005) also adopted site context as site general, such
as the location and operating hours of places of interest, and site current, which ties in
with the user context such as the popularity of a site adapted by the user’s interests. It
is also important to observe that only Hinze and Buchanan (2005) indicated the
interaction between the categories. They proposed that these contexts are
inter-dependent when used for information adaptation.
As opposed to Hinze and Buchanan (2005) and Feng et al. (2004), Korpipaa et al.
(2003) levelled out the categories and sub-categories. They placed location, time,
environment, user and device at the same level. Any newly discovered contexts could
be added by expanding the architecture breadth-wise. They proposed that each context
type can be further refined into sub-context types and properties at the lowest level:
.environment:sound:intensity, with a value of {silent, moderate, loud};
.environment:light:intensity, with a value of {dark, normal, bright};
.environment:light:type, with a value of {artificial, natural}; and
.environment:temperature, with a value of {cold, normal, hot}.
In this case, the environment is a context type. Sound and light are sub-context types of
the environment. Intensity, types and temperature are context properties in their own
group of context and sub-context types.
Finally, Wang et al. (2004) proposed an ontology to model context information. In
their context ontology (CONON), four main context types are identified:
(1) location;
(2) user;
(3) activity; and
(4) computational entity.
AP
61,6
568
Each context type was further classified into sub-context types such as service,
application, device, network and agent for computational entity. It is interesting to
observe that they suggested that this set of general contexts with specific features
could be applied in different domains. For example, the location context has
sub-categories of indoor and outdoor space. For the home domain, indoor space
includes building, room, corridor and entry, while outdoor space includes garden and
dooryard.
All four of the organisation methods reviewed adopt the hierarchical tree as the
common approach for organisation. From the analysis of techniques, it is observed that
modelling of contextual information using the hierarchical tree allows expansion and
easy addition of new context types and properties. This is consistent with the findings
of Peng and Choi (2002), who suggested that a tree structure allows easy addition of
new categories and facilitates searching for information. Fan et al. (2006) also
suggested that hierarchical classification can be used to index information for easy
retrieval. They adopted it to index images differently, depending on the context of use.
In the hierarchical tree, items can be indexed in a nested manner and can be used to
support techniques to bridge the gap between users’ understanding and the actual
classification of media clips for easy retrieval.
These organisation methods also focused on classifying contextual information to
better support the design of the mobile tourism application. However, to the best of our
knowledge, there is no work on validating whether this information fulfils tourists’
needs.
3. Research methodology
The present research employs two research methodologies. First, a review of
context-aware mobile tourism applications was used to collate the existing list of
contextual information adopted by these applications. Next, two focus groups were
conducted to gather tourists’ information needs so as to validate and improve on the
existing list.
3.1 Review of existing context-aware mobile tourism applications
The objective was to collate and synthesise the context types and contextual properties
adopted by existing context-aware mobile tourism applications. This review included
observations of the use of contextual information in these applications. This was done
through reading and extracting context-related information from journal articles
published on these applications. The context types were extracted from 25 articles as
referenced in Table I. These articles have frequently been referred to in context-related
research. The contextual information related to mobile tourism applications was
extracted from 12 journal articles as referenced in Table II. These 12 articles gave
comprehensive descriptions of the contextual information used by well-recognised
mobile tourism applications. The review also included a sampling of printed materials
and websites to obtain a general understanding of how these traditional media relate to
the various contexts. These included four official tourism websites from countries
popular among tourists and three printed tourist brochures commonly used by them
(listed in Table II).
The classification of context type and properties was done through a preliminary
assessment and will be confirmed in future research through inter-coder agreement to
Classifying
contextual
information
569
Domain Location Temporal Identity Environmental Social Proximity Network Activity Device Physiology Cognitive
Research findings on usage of context in context-aware applications
Dey and Abowd (2000) Survey on
context £££ £
Korkea-aho (2000) Survey on
context £££ £ ££ £££
General purpose context-aware applications
Ghosh-Dastidar et al.
(2007); epileptic seizure
detection
Medical
£
Jung et al. (1997); detect
drowsiness in drivers
Medical
£
Kjeldskov et al. (2004);
MobileWARD
Medical
££
Pousman et al. (2004);
event planner
Office
automation £££ £
Fogarty et al. (2004);
MyVine
Office
automation £ £
Schilit et al. (1994);
PARCTAB
Office
automation £££££
Elliott and Tomlinson
(2006); Personal
soundtrack
Edutainment
£
Dornbush et al. (2007);
XPOD
Edutainment
££ £
Jeong and Lee (2007);
Context-aware HCI for
ubiquitous learning
Edutainment
££ ££
Context-aware mobile tourism applications
Cheverst et al.
(2000a, b, c, 2002);
GUIDE
Outdoor
guide
£££
(continued)
Table I.
Categories of
context-aware
applications and their
context
AP
61,6
570
Domain Location Temporal Identity Environmental Social Proximity Network Activity Device Physiology Cognitive
Kamar (2003); m-
ToGUIDE
Outdoor
guide £££ £
Pashtan et al. (2003);
CATIS
Outdoor
guide £££
Poslad et al. (2001);
CRUMPET
Outdoor
guide ££ ££
Roth (2002); Pinpoint Outdoor
guide £££ £
Hinze and Buchanan
(2005), Hinze and
Voisard (2003); TIP
Outdoor
guide
££ ££
Setten et al. (2004);
COMPASS
Outdoor
guide £££ £
Cano et al. (2006);
UbiqMuseum
Indoor guide
££ £
Roffia et al. (2005);
MUSE
Indoor guide
££
Hsi (2002); electronic
guidebook
Indoor guide
££
Table I.
Classifying
contextual
information
571
Mobile tourism applications Official tourism websites Print material
Properties GUIDE
a
m-
ToGUIDE
b
CATIS
c
CRUMPET
d
PinPoint
e
TIP
f
UbiqMuseum
g
COMPASS
h
MUSE
i
Electronic
Guidebook
j
California
k
Singapore
l
New
Zealand
m
Australia
n
Toronto
o
YES
Guide
p
Frommer’s
q
DK
Eyewitness
r
1 Users’ interests
(e.g. history,
architecture) I I I I I I I I I I I I
2 Current location L L L L L L L L L L
3 Time of the day,
current date T T T T T
4 Nearby attractions L L L L
5 Weather EEEE
6 Types of device,
screen resolution,
colour D D D D
7 Duration of stay I II
8 Name I I I
9 Preferred language III
10 Types of network N N N
11 Latest happening
events TT
12 Travelling speed L L
13 Country of origin II
14 Purpose of trip
(leisure, LOHAS,
MICE) I I
15 Attractions visited
by tourists I I
16 Last visited date of
a place II
17 Favourite cuisine I I
18 Refreshment
preferences I I
19 Dietary
preferences I I
20 Preferred
information level
of detail II
21 Elapsed time since
last measured
location (e.g. three
hours ago) L
22 Events round the
year T
(continued)
Table II.
Contextual properties and
their classification in
mobile tourism
applications, websites
and printed media
AP
61,6
572
Mobile tourism applications Official tourism websites Print material
Properties GUIDE
a
m-
ToGUIDE
b
CATIS
c
CRUMPET
d
PinPoint
e
TIP
f
UbiqMuseum
g
COMPASS
h
MUSE
i
Electronic
Guidebook
j
California
k
Singapore
l
New
Zealand
m
Australia
n
Toronto
o
YES
Guide
p
Frommer’s
q
DK
Eyewitness
r
23 Distance between
location and
services L
24 Travelling
direction L
25 Current mode of
transport L
26 Information-
seeking trend I
27 Groups’ interests S
28 Group name S
29 Age I
30 Preferred price
range I
31 Number of
repeated visits I
32 People nearby S
33 Recommendations
and reviews by
other tourists S
34 Travelling
companions – who
are they? S
35 Agenda I
36 Shopping list I
37 User activity (e.g.
walking, driving) I
38 User status (e.g.
free or busy) I
39 Traffic/road
condition E
40 Storage, download
and display
capability D
Notes: T, time; I, identity; L, location; E, environmental; S, social; D, device; N, network.
a
GUIDE (Cheverst et al., 2000a, b, 2002);
b
m-ToGUIDE (Kamar, 2003);
c
CATIS (Pashtan et al., 2003);
d
CRUMPET (European Media Laboratory, 2006; Poslad et al., 2001);
e
PinPoint (Roth, 2002);
f
TIP (Hinze and Buchanan, 2005; Hinze and Voisard, 2003);
g
UbiqMuseum (Cano et al., 2006);
h
COMPASS (Setten et al., 2004);
i
MUSE (Roffia et al., 2005);
j
Electronic Guidebook (Hsi, 2002);
k
California (see www.visitcalifornia.com);
l
Singapore (see www.visitsingapore.com);
m
New Zealand (see www.newzealand.com);
n
Australia (see www.australia.com);
o
Toronto (see www.torontotourism.com);
p
YES Guide (YES, 2007);
q
Frommer’s (Frommer, 2007);
r
DK Eyewitness (DK Eyewitness, 2003)
Table II.
Classifying
contextual
information
573
ensure that the context type and properties are analysed consistently among other
analysts when given a set of criteria (Lombard et al., 2002). A total of 40 contextual
properties are supported by ten mobile tourism applications, five tourism websites and
three printed materials, of which each contextual property is classified under different
context types including identity, location, environmental, social, network, device and
temporal context. The relative importance of each context type is measured by the
number of contextual properties supported by the mobile tourism applications.
3.2 Focus group findings to understand tourists’ information needs
The objective of the two focus groups was to explore tourists’ information needs and
understand the problems they encountered during their trips so as to validate the gaps
in existing context-aware mobile tourism applications. From their information needs,
contextual information can be extracted to validate and extend the list of context types
and context properties derived from review conducted on the existing mobile tourism
applications. Two focus groups were held, each with eight participants aged between
20 to 35 years old. Each group had an equal number of male and female participants.
Participants were working professionals with both leisure and business travelling
experiences. The results gathered from the first session were used to refine and
enhance the effectiveness of the second session. Each focus group consisted of two
parts. In the first part, participants were asked to share their information needs
through scenario-based questions. In the second part, they were asked to identify
contextual properties supporting their information needs based on these questions.
4. Findings and analyses
This section reports the findings from the review of mobile tourism applications and
the focus group. It will describe the contextual information derived from the two
studies, while Section 5 presents the derived model.
4.1 Review of existing context-aware mobile tourism applications
4.1.1 Context types in context-aware mobile applications. This section discusses the
contextual information used in context-aware applications. These applications were
grouped by their domains and purposes, which included the medical field, office
automation, edutainment, outdoor guides and indoor guides. The medical group was
selected for its unique design consideration, which included users’ physiological data
(Raskovic et al., 2004). The office automation group was selected because their daily
activities may be similar to those activities carried out by tourists (Perttunen and
Riekki, 2005). The edutainment group was selected because they might involve useful
features and information relevant to tourist applications, such as users’ preferences
and their location (Dornbush et al., 2007; Jeong and Lee, 2007).
The context types adopted by these context-aware applications were extracted and
compiled as parameters used in the evaluation as shown in Table I. The context types
include:
.The location (proximity) context – describes the data varying based on the user’s
location.
.The temporal context – describes the temporal characteristics of the application.
The identity context refers to the profile of the users. Some examples include
users’ interests, preferred language and gender.
AP
61,6
574
.The environmental context – describes the surroundings of the users.
.The social context – refers to the social setting of users, which includes their
travelling companions and people nearby.
.The network context – describes the available network resources for the
applications.
.The activity context – refers to the tasks currently executed by the users.
.The device context – describes the capabilities and features of the device.
.The physiology context – refers to information about users’ physiological status
such as blood pressure and heart rate.
.The cognitive context – describes the user’s state of mind when using
context-aware applications. It includes consideration of users’ emotions, current
mindset and stress level.
As expected, it was observed that context-aware applications adopt different context
types largely based on the purpose of the applications and users’ needs. As stated by
Kaasinen (2003), it is difficult to identify the context types necessary for context-aware
applications. While there is no standard set of context types designated for
applications in each field, context design can be improved by considering a suitably
comprehensive set of contexts. This is supported by Christensen et al. (2006), who
explained that gathering more contextual information will not necessarily help
context-aware applications meet users’ needs. The key here lies in how to adopt the
appropriate context types, and interpret and use them in the applications. Thus, a
comprehensive typology of contextual information proposed in this paper will serve as
a useful guide for developers deciding which contextual information is to be
incorporated in the design of their context-aware applications.
4.1.2 Contextual properties supported by mobile tourism applications. This section
reports on the context types and their respective properties commonly adopted by the
mobile tourism applications as shown in Table II. From the analysis of literature, no
mobile tourism application has adopted properties under the physiology context and
activity context. Put differently, mobile tourism applications are not sensitive to the
human physiological state and current activity, probably because this sensitivity is not
required by users.
4.2 Focus group findings to understand tourists’ information needs
The focus groups gathered general tourists’ information needs by walking through
four scenarios which represent commonly executed tasks. The scenarios included
visiting places of interest, watching cultural performances, purchasing souvenirs and
selecting restaurants. The participants suggested grouping the information needs into
the context types (i.e. temporal, identity, location, environmental and social). The
information needs and grouping were presented in mind-maps. The mind-map shown
in Figure 1 is an example of information needs when selecting restaurants. Other
mind-maps have similar characteristics and are not shown here for brevity.
Participants grouped related information into restaurant settings, restaurant services,
restaurant requirements, when (timing), how to get there, factors affecting what to eat
and search-by factors. As shown in Figure 1, they identified information related to the
five context types (i.e. temporal, identity, location, environmental and social). In the
Classifying
contextual
information
575
temporal context, they included time of day and time of year. In the identity context, they
included preferences such as preferred price range, preferred transport, preferred portion
size, acceptable waiting time, and acceptable hygiene level. In the location context, they
included nearby available food and route to the nearest restaurant. Finally, they
classified what others were eating and rating by others under the social context.
The focus group participants also decided that the search for restaurants should
take into consideration information from all the context types (i.e. temporal, identity,
location, environmental and social).
5. The proposed contextual information model – TILES
From the findings presented in previous sections, we propose our contextual
information model in this section. The TILES model encapsulates contextual
information, which includes context types and their respective contextual properties.
The acronym TILES stands for the five context types, i.e. temporal, identity, location,
environmental and social. The TILES model focuses on mobile tourism guides at the
application level, supported by software applications. It does not include device and
network contexts because sensing these contexts involves detecting and reading the
hardware’s configuration. It also does not include the physiology context because that
involves sensing the human body’s condition. The device, network and physiology
contexts were also not identified during the focus group, probably because participants
did not see the need to filter information by these contexts. Table III reveals the widely
accepted properties in each context type that are emphasised in the TILES model.
Table III includes properties adopted by existing mobile tourism applications, findings
Figure 1.
Tourists’ information
needs when selecting
restaurants
AP
61,6
576
Literature
review
Focus
group
Temporal – data varying according to time
Current time of day and year UU
Latest happening events U
Events round the year U
Seasons of the year U
Identity – data varying according to user’s identity
User interests UU
Profile – name, birth date, country of origin and age UU
Preferred language UU
Duration of stay U
Purpose of trip U
Attractions already visited by tourists U
Last visited date U
Preferred types of food – refreshments, dietary, cuisine UU
Preferred information level of detail U
Information-seeking trend U
Preferred price range UU
Number of repeated visits U
Tour schedule UU
Current activity – walking, driving, etc. U
User status – free or busy U
Tourists’ free time U
Cravings U
Acceptable waiting time U
Acceptable security level U
Preferred transport mode U
Preferred quality level such as service quality, meal portion size U
Acceptable hygiene/safety level U
Location – data varying according to user’s location
Current location UU
Nearby attractions UU
Travelling speed U
Age of last measured location U
Distance between location and services U
Travelling direction U
Current mode of transport U
Environmental – data varying according to user’s environment
Weather – temperature, climate, humidity, air quality UU
Traffic/road conditions UU
Available seats U
Social – data varying according to user’s social setting
Group’s interests U
Group name U
Nearby people – tour members, friends and like-minded strangers U
Recommendation, reviews by other tourists UU
Travel companions – who are they? UU
Pictures and video clips posted by others U
Table III.
Refined TILES model
Classifying
contextual
information
577
from existing research, as well as properties gathered from the focus group (indicated
by tick marks in the columns of Table III).
The following explains the context types and properties, and how they are verified
and supported by our findings.
5.1 Temporal
Time of day and current date are temporal context properties used to describe the
current time of the day and current date. They are frequently used in information
adaptation to determine relevant events or activities happening in the near future. TIP
(Hinze and Voisard, 2003; Hinze and Buchanan, 2005) is a tourist guide that
incorporates time of day and the location of tourists to suggest tour plans that include
relevant tour activities such as conference talks, lunch venues and places of interest. It
was also observed that m-ToGuide (Kamar, 2003), CATIS (Pashtan et al., 2003), GUIDE
(Cheverst et al., 2000c) and Pinpoint (Roth, 2002) also support this contextual property.
Some of the focus group participants suggested that their selection of places of
interest such as restaurants depended on whether they were planning for breakfast,
lunch, dinner or a late-night snack. Therefore, it was essential to consider the time of
day when making decisions about restaurants. Tourists might also like to participate in
seasonal or festive events, such as visiting a Chinese garden during the lantern festival.
In this case, the event and seasons of the year in the temporal context would affect their
decisions on places of interest. As one participant mentioned: “I used to schedule my
Hong Kong tour trip around the month of June to catch the summer sales”.
The property of current time of day and year was also included in the design of
GUIDE (Cheverst et al., 2002), m-ToGuide (Kamar, 2003), CATIS (Pashtan et al., 2003),
Pinpoint (European Media Laboratory, 2006; Poslad et al., 2001) and TIP (Hinze and
Buchanan, 2005).
5.2 Identity
The property of users’ interests refers to their interests and concerns. These might
include topics such as architecture, history and shopping. In the review, all five mobile
tourism applications supporting this property classified it under the identity context
(Cheverst et al., 2000a, b, c; Kamar, 2003; Poslad et al., 2001; Hinze and Voisard, 2003;
Hinze and Buchanan, 2005; Setten et al., 2004). It was also interesting to observe that all
the official tourism websites and printed media, except the California tourism website
(see: www.visitcalifornia.com), aimed to tailor their tourist information according to
the interests of travellers. This is confirmed by Myrhaug et al. (2004), who have
highlighted that users’ contexts should include their personal preferences. Goker and
Myrhaug (2008) also represented the user’s situation from the user’s perspective. In an
example, they included user’s preference as a property.
The preferred language is a property that describes the language best understood
by tourists. Among all the mobile tourism applications, only Ubiqmuseum (Cano et al.,
2006) was designed to support different languages and allow tourists to choose their
preferred one. In terms of websites, Singapore (see www.visitsingapore.com) and New
Zealand (see www.newzealand.com) also supported the preferred language of tourists.
The duration of trip property is supported by m-ToGuide (Kamar, 2003), the official
Singapore tourism website (see www.visitsingapore.com) and the YES guide print
material (YES, 2007). Depending on the tourists’ trip duration, different activities and
AP
61,6
578
tour agenda are recommended so that the tourists can visit as many places of interest
as possible during their stay.
In COMPASS (Setten et al., 2004), third-party services were employed to consider
users’ status and to provide appropriate information depending on whether the user is
free or busy. They also adapted tourist information according to the user activity; for
example, if the user is driving, the amount of time required to get from the current
location to the destination will be shorter than if he or she is walking.
Some of the focus group participants mentioned that they would consider the price
of services, such as entrance fees and menu prices, and quality of services such as
duration of show and portion sizes in the restaurant, ensuring that they were within
their acceptable price range and quality of service. The acceptable price range property
was consistent with the findings in the literature review and was supported by Pashtan
et al. (2003). Participants were also concerned with the hygiene or safety conditions of
the place of interest and the average waiting time. They also agreed that acceptable
hygiene or safety level and waiting time duration varies according to an individual’s
preferences. Many of them also indicated that their selection of places of interest
sometimes depends on their cravings, such as a sudden preference for a certain type of
food or to watch a performance. One participant mentioned that he would not mind
travelling further to satisfy his cravings. Concern was expressed about whether the
physical surroundings of places of interest and the journey there were safe – that is,
whether the place is within their acceptable level of security. Another participant said:
“For example, when I was visiting San Francisco, I chose to dine in my hotel cafeteria
instead of a highly recommended restaurant in Chinatown ... It was already 8.00 pm
and I didn’t feel comfortable travelling to the other side of town alone”.
Finally, the identity context also includes contextual properties that are least
commonly adopted, such as:
.country of origin;
.purpose of trip;
.attractions last visited by tourists;
.the time a place was last visited;
.favourite cuisine;
.information-seeking trend;
.age;
.affordable price range;
.number of repeated trips;
.agenda;
.shopping list;
.refreshment and dietary preferences; and
.preferred level of detail for information.
5.3 Location (proximity)
The property of current location refers to the existing physical position of tourists. This
is a widely adopted property and is supported by all ten mobile tourism applications
reviewed in our work. Traditional media, websites and printed materials do not
Classifying
contextual
information
579
support this because these media do not have positioning technology, which is
essential to provide this contextual information. This property is also confirmed by
Raper et al. (2007) and Jose et al. (2003), who emphasised the importance of using
current location as the way to measure the relevancy of information.
Tourists’ nearby attractions is also a commonly supported property that refers to
places of interest near the tourist’s current location. It was adopted by CRUMPET
(Poslad et al., 2001), COMPASS (Setten et al., 2004), GUIDE (Cheverst et al., 2000a) and
Ubiqmuseum (Cano et al., 2006). All mobile tourism applications had classified nearby
attractions under the location context.
The focus group participants were concerned about how to get to places of interest
and whether the journey was handicap-friendly. They also wanted to know the
cheapest as well as the fastest available mode of transport. Some wanted to know
whether it was within walking distance and along their planned route. All agreed that
this information could only be available given their current location. From the current
location, the mobile tourism application would be able to calculate the path, cost and
time needed to get to the destination and work out whether it was within walking
distance. They also agreed that the location of the place and transportation schedule is
part of the domain information.
The focus group also gathered comments about the location context, which
included:
I would like to know the shopping malls near me. I would also like to know the distance
between my current location and the nearby mall; it should be measured by walking time,
driving time or the number of blocks away, and not kilometres.
These comments are supported by Mountain and Macfarlane (2007), who included
travelling time as a context property in the design of their mobile information retrieval
solution.
Likewise, the properties of distance between location and services and travelling
direction are also confirmed by Mountain and Macfarlane (2007), who took into
consideration the distance between the users’ current location and the information
source. In addition, our analyses show that there was also a set of less commonly
adopted contextual properties. They included age of last measured location,travelling
speed,distance between location and services and travelling direction.
5.4 Environmental
The property of weather includes descriptions on temperature, humidity, climate and
air quality. Setten et al. (2004) was the only mobile tourism application that supported
this property and had placed it under the environmental context. Their design was
consistent with the official tourism websites of Singapore (see: www.visitsingapore.
com), Toronto (see: www.torontotourism.com) and Australia (see: www.australia.com),
which also published information on their current weather. Setten et al. (2004) also
included the property of traffic condition to take into consideration the current flow of
traffic. If there was a traffic jam along the tour route, their application would be able to
detect this and recommend a different route.
When selecting places of interest, the focus group participants wanted to know the
availability of facilities such as height limitations in a theme park and types of liquor
served in a restaurant. The participants also highlighted that they were concerned with
AP
61,6
580
the current weather. They also wanted to know how to get to a place of interest using
the quickest route, thus requiring information about the current road conditions such
as traffic jams.
Both properties of weather and traffic/road conditions were supported by focus
group findings. One participant explained: “When the weather is nice and cool, I might
select restaurants with alfresco dining and enjoy a hot beverage. However, when the
weather is hot, I would prefer cold drinks in an air-conditioned environment”.
Another participant suggested: “It may be nice to have recommendations on
suitable clothing”. Most participants agreed that the weather would affect their
selection of restaurants. One commented: “When it is freezing cold, I will fill my tour
agenda with indoor activities such as museum visits and shopping. But if it is warm, I
may go for wind surfing”. Another added: “Of course, if there is a traffic jam along the
route to the beach, we can change our destination”.
5.5 Social
Travelling companions constitutes a property supported by GUIDE (Cheverst et al.,
2002) to describe people who are travelling together with tourists. Their design
suggested different places of interest depending on their travelling companions. Other
tourists nearby was also a property in the social context. Cheverst et al. (2002)
incorporated features indicating people near the tourists. The design allowed tourists
to prompt these people for comments and ratings of the places of interest where they
were located. For example, a tourist could prompt another person located at a nearby
cafe
´for comments. The comments given would be ranked by the relevancy, such as
travelling companions with the assumption that those travelling with family would
appreciate comments from tourists with similar travel companions.
It was also interesting to find that all focus group participants wanted to read about
comments and ratings by other tourists. Two of them said they make decisions by
considering what others were doing. They would visit popular places of interest. They
decided that what others recommend should be classified as part of the social context.
One participant said: “It is probably due to some kind of herd mentality ... but it is a
natural social behaviour to find out what others are doing and to follow the crowd”.
This observation is supported by Raper (2007), who indicated that people have a social
urge to join in with others if they are interested in the activity taking place.
Another participant also suggested:
The buddy finder service provided by the Singtel mobile service providers is useful during
the tour because I like to make new friends during my trip. The new friends/buddies can then
make recommendation on what is worth visiting and where to get value-for-money souvenirs.
In this context type, there were also two least commonly supported properties. They
were the group name, supported by Cheverst et al. (2002), and group interest, supported
by Poslad et al. (2001).
6. Discussion and conclusion
This work addresses the lack of understanding of context and tourists’ goals. It
reviews the contextual information supporting mobile tourism applications and
gathers tourists’ information needs through a focus group study. Based on our
findings, this paper proposes a contextual information model – TILES.
Classifying
contextual
information
581
Existing work such as GUIDE and COMPASS (Cheverst et al., 2000a, b, c; Setten
et al., 2004) have emphasised sensing and adapting information based on samples of
contextual information. Work on organising contextual information such as CONON
(Wang et al., 2004) has focused on organising portions of contextual information into a
model to support their context-based design. To the best of our knowledge, there is
little work that puts together a comprehensive list of context types and contextual
properties that can be used in the context-based design of mobile tourism application.
Thus, the TILES model is proposed to fill this gap.
The TILES model can also used by researchers to design new methods of meeting
users’ information needs in the tourism domain through context-aware applications. It
also adds to and supports existing research in other related areas such as working on
effective information retrieval methods in mobile tourism applications. The findings on
tourist information needs in the social context can also be used in the field of social
networking, for example deciding on the social contextual information to be presented
to social network users. The search criteria, shown in the search branch of the
mind-map (section 4.2), will help mobile service providers to design useful search
functions in their applications. An in-depth understanding of tourist preferences and
tourists’ information needs will also help providers to deliver better services to their
users, for example filtering information according to tourists’ preferences.
There are some limitations that should be addressed in future work. In the review,
contextual information is extracted from existing mobile tourism applications. The
analysis process was challenging because the designers of these mobile tourism
applications classified the contextual properties under different context types. The
analysis work was done without inter-coder agreement, but was mitigated by
triangulation through other methods such as focus groups. Nevertheless, in future
work, inter-coder analysis will be introduced to confirm and validate the findings.
Future work will also include refining the contextual information framework (TILES)
and exploring ways to fill the gaps of existing mobile tourism applications. The focus
group was conducted through a group of participants with travelling experience. Our
analysis is based on their general travelling experiences; it does not include differences
in their information needs by gender, age group and type of travel, such as leisure or
business. Thus, there is also a need to further understand and analyse contextual
information based on specific tourist profiles.
This study forms part of our ongoing research to determine effective mobile user
interface designs supporting features associated with contextual information. This
paper proposed a set of contextual information needed to support mobile tourism
applications, which will be used as the basis of the design of our user interface in the
next stage of work. This next stage will focus on using social contextual information to
filter information. It is expected to lead to the development of effective user interface
designs for visualising and manipulating social contextual information. This aspect of
work on social context based design and research is important as we embrace social
computing on an increasing basis in the future.
References
Albers, M. and Kim, L. (2002), “Web design issues when searching for information using
handheld interfaces”, Technical Communication, Vol. 49 No. 3, pp. 314-29.
AP
61,6
582
Borlund, P. and Ingwersen, P. (1997), “The development of a method for the evaluation of
interactive information retrieval systems”, Journal of Documentation, Vol. 53 No. 3,
pp. 225-50.
Cano, J., Manzoni, P. and Toh, C.K. (2006), “Ubiqmuseum: a Bluetooth and java based
context-aware system for ubiquitous computing”, Wireless Personal Communications,
Vol. 38 No. 2, pp. 187-202.
Cheverst, K., Mitchell, K. and Davies, N. (2002), “Exploring context-aware information push”,
Personal and Ubiquitous Computing, Vol. 6 No. 4, pp. 276-81.
Cheverst, K., Davies, N., Mitchell, K. and Friday, A. (2000a), “Experiences of developing and
deploying a context-aware tourist guide: the GUIDE project”, Proceedings of the 6th
Annual International Conference on Mobile Computing and Networking, Boston, ACM
Press, New York, NY, pp. 20-31.
Cheverst, K., Mitchell, K., Davies, N. and Smith, G. (2000b), “Exploiting context to support social
awareness and social navigation”, ACM SIGGROUP Bulletin, Vol. 21 No. 3, pp. 43-8.
Cheverst, K., Davies, N., Mitchell, K., Friday, A. and Efstratous, C. (2000c), “Developing a
context-aware electronic tourist guide: some issues and experiences”, Proceedings of the
SIGCHI Conference on Human Factors in Computing Systems, The Hague, ACM Press,
New York, NY, pp. 17-24.
Christensen, J., Sussman, J., Levy, S., Bennett, E.B., Wolf, T.V. and Kellogg, W.A. (2006), “Too
much information”, ACM Queue, July/August, pp. 51-7.
Dey, A.K. and Abowd, G.D. (2000), “Towards a better understanding of context and
context-awareness”, GVU Technical Report GIT-GVU-99-22, available at: http://dsv.su.se/
fuse/int8/docs/context.pdf (assessed 10 August 2009).
DK Eyewitness (2003), DK Eyewitness Guide for San Francisco, Dorling Kindersley, London.
Dornbush, S., English, J., Oates, T., Segall, Z. and Joshi, A. (2007), “XPod: a human activity aware
learning mobile music player”, Proceedings of the Workshop on Ambient Intelligence, 20th
International Joint Conference on Artificial Intelligence, Hyderabad, UMBC Press,
Baltimore, MD, pp. 324-31.
Elliott, G.T. and Tomlinson, B. (2006), “Personal soundtrack: context-aware playlists that adapt
to user pace”, Proceedings of Conference on Human Factors in Computing System, Quebec,
ACM Press, New York, NY, pp. 736-41.
European Media Laboratory (2006), “CRUMPET”, available at: www.eml-development.de/
english/research/crumpet/index.php (accessed 10 December 2006).
Fan, J., Gao, Y. and Luo, H. (2006), “Hierarchical classification for automatic image annotation”,
Proceedings of the 30th Annual ACM SIGIR Conference on Research and Development in
Information Retrieval, Amsterdam, ACM Press, New York, NY, pp. 111-8.
Feng, L., Apers, P. and Jonker, W. (2004), “Towards context-aware data management for ambient
intelligence”, Proceedings of the 15th International Conference on Database and Expert
Systems Applications (DEXA), Zaragoza, Springer, Heidelberg, pp. 1036-41.
Fogarty, J., Lai, J. and Christensen, J. (2004), “Presence versus availability: the design and
evaluation of a context-aware communication client”, International Journal of Human
Computer Studies, Vol. 61 No. 3, pp. 299-317.
Frommer (2007), Frommer’s San Diego 2007, Wiley, Hoboken, NJ.
Ghosh-Dastidar, S., Adeli, H. and Dadmehr, N. (2007), “Mixed-band wavelet-chaos-neural
network methodology for epilepsy and epileptic seizure detection”, IEEE Transactions on
Biomedical Engineering, Vol. 54 No. 9, pp. 1545-51.
Classifying
contextual
information
583
Goker, A. and Myrhaug, H. (2008), “Evaluation of a mobile information system in context”,
Information Processing and Management, Vol. 44 No. 1, pp. 39-65.
Hinze, A. and Buchanan, G. (2005), “Context-awareness in mobile tourism information systems:
challenges for user interaction”, Proceedings of the Workshop on Context in Mobile HCI, in
Conjunction with Mobile HCI, Salzburg, ACM Press, New York, NY, pp. 257-66.
Hinze, A. and Voisard, A. (2003), “Location and time-based information delivery in tourism”,
Proceedings of Conference of Advances in Spatial and Temporal Databases (SSTD),
Santorini Island, Springer, Heidelberg, pp. 489-507.
Hsi, S. (2002), “electronic guidebook: a study of user experiences using mobile web content in a
museum setting”, Proceedings of the IEEE International Workshop on Wireless and Mobile
Technologies in Education (WMTE), Va
¨xjo
¨, IEEE Press, Piscataway, NJ, pp. 48-54.
Jeong, C. and Lee, E. (2007), “Context aware human computer interaction for ubiquitous
learning”, Proceedings of Conference on HCI International, Human Interface, Part II,
Beijing, Lecture Notes in Computer Science, Vol. 4558, Springer, Heidelberg, pp. 364-73.
Jose, R., Moreira, A., Rodrigues, H. and Davies, N. (2003), “The AROUND architecture for
dynamic location-based services”, Mobile Networks and Applications, Vol. 8 No. 4,
pp. 377-87.
Jung, T., Makeig, S., Stensmo, M. and Sejnowski, T. (1997), “Estimating alertness from the EEG
power spectrum”, IEEE Transactions on Biomedical Engineering, Vol. 44 No. 1, pp. 60-9.
Kaasinen, E. (2003), “User needs for location-aware mobile services”, Personal and Ubiquitous
Computing, Vol. 7 No. 1, pp. 70-9.
Kamar, A. (2003), “Mobile tourism guide (m-ToGuide), Deliverable 1.4, Project final report”,
available at: ftp://ftp.ifs.uni-linz.ac.at/pub/diplomathesis/gruen.pdf (accessed 1 March
2007).
Kawash, J., Morr, C.E. and Itani, M. (2007), “A novel collaboration model for mobile virtual
communities”, International Journal of Web Based Communities, Vol. 3 No. 4, pp. 427-47.
Kjeldskov, J. and Skov, M.B. (2004), “Supporting work activities in healthcare by mobile
electronic patient records”, Proceedings of the 6th Asian Pacific Conference on Computer
Human Interaction, Rotorua, Springer, Heidelberg, pp. 191-200.
Korkea-aho, M. (2000), “Context-aware applications survey”, available at: http://users.tkk.fi/
,mkorkeaa/doc/context-aware.html (accessed 2 October 2007).
Korpipaa, P., Mantyjarvi, J., Kela, J., Keranen, H. and Malm, E. (2003), “Managing context
information in mobile devices”, Pervasive Computing, July-September, pp. 42-51.
Lombard, M., Snyder-Duch, J. and Bracken, C.C. (2002), “Content analysis in mass
communication: assessment and reporting of intercoder reliability”, Human
Communication Research, Vol. 28 No. 4, pp. 587-604.
Mizzaro, S. (1997), “Relevance: the whole history”, Journal American Society for Information
Science, Vol. 48 No. 9, pp. 810-32.
Mountain, D. and Macfarlane, A. (2007), “Geographic information retrieval in a mobile
environment: evaluating the needs of mobile individuals”, Journal of Information Science,
Vol. 33 No. 5, pp. 515-30.
Myrhaug, H., Whitehead, N., Goker, A., Faegri, T.E. and Lech, T.C. (2004), “AmbieSense – a
system and reference architecture for personalised context-sensitive information services
for mobile users”, in Markopoulos, P. (Ed.), Ambient Intelligence, Lecture Notes in
Computer Science, Vol. 3295, Springer, Heidelberg, pp. 327-38.
AP
61,6
584
O’Grady, M.J., O’Hare, G.M.P. and Donaghey, C. (2007), “Delivering adaptivity through
context-awareness”, Journal of Network and Computer Applications, Vol. 30 No. 3,
pp. 1007-33.
Pashtan, A., Blattler, R., Heusser, A. and Scheuermann, P. (2003), “CATIS: a context aware
tourist information system”, Proceedings of the 4th International Workshop of Mobile
Computing, Rostock, available at: www.ece.northwestern.edu/ ,peters/references/IMC.
CATIS.pdf (accessed 1 March 2007).
Peng, X.G. and Choi, B. (2002), “Automatic web page classification in a dynamic and hierarchical
way”, Proceedings of Second IEEE International Conference on Data Mining (ICDM’02),
Maebashi City, ACM Press, New York, NY, pp. 386-93.
Perttunen, M. and Riekki, J. (2005), “Introducing context-aware features into everyday mobile
applications”, Proceedings of Conference of Location and Context-awareness (LoCA),
Oberpfaffenhofen, Springer, Heidelberg, pp. 316-27.
Poslad, S., Laamanen, H., Malaka, R., Nick, A., Buckle, P. and Zipf, A. (2001), “CRUMPET:
Creation of User-friendly Mobile services Personalised for Tourism”, Proceedings of 2nd
International Conference on 3G Mobile Communication Technologies, London, IEEE Press,
Piscataway, NJ, pp. 28-32.
Pousman, Z., Iachello, G., Fithian, R., Maghazy, J. and Stasko, J. (2004), “Design iterations for a
location-aware event planner”, Personal Ubiquitous Computing, Vol. 8 No. 2, pp. 117-25.
Raper, J. (2007), “Geographic relevance”, Journal of Documentation, Vol. 63 No. 6, pp. 836-52.
Raper, J., Gartner, G., Karimi, H. and Rizos, C. (2007), “Applications of location based services:
a selected review”, Journal of Location Based Services, Vol. 1 No. 2, pp. 89-111.
Raskovic, D., Martin, T. and Jovanov, E. (2004), “Medical monitoring applications for wearable
computing”, The Computer Journal, Vol. 47 No. 4, pp. 495-504.
Roffia, L., Raffa, G., Pettinar, M. and Gaviani, G. (2005), “Context awareness in mobile cultural
heritage applications. Smart environments and their applications to cultural heritage”,
Proceedings of the 7th International Conference on Ubiquitous Computing, ACM Press,
New York, NY, pp. 33-6.
Roth, J. (2002), “Context-aware web applications using the pinpoint”, Proceedings of Conference
on IADIS, Lisbon, IADIS Press, pp. 3-10.
Saracevic, T. (1996a), “Relevance reconsidered”, Proceedings of the 2nd Conference on
Conceptions of Library and Information Science, Copenhagen, Information Research,
London, pp. 201-18.
Saracevic, T. (1996b), “Modeling interaction in information retrieval (IR): a review and proposal”,
Proceedings of the 59th Annual Meeting of the American Society for Information Science,
Medford, NJ, Vol. 34, ERIC, Washington, DC, pp. 3-9.
Schilit, B.N., Adams, N. and Want, R. (1994), “Context-aware computing applications”,
Proceedings of Conference IEEE Workshop on Mobile Computing Systems and
Applications, Santa Cruz, CA, IEEE Press, Piscataway, NJ, pp. 85-90.
Setten, M., Pokraev, S. and Koolwaaij, J. (2004), “Context-aware recommendations in the mobile
tourism application COMPASS”, Proceedings of 3rd International Conference Adaptive
Hypermedia and Adaptive Web-based Systems, Eindhoven, Springer, Heidelberg,
pp. 235-44.
Spink, A., Greisdorf, H. and Bateman, J. (1998), “From highly relevant to not relevant: examining
different regions of relevance”, Information Processing and Management, Vol. 34 No. 5,
pp. 599-621.
Classifying
contextual
information
585
Wang, X.H., Zhang, D.Q., Gu, T. and Pung, H.K. (2004), “Ontology based context modeling and
reasoning using OWL”, Proceedings of the 2nd IEEE Annual Conference on Pervasive
Computing and Communications Workshop, Orlando, FL, IEEE Press, Piscataway, NJ,
pp. 18-22.
Wilson, P. (1973), “Situational relevance”, Information Storage and Retrieval, Vol. 9 No. 8,
pp. 457-71.
YES (2007), Your Essential Guide for Singapore, Singapore Tourism Board, Singapore.
AP
61,6
586
To purchase reprints of this article please e-mail: reprints@emeraldinsight.com
Or visit our web site for further details: www.emeraldinsight.com/reprints
