Proposal Model to Construct Adaptive Assessment Items
Héctor Barbosa, Francisco García
Departamento de Informática y Automática, Universidad de Salamanca, Plaza de la Merced s/n, Salamanca, España
Keywords: Software tools for E-learning, E-learning standards, learning materials development.
Abstract: This paper presents the proposal for the development of an Authoring Tool for the definition of Assessment
items for E-learning platform that follows accepted and open standards like IMS an XML. The main goal of
this model is the definition and packaging of semantic learning objects that could be integrated and
deliberated to open e-learning platforms.
In recent years, instructional and educational
institutions have been incorporating information and
communication technologies in learning and
teaching processes in order to increase the quality,
efficiency, and dissemination of education. As long
as those projects cover the needs of individuals in a
particular way, the success and transcendence of
such developments could be incremented by
performing adaptability to each user so the learning
experience can be enhanced.
However, we must ensure that those efforts do
not become groups of isolated isles, so we may
construct standardized tools so they can be
applicable, compatible, and interchangeable between
them. Also, the emerging of the Semantic Web has
allowed the development of systems that could
satisfy the requirements of open systems. One
accepted standard for the development of educative
objects is the IMS (IMS organization, 2004), a
global learning consortium that develops and
promotes the adoption of open technical
specifications for interoperable learning
technologies that become the standards for
delivering learning products worldwide. Among the
inherent importance of the developing of e-learning
platforms, we want to emphasise in the role of the
assessment activity inside the e-learning process. We
want to concentrate in this task, and see how it can
help to improve the e-learning process to all the
participants: students, teachers, and content
Traditionally, assessment activity has been seen like
a task aside of the e-learning process and there is a
danger in focusing research on assessment
specifically (Booth, 2003), as this tends to isolate the
assessment process from teaching and learning in
The results of the test made by the students could
allow an adecuation of the web site that reflects the
new knowledge topics or the new syllabus that will
be taken. According to the Australian Flexible
Learning Framework (Backroad Connections, 2003),
assessment, especially when is included within a real
learning task or exercises, could be an essential part
of the learning experience, giving to the entire Web
site the characteristic to adapt itself to the needs of
the users. This could be an interesting feature of an
educational Web site because the improvement of
the online teaching experience by giving to the
student a convenient feedback, re-adaptation of the
educative content to the new knowledge level,
setting user-tailored content information.
For the student, the assessment activity informs
progress and guide learning; also, it is essential for
the accreditation process and measures the success
of the student. Assessment tasks can be seen as the
active components of study, also assignments
provide learners with opportunities to discover
whether they understand and not, if they are able to
Barbosa H. and García F. (2006).
AN AUTHORING TOOL TO DEVELOP ADAPTIVE ASSESSMENTS - Proposal Model to Construct Adaptive Assessment Items.
In Proceedings of WEBIST 2006 - Second International Conference on Web Information Systems and Technologies - Society, e-Business and
e-Government / e-Learning, pages 379-382
DOI: 10.5220/0001242803790382
perform competently and demonstrate what they
have learnt in their studies. Furthermore, the
feedback and grades that assessors communicate to
students serve to both teach and motivate (Thorpe,
From our point of view, the assessment activity
could be considered as an integrator step that help
the entire process to get self adecuation to the user
needs, giving feedback to both the student and the
Nowadays, it is necessary to produce educative
Internet-based systems that permit the dissemination
of the education, covering the needs of diverse
learning group profiles. To obtain this, it is desirable
that such systems perform automatic task to adapt
itself to each user, disconnecting the content from its
presentation by using a semantic approach rather
than a syntactical one, defining a meaningful web.
In consequence, learning systems must be
flexible and efficient, and one way to accomplish
that is to be an open and standardized system. We
want to focus on the following standards by giving
their general characteristics and their support
features for a learning system:
One aim is to make those systems to work in
adaptive learning systems and, given the fact that the
assessment activity is an important and integral part
of the e-learning process; it is desirable that this
evaluation activity could be adaptable as well. If we
want the assessment to be interoperable, compatible,
and shareable, we must have to develop a
standardized tool (Barbosa, 2005).
For the development of our Adaptive Assessment
Tool, we outlined a model that will help us in the
process. This model (figure 5), have 3 levels of
definition, starting with the first level which shows
the model in its most abstract definition, ending with
the third level which is the most concrete definition
of the same model.
4.1 First Level of Definition
We will explain the first and the inner level of
definition (fig. 2), according to the complete figure
showed in the third level of definition (fig. 4):
Figure 1: First level of definition.
a) Learning Object Definition: the process starts
with the definition of the main component of a
test: the learning object that will contain a single
assessment item or question. It is convenient to
define single questions to ensure a convenient
granularity of the elements for the next step of
the process: the construction of the test itself. To
ensure the reusability, interoperability,
durability, and accessibility of these objects, they
will be defined using open standards, following
the definition of semantic objects using XML
and the IMS, IMS QTI specifications.
b) Test Construction: Here the assessment items are
selected from the repository to construct the test
that will be delivered to the students, selecting
the questions based on the subject and unit that
will be evaluated. At this point of the process,
the question items are packaged following the
SCORM specification.
c) Delivery and score of the test: Once we have a
group of assessment items to conform a
deliverable test, is time to present it to the
student through an LMS. In this process is where
the adaptation is made in an interaction process.
The responses to each question are saved to
obtain the final score for each student.
d) Student Model Update: In the LMS, the final
score is taken into account to update the student
model so the learning environment could adapt
to the student in the next interaction process.
4.2 Second Level of Definition
Following the four quadrants of the first level of
definition, we will explain them in the second level
(fig.3), in a more concrete definition:
Figure 2: Second level of definition.
a) Learning Object Definition: In this level we are
defining the first interface of the system, that
allow to the educational instructor or teacher to
capture each assessment item directly or by
make a relation to external sources, assigning a
categorization of difficulty that will help in the
adaptation process. After that, the assessment
item will be packaged with a manifest (using
IMS CP/LD specification), links to external
sources, the assessment item itself (using IMS
QTI) and the definition of difficulty for the
assessment item (using XML), and stored in a
semantic object repository.
b) Test Construction: We use the second interface of
the system to allow the instructor or teacher to
generate the test by capturing the subject and unit
of that subject that will be evaluated. This process
constructs the deliverable test by selecting all the
assessment assets from the semantic objects
repository and transforming all the assessment
items to a XML format. By doing this, we have a
single file in XML format containing all the
assessment items to construct single tests,
allowing an adaptation process.
c) Delivery and score of the test: In the ALE, the
final XML file containing all the assessment
items, the test is presented to the student. A third
interface is used that allows to the student to
identify her/himself. In this phase, the students
answer the questions as they are presented by the
system. At the end of the exam, the final score
could be showed to the student.
d) Student Model Update: The ALE, according to
its own process could update the student model
and the curriculum to save the final score of the
test already performed by the student.
4.3 Third Level of Definition
In this level (fig. 4), we outline the interaction and
adaptation process. In the event that the ALE do not
Figure 3: Third level of definition.
Assessment Items
conform an adaptation process for external assets,
we could suggest one, following the logic (Eri-CAE
Network, 2003) given below:
a) All items that have not yet been administered are
evaluated to determine which will be the best
one to administer next, given the currently
estimated ability level.
b) The “best” next item is administered and the
student responds.
c) A new ability level is computed based on the
responses of the student.
d) Steps 1 to 3 are repeated until a stopping
criterion is met.
Questions would be given randomly to eliminate
cheating. The new ability level is computed based on
the equations (Mia, 1998), used in the MANIC
system (op. cit.) during the quiz sessions to compute
the new estimated ability level (see table 1).
To determine the first OldValue at the beginning
of the test, the student is given with a pre-test
question, after that, for every question, this value
will be updated, based on student’s response.
This could be an automatic and independent
process from the ALE, which could be performed by
a software agent. If the test will be able to use
external multimedia assets, it is also possible to use
an external agent stored in the same server, which
starts the adaptation process in the client’s machine.
Online assessment is an important step inside the e-
learning process because gives convenient feedback
to all participants in the process, helping to improve
the learning and teaching experience.
In this paper, we wanted to emphasize the role of
the assessment inside the e-learning process and
defining the factors of importance to the main
elements that participate in this process: the
educative content and adaptation process, the users
or students and the teachers and assessors. We think
that the assessment activity takes place in a specific
point of the process as we show it in the figure I, and
we conceptualized the activity as the link that closes
the chain of the e-learning process.
Adaptability is another key factor in assessment.
Given the fact that assessment is an important
element of the e-learning process and that this
process looks to be interoperable, then we can think
that the assessment tool could be used with different
educative content administrators with different
conceptualizations and ways to design and apply a
test for their students. To face this situation it is
necessary to develop an assessment tool that give
several ways to design an test with different types of
resources, different kind of assessments, group of
students, kind of questions, managing schedules, etc.
Under this conceptualization, we want to create
an Adaptive Assessment Tool (AAT) that could take
into account the specific characteristic of the HyCo
system and be intrinsically part of it.
We want to thank to the group GRIAL (Research
Group in Interaction and eLearning) of the
University of Salamanca for their contributions and
ideas for the development of this work.
In addition, we want to thank to the Education
and Science Ministry of Spain, National Program in
Technologies and Services for the Information
Society, Ref.: TSI2005-00960.
Héctor Barbosa thanks the National System of
Technological Institutes (SNIT – Mexico) for its
financial support.
Backroad Connections Pty Ltd. (2003). Assessment and
Online Teaching. Australian Flexible Learning
Framework, Quick Guides Series. 3 (1), 20.
Barbosa, H., Garcia F.. (2005). Importance of the online
assessment in the e-learning process. 6th International
Conference on Information Technology-based Higher
Education and Training ITHET 2005. CD version.
Booth, R., Berwyn, C.. (2003). The development of
quality online assessment in vocational education and
training. Australian Flexible Learning Framework. 1
(1), 17.
Eri-CAE. (2003). Computer adaptive testing tutorial.
Available: Last accessed 17 Sep
IMS Global Learning Consortium. (2002). IMS Question
& Test Interopeability: ASI Information Model
Specification. Available:
question/qtiv1p2/imsqti_asi_infov1p2.html. Last
accessed 20 Nov 2005.
Mia, K., Stern & B.P. Wolf. (1998). Curriculum
sequencing in a Web-based tutor. Lecture notes in
Computer Science. 1452 (1), 574-583.
Thorpe, M.. (2004). Assessment and ‘thirth generation’
distance education. Exploring assessment in flexible
delivery of vocational education and training program,
Australian National Training Authority. 2 (1) 20.