WEB INFORMATION GATHERING TASKS AND THE USER
SEARCH BEHAVIOUR
Anwar Alhenshiri, Michael Shepherd, Carolyn Watters and Jack Duffy
Faculty of Computer Science, Dalhousie University, 6050 University Ave., Halifax, NS, Canada
Keywords: Information Retrieval, Web, Tasks, Information Gathering, Behaviour.
Abstract: The research described in this article is an attempt to characterize the kind of search behaviour users follow
while gathering information on the Web. Information gathering on the Web is a task in which users collect
information; possibly from different sources (pages); more likely over multiple sessions to satisfy certain
requirements and goals. This process involves decision making and organization of the information gathered
for the task. Information gathering tasks have been shown to be search-reliant. Therefore, identifying the
kind of search behaviour users choose for this kind of task may lead to supporting Web information
gathering tools as recommended in the findings of this research. The results of the user study reported in this
paper indicate that the user search behaviour during Web information gathering tasks has characteristics of
both orienteering and teleporting behaviours.
1 INTRODUCTION
To categorize user activities on the Web, researchers
often apply models of information seeking (Ellis,
1993; Marchionini, 1997; Choo et al., 1998).
However, because Web users and Web technologies
evolve rapidly, those models may be obsolete. The
content of the Web—as well as its users—change
over time due to the emergence of new genres,
topics, and communities on the Web (Santini, 2006).
Existing information seeking models have attempted
to categorize user activities. More recent models
have emerged to focus on the narrower behaviour of
users with particular tasks.
There have been different studies in which the
types of activities users perform on the Web were
identified and categorized into higher level tasks.
Examples of models concerning user tasks on the
Web include Broder’s taxonomy (Broder, 2002),
Rose and Levinson’s classification (Rose and
Levinson, 2004), Sellen’s model (Sellen et al.,
2002), and Kellar’s categorization of information
seeking tasks (Kellar et al., 2007). The results of
those studies indicate that each task can be further
studied for understanding the subtasks involved in
the overall task.
Alhenshiri et al. (2010) presented a model in
which the task of Web information gathering was
divided into subtasks each of which involves
activities of similar nature that users perform on the
Web during the task. The process of information
gathering on the Web has been shown to heavily rely
on search and organization of information for the
task (Alhenshiri et al., 2011). The search part of the
process includes activities users perform to locate
pieces of information required in the task which may
involve locating information from different sources,
locating related information to the already located
pieces, and re-finding information in multi-session
tasks (Alhenshiri et al., 2010).
When searching for information on the Web,
users orienteer, teleport, or do both (Teevan et al.,
2004). In the former, users start at a certain page (or
site) and continue searching for information by
following the hierarchy of hyperlinks to find
relevant information. In the latter, users rely heavily
on frequent submissions of search queries to search
engines (or through search features provided on Web
pages) to find relevant information. These two types
of behaviour have been studied by Teevan et al.
(2004) who showed that 61% of user search
activities did not involve keyword search, denoting
orienteering behaviour. Only in 39% of the search
activities, teleporting behaviour was involved.
This paper re-examines the findings of Teevan et
al. (2004) in the case of searching for information
during information gathering tasks on the Web. This
323
Alhenshiri A., Shepherd M., Watters C. and Duffy J..
WEB INFORMATION GATHERING TASKS AND THE USER SEARCH BEHAVIOUR.
DOI: 10.5220/0003910303230331
In Proceedings of the 8th International Conference on Web Information Systems and Technologies (WEBIST-2012), pages 323-331
ISBN: 978-989-8565-08-2
Copyright
c
 2012 SCITEPRESS (Science and Technology Publications, Lda.)

paper builds on the findings in Alhenshiri et al.
(2010) and investigates the characteristics of user
search behaviour during Web information gathering
tasks. The study described in this paper was also
intended for investigating other aspects of
information gathering on the Web that are reported
in Alhenshiri et al. (2012). The research in this paper
attempts to answer the following questions: (i) Do
users gathering Web information follow a specific
kind of search behaviour (orienteering or
teleporting)? And how can identifying the user
search behaviour benefit the design of future
information gathering tools intended for the Web?
The paper is organized as follows. Section 2
explores the research rationale. Section 3 illustrates
the research study. Section 4 discusses the study
results and findings. The paper is concluded in
Section 5.
2 RESEARCH RATIONALE
Information seeking models have focused on
identifying activities users perform while they
attempt to locate information of interest. The Web
has been treated as a special case in some of the
older models such as Ellis’s (1993). Ellis (1993)
concluded that there are several main activities
applicable to hypertext environments of which the
Web is one. Those activities represent user actions
during seeking information that is not previously
known to the user and which is aimed to increase the
user’s knowledge. Marchionini (1997) stated that the
process of information seeking consists of several
activities (sub-processes) that start with the
recognition and acceptance of the problem and
continues until the problem is either resolved or
abandoned. Wilson and Walsh’s (1996) model of
information behaviour differs from many of the prior
models since it suggests high-level information
seeking search processes: passive attention, passive
search, active search, and on-going search. Although
these models provide accurate characterizations of
users’ information seeking activities, several
activities that users perform on the Web usually are
not included in the model. The variations of those
models and the continuous modifications make it
difficult to choose an appropriate characterization.
Several other frameworks have been suggested to
understand and model the different activities users
perform specifically on the Web while seeking
information. Rose and Levinson (2004) attempted to
identify a framework for user search goals using
ontologies in order to understand how users interact
with the Web. Their findings indicated that users'
goals can be informational, navigational, or
transactional. Similarly, Sellen et al. (2002) found
that user activities can be categorized into finding,
information gathering, browsing, performing a
transaction, communicating, and housekeeping.
Moreover, Broder (2002) studied different user
interactions during Web search and identified three
types of tasks based on the queries submitted by
users. Those types are: navigational, informational,
and transactional. In addition, Kellar et al. (2007)
investigated user activities on the Web to develop a
task framework. The results of their study indicated
that the four types of Web tasks are: fact finding,
information gathering, browsing, and transactions.
Based on the different classifications of Web
tasks, research showed that information gathering
tasks represent a great deal of the overall tasks on
the Web (61.5% according to Rose and Levinson,
2004). Therefore, Alhenshiri et al. (2010) developed
a model in which the subtasks underlying the overall
task of information gathering were identified. Their
research indicated that information gathering is
heavily search-reliant. Prior to this model, Amin
(2009) identified different characteristics in Web
information gathering tasks. Information gathering
was shown to be a more complex task than keyword
search tasks. The terms ‘information gathering’
imply different kinds of search including
comparison, relationship, exploratory, and topic
searches as well as combinations of more than one
type of search (Amin, 2009). Information gathering
tasks are characterized, in part, by having high-level
goals and requiring the use of multiple information
resources (Alhenshiri et al., 2012).
Teevan et al. (2004) identified two types of
search behaviour (viz. teleporting and orienteering)
in e-mails, personal documents, and the Web. In the
former, a searcher is most likely to use keywords
while seeking information. In the latter, a sequence
of steps and strategies is adopted to reach the
intended information, i.e. usually by starting search
at a particular URL and continuing on the Web
hierarchy by following links on Web pages. In this
paper, the two types of behaviour are further
considered in the case of gathering information from
the Web. The goal of this consideration is to decide
on the significance of which type of behaviour for
the information gathering tasks and to eventually
recommend design properties for tools intended for
Web-based information gathering tasks.
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
324

3 RESEARCH STUDY
Information gathering tasks have been shown to be
heavily search-reliant (Amin, 2009; Alhenshiri et al.,
2012) and very popular on the Web as discussed
above. Therefore, the user study discussed in this
section was conducted. The study was meant to
conclude on the kind of behaviour users follow
when performing Web information gathering tasks
which would lead to developing support for the
design of tools intended for this type of task. To
identify the kind of search behaviour users followed
during the task of information gathering, the analysis
in the study considered: (i) the number of URLs
users typed-in to start searching for information; (ii)
the number of keyword queries they submitted; (iii)
the number of links they followed on the Web
hierarchy to locate information for the task; and (iv)
correlations among those factors.
3.1 Study Design and Population
The design of the study was complete factorial and
counter-balanced with random assignment of tasks
to participants. There were 20 participants in the
study, equally split between graduate and
undergraduate students in Computer Science at
Dalhousie University. The study used a special
version of the Mozilla Firefox browser
(http://www.mozilla.com)
called DeerParkLogger,
which was designed at Dalhousie University. This
browser has the ability to log all user interactions
during the task.
3.2 Study Tasks
The study used four information gathering tasks that
were similar in terms of the complexity of the task
and different with regard to the task topic. Each task
was created following the guidelines described by
Kules and Capra (2008) and summarized in the
following:
• The task description should indicate uncertainty,
ambiguity in information need, or need for
discovery.
• The task should suggest knowledge acquisition,
comparison, or discovery.
• It should provide a low level of specificity about
the information required in the task and how to
find such information.
• It should provide enough imaginative contexts for
the study participants to be able to relate and apply
the situation.
To ensure the equality of the tasks with regard to
the complexity level, a focus group met twice to
analyze the tasks and make the necessary
modifications based on: the time needed to complete
the task, the amount of information required to be
gathered, the clarity of the task description, and the
possible difficulties that the user may encounter
during gathering.
3.2.1 Information Gathering Task Example
Part 1. You heard your friends complaining about
bank account service charges in Canada. You are not
sure why they are complaining. You want to do
research on the Web to find out more about bank
account service charges in Canada. State your
opinion about the charges and your friends’
complaints. Keep a copy of the information you
found to support your argument. Provide at most
five links to pages where you found the information.
Keep the information for possible re-use in a
subsequent task.
Part 2. After you found out about the bank service
charges in Canada, you want to compare account
service charges of Canadian banks to those applied
by American banks. Search the Web to find
information about banks in the US. Find at most
information from five pages on the Web. Provide a
comparison of service charges in both countries. Use
the information you kept in the previous task about
the Canadian banks. You should keep a copy of all
relevant information you found for both tasks.
3.3 Study Methodology
Every participant was randomly assigned two tasks
each of which was divided into two parts as shown
in the example above. The reason for splitting each
task was to encourage participants to re-find
information for the second subtask that was
preserved (kept) during the first subtask. The issue
of re-finding is beyond the scope of this paper. Other
aspects including re-finding information are reported
in Alhenshiri et al. (2012). The study had two
questionnaires, a pre-study and two post-task
questionnaires. All user activities were logged
during the study for further analysis.
3.4 Study Results
The user behaviour and its correlation with the kind
of activities users perform during the task of
information gathering were expected to yield certain
findings that would help with the design of future
gathering tools. The results reported in this paper
WEBINFORMATIONGATHERINGTASKSANDTHEUSERSEARCHBEHAVIOUR
325

concern attempts to identify the user search
behaviour during Web information gathering tasks.
Users in the study followed either or both of two
types of search behaviour that were discussed in the
work of Teevan et al. (2004). Those types are
orienteering and teleporting. In the former, a user
starts the search at a specific URL, and continues by
following links on Web pages to find and gather
information. Users of this type of behaviour are
usually expected to follow more links on the Web
and submit fewer search queries to search engines.
In the latter, the user tends to rely on the submission
of search queries more often to locate information.
The user in this case relies less on following
hyperlink connectivity on the Web.
To decide on the type of behaviour users
followed during the tasks, the analysis of the data
considered the number of URLs typed-in, the
number of search queries submitted, the number of
links followed during the task (click behaviour), and
correlations among those factors.
3.4.1 Using Typed-in URLs
The analysis of the data took the number of URLs
participants typed in to start searching for the task
requirements as a distinguishing factor between
orienteering and teleporting behaviour users. Based
on the average URLs typed in, 70% of the study
participants (14 users) were identified to have
followed teleporting behaviour to accomplish the
tasks. Only 30% (six users) were identified to have
followed orienteering behaviour. The difference
between the two proportions of participants was
significant according to the z-test results (z=1.96,
p<0.03). The actual data regarding the typed-in
URLs from the study are shown in Figures 1 and
Table 1. Six users who typed-in more URLs (above
average) were considered teleporting behaviour
users while the remaining users were considered
orienteering behaviour users. Due the fact that the
average URLs typed in did not draw a clear line
between two completely different kinds of behaviour
based on the data in the study, the analysis went to a
different criterion and the number of queries
submitted was tried as a distinguishing factor
between the two kinds of search behaviour.
3.4.2 Using Submitted Queries
The second factor used to determine which
proportion of participants followed which kind of
search behaviour during the study was the number of
queries submitted for accomplishing the tasks. As
shown
in Figure 2 and further illustrated in Table 2,
Figure 1: URLs typed-in by users to start searching for
information.
Table 1. Typed-in URLs.
Type of
behaviour
Participant
Number of URLs
typed in
participants
identified as
orienteering
behaviour
users
P2
17
P10
13
P11
8
P1
6
P20
6
P3
5
participants
identified as
teleporting
behaviour
users
P14
4
P8
3
P12
3
P13
3
P15
3
P6
2
P17
2
P8
2
P7
1
P9
1
P19
1
P4
0
P5
0
P16
0
࢞
ഥ
4
s 4.3
by taking the average number of queries submitted
during the study as a distinguishing factor, half of
the participants were considered as orienteering
behaviour users while the other half as teleporting
behaviour users. As a result, the two groups
resulting from using the number of queries
submitted as a distinguishing factor did not agree
with the two groups that resulted from using the
number of typed-in URLs. The analysis used the
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
326

average number of queries to distinguish users with
the two types of search behaviour which was not a
reliable choice due to the closeness of the numbers
of queries in each group to the average.
Table 2: Queries submitted during the study.
Type of
behaviour
Participant
Number of queries
submitted
participants
identified as
teleporting
behaviour
users
P4 23
P7 19
P10 15
P11 15
P6 13
P20 9
P5 8
P8 8
P9 8
P17 8
participants
identified as
orienteering
behaviour
users
P18 7
P19 6
P13 5
P15 5
P12 4
P16 4
P3 3
P1 1
P2 0
P14 0
Since the analysis yielded different
categorization in the case of using search queries as
an alternative to URLs typed in by the user, the
number of links followed by users in the latter case
was considered for analysis. The reason why the
number of links followed on the Web hierarchy was
considered in the case of using search queries only
and not in the case of URLs typed in is the number
of participants that would result from the
classification. In the case of using URLs typed in,
the number of orienteering behaviour users turned
out to be too small (only six participants). The use of
such small group may yield insignificant findings
when taking a step further in the analysis by
involving the links followed on the Web hierarchy
during the study. However, the use of queries
submitted as a distinguishing factor between
orienteering and teleporting behaviour users created
two similar groups (10 participants in each).
Therefore, it was selected with the analysis of linked
followed.
3.4.3 Number of Links Followed
Furthermore, by looking at the number of links each
group (orienteering or teleporting) followed on the
Web during the study, there was almost no
difference between the two groups of participants
distinguished by query submissions (ANOVA,
F(1,18)=1.81, p=0.19) as shown in Table 3.
Table 3: Links followed by users: the case of using search
queries.
A
bove avera
g
e queries
(Teleporting)
Below avera
g
e
queries
Par
t
ici
p
an
t
lin
k
s
Par
t
ici
p
an
t
lin
k
s
P4 84 P18 61
P7 46 P19 60
P10 28 P13 73
P11 77 P15 22
P6 41 P12 2
P20 25 P16 22
P5 90 P3 53
P8 34 P1 11
P9 76 P2 42
P17 49 P14 57
࢞
ഥ
55
࢞
ഥ
40.3
s 24.4 s 24.3
ANOVA
,
f
=1
.
81
,
p
=
0
.
19
This finding indicates that: either users’
behaviour had characteristics of both orienteering
and teleporting search; or the average number of
search queries did not suffice for distinguishing the
‘expected’ two groups of users. Theoretically,
orienteering behaviour users submit fewer queries
than teleporting behaviour users. The difference was
between the number of queries submitted by the two
groups was significant according to a single-factor
ANOVA (F(1,18)=23.82, p<0.0002). Nonetheless,
the difference with regard to the number of links
followed was not significant.
3.4.4 Measuring Correlations
To further ensure that the user search behaviour was
hard to identify in the case of Web information
gathering tasks in the study, the analysis of the data
involved measuring the correlation between the
number of typed-in URLs and the number of queries
submitted by the study participants. We used the
Pearson Product Moment correlation test. We
considered measuring the correlation between
queries submitted and URLs typed in for all users at
first and then we followed by measuring the
correlations for each group of users identified as
either orienteering or teleporting users using the
WEBINFORMATIONGATHERINGTASKSANDTHEUSERSEARCHBEHAVIOUR
327

number of typed-in URLs and then using the number
of queries submitted.
The results concerning the correlation between
typed-in URLs and queries submitted during the
study for the entire group of users showed that there
was a very strong positive correlation between the
two groups of data (r=0.95, p<0.00001). Please refer
to Tables 1 and 2 for data. This relationship
contradicts the expected since a strong positive
correlation means that the more queries users
submitted, the more URLs they typed in while
gathering the information. This can be related to the
nature of the user and their activities during the
study. However, it is hard to distinguish one kind of
behaviour or the other as a result of this relationship.
For further assurance, we tackled the issue from a
different perspective by considering that there
actually exist two groups of users with two different
types of behaviour. Those two groups are first
distinguished by the number of URLs typed in, and
second by the number of queries submitted.
Table 4: Pearson (r) correlation test results in the case of
using typed-in URLs (teleporting).
Teleportin
g
Partici
p
ants
Queries
submitted
URLs typed-in
P14 0 4
P8 3 3
P12 1 3
P13 2 3
P15 2 3
P6 6 2
P17 3 2
P18 3 2
P7 13 1
P9 3 1
P19 2 1
P4 17 0
P5 4 0
P16 1 0
Pearson Product Moment ( r = -0.5, p<0.07 )
The results of the Pearson test shown in Table 4
indicate that there was a moderate relationship
between the number of URLs typed in and the
number of queries submitted with inverse
association between the two variables. The
participants shown in Table 4 are those initially
identified as teleporting behaviour users using the
number of typed-in URLs. For orienteering
behaviour users, the results of the Pearson test are
shown in Table 5. Those results indicate that almost
no correlation exists between the queries submitted
and the URLs typed-in.
Table 5: Pearson (r) correlation test results in the case of
using typed-in URLs (orienteering).
Orienteerin
g
Partici
p
ants
Queries
submitted
URLs typed-in
P2 0 17
P10 8 13
P11 6 8
P1 0 6
P20 5 6
P3 1 5
Pearson Product Moment (r = 0.04, p<0.94)
Table 6: Pearson (r) correlation test results in the case of
using submitted queries (teleporting).
Teleportin
g
p
artici
p
ants
Queries
submitted
URLs t
y
ped
in
P4 23 0
P7 19 1
P10 15 13
P11 15 8
P6 13 2
P20 9 6
P5 8 0
P8 8 3
P9 8 1
P17 8 2
Pearson Product Moment (r=0.04, p<0.91)
The analysis went to the use of the number of
queries to decide on the two groups of users
expected to follow one kind of behaviour or the
other. The data is shown in Tables 6 and 7. There
was almost a zero correlation between the submitted
queries and the typed-in URLs in the case of
participants identified as teleporting behaviour users
using the number of queries submitted as a
distinguishing factor (Table 6). In the case of
orienteering behaviour users identified also using the
number of queries submitted, the correlation was
strong indicating that an inverse relationship existed
(Table 7). However, this was only for half the
number of participants since in the case of the rest of
participants the correlation was close to zero.
The use of the correlation tests was a different
investigation step to ensure that the search behaviour
of the users in the study—while performing the
given information gathering tasks—was hard to
identify as either orienteering or teleporting. To this
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
328

Table 7: Pearson (r) correlation test results in the case of
using submitted queries (orienteering).
Orienteerin
g
p
artici
p
ants
Queries
submitted
URLs t
y
ped
in
P18 7
2
P19 6
1
P13 5
3
P15 5
3
P12 4
3
P16 4
0
P3 3
5
P1 1
6
P2 0
17
P14 0
4
Pearson Product Moment (r= -0.67, p<0.04)
point, the findings indicate that users’ search
behaviour may have had characteristics of both
orienteering and teleporting behaviours. However,
the use of averages (URLs typed in or queries
submitted) may not be sufficient. For example, it
might have not been invalid to put a user who
submitted seven queries (too close to the average of
eight queries) in the section of orienteering
behaviour users only because of a one-query
difference. Therefore, we selected another portion of
users in the study that is not centred around the
mean (i.e. outliers) even though we expected not to
have enough participants in groups categorized as
outliers.
3.4.5 Using Outliers with Correlations
Even though the use of correlations between queries
submitted and URLs typed-in by users during the
study further demonstrated that it was hard to draw a
line between orienteering and teleporting behaviour
users in the study, we took the investigation a step
further. In this step, the outliers in both cases: the
typed-in URLs and the queries submitted during the
study were considered.
In the case of using typed-in URLs, the outliers
were taken apart from the rest of the data by
considering numbers of URLs greater than 1.5 the
upper quartile (from Tables 1) and numbers of URLs
less than 1.5 the lower quartile. The results of this
selection are shown in Table 8. This table contains
the outliers with respect to the number of URLs
typed-in on both sides (shaded for clarification). The
table also contains the number of queries submitted
by each participant and the number of links followed
on the Web hierarchy.
Table 8: Outliers data (typed-in URLs).
Participant
Typed-in
URLs
Submitted
Queries
Links
Followed
P16 0 4 22
P5 0 8 90
P4 0 23 84
P19 1 6 60
P9 1 8 76
P7 1 19 46
P11 8 15 77
P10 13 15 28
P2 17 0 42
To ensure whether one type of behaviour or the
other (orienteering/teleporting) was followed, three
correlations were computed using Pearson Product
Moment. The correlation between the number of
typed-in URLs and the number of submitted queries
turned out to be weak and negative (r = -0.25,
p=0.51). The correlation between the number of
typed-in URLs and followed links was also weak
(r=0.39, p=0.29). The correlation between the
submitted queries and the followed links was weak
(r=0.29 and p=0.44).
The results show that there was no indication of
any specific type of behaviour by any group of
users. The weak correlations demonstrate that no
relationship can be explained by any of the factors
involved in the correlations except for the
relationship between queries submitted and links
followed which turned out to be weak. Users who
follow teleporting behaviour by relying on query
submissions usually tend to follow fewer links on
the hierarchies of websites than users who start
searching by typing in URLs. However, users who
relied on typing in URLs were not shown to have
made a significant use of the strategy of following
link hierarchies on the Web as shown by the test
results.
Furthermore, the analysis considered the outliers
in the case of using the number of queries submitted
by users during the study. The results are shown in
Table 9. The table contains the number of queries
(for outliers only) submitted by participants
associated with the URLs typed-in and links
followed for each participant. The correlations
between each two of the three factors were
computed using Pearson Product Moment. The
results showed that the correlation between the
number of submitted queries and typed-in URLs was
weak (r=0.41, p=0.31). The correlation between the
WEBINFORMATIONGATHERINGTASKSANDTHEUSERSEARCHBEHAVIOUR
329

submitted queries and the followed links was
moderate and positive (r=0.57, p=0.14). The
correlation between the typed-in URLs and the
followed links was weak and negative (r=-0.25,
p=0.55).
Table 9: Outlier data (submitted queries).
Participant
Submitted
Queries
Typed-in
URLs
Followed
Links
P4 23 0 84
P7 19 1 46
P10 15 13 28
P11 15 8 77
P3 3 5 5
P1 1 6 6
P2 0 17 42
P14 0 4 57
Orienteering behaviour users rely usually on
typing URLs for starting search for information on
the Web. They also follow links on Web pages to
locate information of the interest. The weak and
negative correlation between URLs typed in and
links followed contradicts the definition of
orienteering behaviour. Actually, a stronger
relationship can be seen in the correlation between
submitted queries and followed links, which is
contradictory to the teleporting search behaviour
definition. The only correlation that agrees with the
definitions of search behaviours (orienteering vs.
teleporting) is the correlation between queries
submitted and URLs typed in. Nonetheless, it was a
weak relationship.
4 DISCUSSION
The study used the number of typed-in URLs, the
number of search queries submitted, and the number
of links followed on the Web hierarchy during the
tasks in order to identify the type of behaviour users
followed while performing information gathering
tasks during the study. The results showed that
neither factor was sufficient to make a clear
distinction between the two groups of users with
respect to the search behaviour during the tasks. To
further ensure that no clear signs of either behaviour
could be identified among participants in the study,
the correlation between the typed-in URLs and the
search queries submitted during the study was
measured for the entire group of users, the two
groups distinguished by the number of URLs typed
in, and the two groups distinguished by the number
of queries submitted.
According to the results of the correlation tests, it
was hard to identify which group of participants
followed which type of search behaviour while
performing the information gathering tasks given
during the study. The initial idea behind orienteering
and teleporting behaviours is that one is different
from the other. Users who follow orienteering
behaviour are those who type-in URLs more often
and follow hyperlink connectivity on the Web to
search for information. Users who follow teleporting
behaviour usually rely on the submission of search
queries in order to find information. This type of
users hardly starts searching at a certain URL and
barely follows links on Web pages using a series of
clicks to locate information.
Every time the analysis of the study data
considered one criterion to make a distinction
between the two kinds of behaviour amongst the
study participants, it was hard to conclude on which
group followed which kind of search behaviour. The
results of the analysis indicate that activities users
perform during this kind of task belong to both kinds
of behaviour. Therefore, the type of search
behaviour had no effect on the task and was not
affected by the nature of the Web information
gathering tasks.
Even with the selection of a subset of users that
represented only the outliers in the cases of typed-in
URLs and submitted queries, the correlations
computed among submitted queries, typed-in URLs,
and followed links did not demonstrate that one kind
of search behaviour was dominant in the case of any
group of participants. Interestingly, the relationship
between query submission and following links on
the Web was moderate showing that the same users
had two features from two different kinds of search
behaviour (Table 8).
As a result of the study, any support for
information gathering tasks in terms of building
tools for the task should consider both characteristics
of the two kinds of behaviour. The design should
take into account that users gathering information on
the Web using the current available tools may adopt
varied strategies and use several techniques and
features to accomplish the goal of the task. Users
submit queries at different levels of frequency, open
browser tabs and windows, compare information,
collect information from both actively open Web
pages in the browser and search hits’ summaries,
and use different tools to accomplish the task. They
use search engines and type in URLs to start
searching on the Web hierarchy by following links
on Web pages and sites.
WEBIST2012-8thInternationalConferenceonWebInformationSystemsandTechnologies
330

In future designs of Web tools intended for
information gathering, support should be provided
for allowing users to open multiple URLs in a way
that eases the information comparison process with
which users usually have difficulties when using
browser tabs and windows. Support should also be
provided to users submitting several queries
simultaneously to compare result hit summaries.
Those users used browser tabs and windows and lost
track of information on several occasions in the
study. Moreover, the design should support multiple
activities on the same display for users typing-in
URLs and trying to follow links on Web pages as
they continue to gather information. Finally, the
design of Web information gathering tools should
consider both searching by following the hierarchy
of the Web graph and by submitting search queries
in an efficient manner so that the number of times
users have to switch among applications and tools is
minimized. The significance of the Web information
gathering task necessitates that further work is
needed since current applications including the Web
browser suffer from several pitfalls that degrade the
user’s ability to effectively perform information
gathering tasks on the Web.
5 CONCLUSIONS
The paper discussed the results of a part of a user
study that was intended to reveal the kinds of
behaviour Web users adopt while gathering Web
information. The study results showed that the
search approach for gathering the information
required in the tasks had several characteristics of
both kinds of behaviour. This conclusion reflects
two important points. First, this kind of task is
complicated and requires much effort with several
kinds of activities involved. Second, support is
needed for several activities in the task of Web
information gathering including searching by both
following link hierarchies and frequent query
submission. The support is also required for
comparing information and decision making during
the task.
REFERENCES
Alhenshiri, A., Watters, C., and Shepherd, M. 2012.
Building support for web information gathering tasks,
A paper submitted to the Hawaii International
Conference on System Sciences (HICSS45), (Grand
Wailea, Maui, Hawaii, USA, January 04-07), 2012, to
appear.
Alhenshiri, A., Watters, C., and Shepherd, M. 2010.
Improving web search for information gathering:
visualization in effect. In Proceedings of the 4th
Workshop on Human-Computer Interaction and
Information Retrieval (HCIR2010), New Brunswick,
NJ, USA, 1-6.
Amin, A. 2009. Establishing requirements for information
gathering tasks. TCDL Bulletin of IEEE Technical
Committee on Digital Libraries, Volume 5, Issue 2,
ISSN 1937-7266.
Broder, A. 2009. A Taxonomy of web search. ACM
SIGIR Forum, vol 36, issue 2, 2-10.
Choo, C., Detlor, B., and Turnbull, D. 1998. A behavioral
model of information seeking on the Web--preliminary
results of a study of how managers and IT specialists
use the Web. In Proceedings of the Annual Meeting of
the American Society for Information Science,
Pittsburgh, PA, USA, 25-29.
Ellis, D., Cox, D., and Hall, K. 1993. A Comparison of the
information seeking patterns of researchers in the
physical and social sciences. J. Documentation, vol.
49, issue 4, 356-369.
Kellar, M., Watters, C., and Shepherd, M. 2007. A field
study characterizing web-based information-seeking
tasks. J. the American Society for Information Science
and Technology, vol. 58, issue 7, 999-1018.
Kules, B., Capra, R., and Sierra, T. 2009. What do
exploratory searchers look at in a faceted search
interface? In Proceedings of the 9th ACM/IEEE-CS
Joint Conference on Digital Libraries, Austin, TX,
USA, 313-322.
Marchionini, G. 1997. Information seeking in electronic
environments. Cambridge University Press, New
York.
Rose, D., and Levinson, D. 2004. Understanding user
goals in web search. In Proceedings of the 13th
International Conference on World Wide Web, New
York, NY, USA, 13-19.
Santini, M. 2006. Interpreting genre evolution on the Web.
In Proceedings of the EACL 2006 Workshop, Trento,
32-40.
Sellen, A., Murphy, R., and Shaw, K. 2002. How
knowledge workers use the Web. In Proceedings of
the SIGCHI Conference on Human Factors in
Computing Systems, Minneapolis, Minnesota, USA,
227-234.
Teevan, J., Alvarado, C., Ackerman, M., and Karger, D.
2004. The perfect search engine is not enough: a study
of orienteering behaviour in directed search. In
Proceedings of the 2004Conference on Human Factors
in Computing Systems, Vienna, Austria, 415-422.
Wilson, T., and Walsh, C. 1996. Information behaviour:
an interdisciplinary perspective. British Library
Research and Innovation Report 10, University of
Sheffield, Department of Information Studies,
Sheffield, UK.
WEBINFORMATIONGATHERINGTASKSANDTHEUSERSEARCHBEHAVIOUR
331