LEARNET
A Location-based Social Networking Methodology for Learner Group Forming
Jessica G. Benner, Mohd Anwar and Hassan A. Karimi
Geoinformatics Laboratory, School of Information Sciences,
University of Pittsburgh, 135 N. Bellefield Avenue, Pittsburgh, PA 15260, U.S.A.
Keywords: Mobile Learning, Location-based Learning, Group Formation, Social Navigation Network.
Abstract: The benefits of collaboration in learning have widely been discussed in the literature. Our position is that
location-based social networks can facilitate location-based group formation for learners and support face-
to-face collaboration. In this paper, we present a methodology (LearNet) that is focused on the learner’s past
and current location(s) as a key criterion for recommending peers for collaboration. LearNet is a component
of OnLocEd, a location-based social networking model for online learners that can help learners discover
content, events, and people in proximity to their current location. LearNet utilizes OnLocEd’s location-
based features to recommend learner group forming in the social navigation network system (SoNavNet).
1 INTRODUCTION
The importance of collaboration in learning is
stressed in literature (e.g., Moreno et al., 2007).
Collaboration takes place when learners work with a
group of their cohorts to achieve some learning
outcomes. Thus, group formation is viewed as an
important process of collaboration (Burton et al.,
1997; Barros et al., 2001). In this work, we focus on
location-based group formation to enable learners to
meet each other in-person. This requires
consideration of a learner’s past and current context.
The term context is used in this paper to refer to the
learner’s frequently visited lcoation, the date and
time of the group formation request, and the
learner’s availability for the requested date and time.
Our position is that location-based social networks
(LBSN) can be leveraged to facilitate location-based
group formation.
In this paper, we propose a location-based social
networking methodology for collaboration, called
LearNet, based on the OnLocEd methodology
offered in Anwar et al. (2011) and supported by a
social navigation network platform, SoNavNet
(Karimi et al., 2009). LearNet can facilitate face-to-
face collaboration among learners by recommending
course-centered or interest-centered groups based on
each other’s locations. The contribution of the paper
is LearNet, a methodology for collaboration
designed to leverage OnLocEd’s R3 methodology
and operate within SoNavNet, an LBSN focused on
navigation.
We provide background to our position in Section 2,
introduce the key parameters of the LearNet
methodology in Section 3, and illustrate the process
of group formation using LearNet parameters
through an algorithm and scenarios in Section 4.
Section 5 summarizes the paper.
2 BACKGROUND
As computers miniaturize and the use of mobile
phones increases worldwide, advances in mobile
technology are enabling the application of these
technologies to learning (Gupta and Koo, 2010).
Gentile et al. (2007) observe that mobile devices can
“support students in positioning themselves both in
the physical space and in the community space.” An
emerging learning environment with a focus on
community is the ‘Learning Network’, a specific
kind of online social network to facilitate
communication, participation, and collaboration
among learners (Fetter et al., 2010). This new
environment shows promise for social learning and
collaboration; however, these environments do not
consider any aspect of current or past locations of
the learner as the context for collaboration.
One study of a Mobile Virtual Campus (MVC)
294
G. Benner J., Anwar M. and A. Karimi H..
LEARNET - A Location-based Social Networking Methodology for Learner Group Forming.
DOI: 10.5220/0003959002940298
In Proceedings of the 4th International Conference on Computer Supported Education (CSEDU-2012), pages 294-298
ISBN: 978-989-8565-06-8
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
offers a rule-based clustering approach to location-
based grouping for learning (Tan et al. 2009; Tan et
al. 2010). The work presented on MVC focuses on
the current location of a learner and learning interest
and style. The novelty of our work is the use of the
L-factor paired with knowledge of the learner’s
interests for group formation or clustering.
Clustering is an exploratory process of organizing
objects into groups based on two or more variables
(Finch 2005). In LearNet, we cluster based on a
learner’s location patterns and their learning
interests.
Karimi et al. (2009) developed a special purpose
location-based social network (LBSN) for
navigation experience sharing, called SoNavNet
(i.e., social navigation network). The focus of
SoNavNet is on personalized navigation information
sharing. SoNavNet can support location-based
collaboration among learners. Building on the initial
description of SoNavNet, Karimi et al. (2011)
provide a model for sharing navigation experiences
using a concept called ‘L-factor’. The users of
SoNavNet are assigned an L-factor for each of their
visited locations, and their familiarity with each
location decays as the distance from a visited
location increases. The more a user visits a location,
the larger the L-factor for that location will be and
the strength of their knowledge extends further out
from the location. The L-factor can assist in
location-based grouping and pairing of learners for
collaborative learning activities because it can group
learners based on their location patterns.
Anwar et al. (2011) present a methodology,
designed for SoNavNet, for supporting
collaboration. This methodology, called OnLocEd
(Online Location-based Education), facilitates
recommendations of resources and peers to learners.
The authors emphasize two learning situations,
location-based learning and location-aware learning,
both are supported by the combination of online
social networks, location-based services (LBS), and
mobile technologies, which can result in experiential
and authentic learning activities. The authors
demonstrate that OnLocEd can be used for sharing
learning resources. LearNet serves as a core model
for the OnLocEd methodology utilized in
SoNavNets.
3 LearNet
LearNet is a methodology for location-based
collaboration within a network of learners and
resources. LearNet uses this network to support the
R3 methodology of OnLocEd. This section provides
a description of LearNet parameters.
We describe the LearNet graph G as:
=(,)
(1)
where M and E are two finite sets of nodes and links.
In LearNet, nodes and links represent learners and
resources; a node represents an entity such as a
learner p or a location-dependent resource r (e.g.,
learning artefact or event). Each node is either a
learner or a resource:
={
,
}
(2)
where M
p
is the set of all learners and M
r
, is the set
of all resources. In LearNet, a link can connect a
learner to another learner pp, or a learner to a
resource pr. The links are of two types: learner-to-
learner and learner-to-resource:
={
,
}
(3)
where E
pp
is the set of all learner-to-learner links and
E
pr
is the set of all learner-to-resource links. The
attributes of a learner node are referred henceforth as
a portfolio:
=(Ω,, ,X,C,)
(4)
where U is the learner portfolio containing Ω, the
user profile from SoNavNet, Z, the academic
information shared by the learner, Lf
,
a set of L-
factors for the learner, X, the context information, C,
the completed and active courses, I, and the learner
interests. The profile, Ω, includes a user’s name,
unique ID, home address, contact information, and
account credentials. The academic information
shared by the user includes highest level of
education and degree/program information. The L-
factor in SoNavNet measures the location
knowledge of a user based on their interaction with
the system and is viewed as a learner’s past context.
The set of L-factors in a learner’s profile is:
={
,
,…
}
(5)
where
is the L-factor of a single location and
is:
=(
,
,r,
,n)
(6)
where L
name
is the name of the location; L
xy
is the
coordinate pair for the location; r is a range from
the location; A is the strength of knowledge about
the location; and n is the magnitude of the L-factor
(see Karimi et al. 2011). Current context information
is:
=(
,
,
,
)
(7)
LEARNET-ALocation-basedSocialNetworkingMethodologyforLearnerGroupForming
295
where v
p
is the learner’s availability (available, busy,
etc.), d
p
is the current date, t
p
is the current time, and
L
xy
is the coordinate pair of a learner’s current
location. The set of courses a learner has completed
or is enrolled in can be viewed as C, a set of ċ, or:
ċ = (,
,Þ)
(8)
where ċ is a single course, d is a term number, c
num
is
a unique university class number, Þ is a set of
unique IDs for all learners who have previously
taken the course and Þ
i
is a single peer ID. Learner
interests are topic areas that the user finds
interesting, and the set of learner interests is defined
as:
={
,
,…
}
(9)
where I
i
is a single keyword of interest for a learner.
Finally, attributes for each resource node (these are
location-dependent, not online, resources) include
resource content, location, availability, date and
time.
4 LearNet SCENARIOS AND
GROUPING ALGORITHM
LearNet graph construction is a dynamic process
that occurs as learners interact with SoNavNet.
LearNet starts as a null graph and evolves over time
as learners initiate learner nodes by creating profiles
and resource nodes by searching for resources. Table
1 displays conditions for the construction of nodes
and links in a LearNet.
Table 1: Node and link construction.
Node Type Initialization
Learner Learner submits a profile
Resource Resource is found during a search
Link Type Generation
Learner to
Resource
Learner requests navigation/routing to a
resource, recommends a resource, or sets
a reminder for a resource.
Learner to
Learner
A Learner connects to, communicates
with, or recommends another learner
(‘friends’), or requests
navigation/routing information from
another learner.
Relationships (learner-to-learner and learner-to-
resource links) are created through the actions of the
users. For example, if a learner befriends another
learner a link between their respective learner nodes
is added to the database. Now we will provide an
example scenario to illustrate how nodes and links
are created. A learner submits a profile to
SoNavNet, sharing their academic information,
course list, and interests, and is currently using
SoNavNet to request a reference book for their
upcoming class. Given this scenario, we know that
their portfolio, U, is initialized and a learner node
has been constructed. A book request populates the
resource containers for the resulting resources. Let’s
say the result list includes 10 books. The learner
selects the book they want from the list and requests
a route to its location. At this point, the LearNet
instance includes one learner node and one learner-
to-resource link between the learner and the desired
resource. As the learner continues to interact with
the system making friends and finding resources the
LearNet instance will evolve and grow. Now that we
have briefly shown how a LearNet instance is
created and grows, we will explore LearNet’s
method for collaboration support.
LearNet supports collaboration by helping
learners create location-based groups. The grouping
process is shown in Figure 1.
Figure 1: Dynamic Grouping Process.
A search request, Q, includes a search object and
a match constraint for either a learner or resource,
that is:
=(
,
)
one set must be non-empty
(10)
CSEDU2012-4thInternationalConferenceonComputerSupportedEducation
296
=(,)
both sets must be non-empty
(11)
=(,)
both sets must be non-empty
(12)
= {01, 02, … }
(13)
=(
,
)
one set must be non-empty
(14)
where Q
p
is the learner request, Q
r
is the resource
request, o is an objective for the search, m is a match
constraint of either c
num
, a unique course number, or
I
i
, a learner interest for a learner request, and k
i
∈
is a keyword for a resource request. Example
objective (o) codes are as follows: form a group
(01), find an event (02), find a book (03), and so on.
One example of the recommendation feature in
LearNet is a recommendation requested by a learner.
Recommendations requested by a learner are
requests using the attributes assigned to known items
and retrieving items that match the search criteria.
To illustrate this kind of recommendation, we will
show a technique to create a dynamic study group.
The process for a dynamic grouping request is
shown in Figure 1. The following assumptions are
made: all students attend the same university online
and all students have a SoNavNet account.
The grouping recommendation process is
initiated by a learner, p
i
, who submits a learner
request, Q
p
, with the objective of forming a group
(o=01) and either a course number or an interest
topic (see Equation 11). Once the algorithm has
verified that the objective is group formation, the
process of grouping begins. The algorithm verifies
the provided course number or interest topic and
retrieves the set of L-factors from the requestor’s
portfolio,
(see Equation 5). The L-factor is
critical for grouping due to the knowledge of a
user’s location patterns. This allows for a more
thorough match than using only a learner’s current
context.
Next, the algorithm uses the course number to
retrieve peers Þ (list of students with same course)
or the interest topic to retrieve, V (a set of learners
who have the same interest), such that:
={pI
,pI
,…pI
}
(15)
where pI
is a learner with the same interest. Then, a
set of L-factors are retrieved for all learners in Þ or
V, such that:
Þ
={
,
,
…
}
(16)
where
is the set of L-factors for one matched
learner. Finally, these L-factors are merged into one
set, such that:
=
∪
…∪
(17)
Where
={
,
,
,…
}
(18)
where
is a single L-factor for a single
location-learner pair in the set. The result set of the
intersection of requestor, H, and candidate, V’s L-
factors:
=
∩
(19)
where
is the requestor’s L-factors and
is
the merged set of L-factors for all candidate group
mates in the result set. If g is not empty, the system
will initiate group formation and push the candidate
list to OnLocEd to suggest a physical meeting using
the recommendation model described in Anwar et al.
(2011). Let’s look at a scenario.
Betty is an online learner who needs to form a
group for a course. She could use email but Betty
uses SoNavNet to initiate a grouping request
because she wants to be able to meet her group in-
person. The algorithm gets Betty’s L-factors and
compares them to L-factors for all students in the
course. Once the candidate set is retrieved, it is
passed to OnLocEd which can help the students
coordinate the meeting and navigate to the location.
Betty is very happy to be able to meet in person a
few times during the semester and feels that this will
make it easier to work on a large project for an
online class.
5 SUMMARY
Our position in this paper is that, location-based
social networks (LBSN) can be leveraged to
facilitate location-based group formation for
collaboration in the learning context. We introduce a
location-based social networking methodology for
collaboration, called LearNet, and present a
location-based grouping algorithm.
ACKNOWLEDGEMENTS
We would like to thank Renee Ralke for her
exploration of the topic with the authors.
LEARNET-ALocation-basedSocialNetworkingMethodologyforLearnerGroupForming
297
REFERENCES
Anwar, M., H. A. Karimi, et al. (2011). OnLocEd: Finding
learning resources and communities. CSEDU 2011,
3rd International Conference on Computer Supported
Education. Noordwijerhout, The Netherlands: 204-
210.
Barros, B., Mizoguchi, R. & Verdejo M. F (2001). A
Platform for Collaboration Analysis in CSCL: An
ontological approach. International Conference on
Artificial Intelligence in Education (AIED 2001), May
19-23, 2001.
Burton, M., Brna, P., & Treasure-Jones, T. (1997).
Splitting the collaborative atom: How to support
learning about collaboration. International Conference
on Artificial Intelligence in Education (AIED 1997).
IOS Press: 135-140.
Fetter, S., Berlanga, A. J., and Sloep, P. (2010). Fostering
Social Capital in a Leaning Network: Laying the
Groundwork for a Peer-Support Service. Retrieved
from: http://dspace.learningnetworks.org/handle/1820/
2024
Finch, H. (2005) Comparison of Distance Measures in
Cluster Analysis with Dichotomous Data Journal of
Data Science Volume 3, pp. 85-100
Gentile, M., D. Taibi, et al. (2007). Social knowledge
building in a mobile learning environment. OTM 2007
Ws, Part I, LNCS 4805. Z. T. R. Meersman, P.
Herrero et al. Berlin Heidelberg, Springer-Verlag:
337-346.
Gupta, B. and Y. Koo (2010). "Applications of mobile
learning in higher education: an empirical study."
International Journal of Information and
Communication Technology Education 6(3): 75-87.
Karimi, H. A., J. G. Benner, et al. (2011). A Model for
Navigation Experience Sharing through Social
Navigation Networks (SoNavNets). In Proceedings of
IEEE IRI 2011 (International Workshop on Issues and
Challenges in Social Computing -- WICSOC '11). Las
Vegas, USA: 557-560.
Karimi, H. A., B. Zimmerman, et al. (2009). SoNavNet: A
Framework for Social Navigation Networks.
International Workshop on Location Based Social
Networks (LBSN’09). Seattle, WA.
Moreno L., Gonzalez C., Castilla, I., Gonzalez E., Sigut, J.
(2007). Applying a constructivist and collaborative
methodological approach in engineering education.
Computers & Education. Elsevier: 891-915.
Tan, Q., Kinshuk, Jeng, Y.L., and Huang, Y.M. (2010) “A
Collaborative Mobile Virtual Campus System Based
on Location-Based Dynamic Grouping” in 10th IEEE
International Conference on Advanced Learning
Technologies, pp. 16-18
Tan, Q., Kinshuk, Kuo, Y.H., Jeng, Y.L., Wu, P.H.,
Huang, Y.M., Liu, T.C. and Chang, M. (2009)
“Location-based Adaptive Mobile learning Research
Framework and Topics” in IEEE International
Conference on Computational Science and
Engineering pp. 140-147
CSEDU2012-4thInternationalConferenceonComputerSupportedEducation
298
