A Domain Engineering For Interactive Learning Modules

A Domain Engineering for Interactive Learning
Modules
Danilo L. Dalmon and Leônidas O. Brandão
Instituto de Matemática e Estatística, Universidade de São Paulo
Department of Computer Science
Rua do Matão 1010, Cidade Universitária, São Paulo, SP - Brazil
Email: ddalmon@ime.usp.br, leo@ime.usp.br
Anarosa A. F. Brandão
Escola Politécnica, Universidade de São Paulo
Department of Computer and Digital Systems Engineering
Email: anarosa.brandao@poli.usp.br
Seiji Isotani
Instituto de Ciências Matemáticas e de Computação, Universidade de São Paulo
Laboratory of Software Engineering
Email: sisotani@icmc.usp.br
For creating educational systems, each developer usually applies different approaches to specific
situations, including advanced software engineering techniques. In the case of systems that provide
interactivity-intense assignments, problems during their development include difficulties to manage
component repositories and the absence of a systematic process to support code reuse. Interactive Learning
Modules (iLM) are systems designed in this context, which also provide key functionalities to facilitate the
teacher’s work. To address the problems faced during theirs development, a Domain Engineering is
proposed to build iLM. In order to define the iLM software family, this paper presents the core features of
existing systems, describes the method used to produce an application framework and how to instantiate it.
Restructuration of existing iLM using the proposed Domain Engineering is reported with initial high gains
in productivity and system quality.
Classification: D.2.2 Design Tools and Techniques; D.2.11 Software Architecture; D.2.13 Reusable
Software; K.3.1 Computer Uses in Education.
Keywords: Educational Software; Domain Engineering; Application Framework; Interactive Learning
Modules.
1. INTRODUCTION
The problem of developing educational software is not new. Nicolson and Scott (1986) analyzed educational
software development cycles and reported that the technology developed is often inadequate for both teachers
and learners. Educational Software are tools created to be used by students and teachers, and they are
expected to improve students’ knowledge about some subject and/or help teachers to produce better
educational content (Tchounikine, 2011).
In order to establish a clear goal for any educational software, the literature points out the importance of
interactivity to increase students’ motivation and learning outcomes. For instance, many authors address the
improvement in students’ understanding through their interactivity with the learning content (Isotani and
Brandão, 2008, Kortenkamp, 2004, Sims, 1997, and Ruthven, Hennessy and Deaney, 2007). Also,
interactivity through quick feedback for each student’s response is important to keep them motivated
(Hentea, Shea and Pennington, 2003, Hara and Kling, 1999).
Interactive Learning Modules (iLM), formerly known as e-LM (Rodrigues, Brandão and Brandão, 2010),
are educational software that focus on these interactive features and the main characteristics of which are:
running in web-browsers, presenting authoring tools, promoting interactivity-intense assignments and

providing communication with Learning Management Systems (LMS) under a simple protocol (Dalmon,
Isotani, Brandão and Brandão, 2011a).
Developing educational software is expensive and time-consuming. Moreover, to produce new systems
that provide interactivity demanded by teachers and educators may imply higher costs, related to specifying,
developing and testing the system with the final users, teachers and students. An approach to reduce costs
and improve maintainability during software development is code reuse. Nevertheless, without a systematic
process to support code reuse, this approach could be ineffective, even considering a small group of people
that share almost the same knowledge. This was occurring in our research group, in which five iLM were
developed: iGeom (Isotani and Brandão, 2008), iGraf (do Prado, 2006), iCG, iComb (Eisenmann and
Brandão, 2009) and iVProg (Kamiya and Brandão, 2009) without the adoption of a systematic software
process and people organization.
In addition, the development of educational systems has some differences from other specific-area
software in some aspects (Tchounikine, 2011), which makes code reuse difficult (Spalter and van Dam,
2003). These differences include components with numerous granularities, self managed repositories, ill-
defined quality criteria and social issues, such as team organization and project management. Many attempts,
including applying software engineering techniques, were made to solve these problems (Roschelle, Kaput,
Stroup and Kahn, 1998, Spalter and van Dam, 2003, Ateyeh and Lockermann, 2006), but some issues still
remain, such as systematic use and reuse of components available in repositories, limited interactivity
provided, and underlying pedagogical approaches that prevent teachers from using systems in broader
educational contexts.
In order to overcome many of these problems, a Software Product Line (SPL) approach was chosen for
developing the iLM software family. The intended outcome of this research is to provide a Domain
Engineering for improving the development of these systems. This technique defines team organization and a
systematic development process to improve reuse and quality of a group of similar software (Clements and
Northrop, 2001). Domain Engineering can be used to: (i) design the refactoring of a set of existent iLM; to
(ii) develop new iLM; and to (iii) add new domain-independent features to them. An application framework
was used to manage variability in Domain Engineering design and implementation (Dalmon et al., 2011a).
Our main goal is to report the development of a Domain Engineering for a specific family of educational
software, the iLM, in order to highlight lessons learned and development results. It is not our goal to
contribute to the SPL field of research. The work presented is inserted in a research project the long term goal
of which is to develop and to evaluate a SPL for iLM. This evaluation will be possible after several years of
SPL use so there will be data about software quality, development, maintenance and evolution under the
influence of this technique. Currently, the first version of Domain Engineering is completed and Application
Engineering is starting, which provides data about the initial influences of the SPL technique on the
maintenance of existing iLM.
The text is organized as follows. Section 2 presents some approaches from the literature to the
development of educational software, highlighting the main challenges focusing on code reuse. Section 3
describes existing iLM as a product family. The method and the development of the Domain Engineering are
detailed respectively in sections 4 and 5. Section 6 reports current Application Engineering results and a first
evaluation. The paper ends with some conclusions and future directions.
2. EDUCATIONAL SOFTWARE DEVELOPMENT AND CODE REUSE
Developing high quality educational software is expensive, time-consuming and is usually done in
exceptional contexts, such as research projects at universities, by students without professional experience
(Spalter and van Dam, 2003). Studies indicate that, in some cases of highly computer-managed learning, it
can take more than 100 hours of programming for each hour of instruction (Aleven, Mclaren, Sewall and
Koerdinger, 2008). For being influenced by different factors in design and development in comparison with
regular enterprise systems, educational software developers elaborated specific techniques for code reuse. In
this section, we present some related work that report several of these factors and techniques.

Software used for educational purposes can be divided into three groups (Mor and Winters, 2007): (a)
those intended to organize and to present content, such as LMS; (b) those intended to provide assignments,
such as tutors and simulators; and (c) those not intended to be educational despite being used as such, such as
text editors or Internet search engines. This categorization serves mainly to restrict the context of this
analysis. This paper relies on the group of educational software which provides interactive assignments.
In fact, we are interested in the development of systems that provide domain-specific features to support
teachers’ everyday work, as well as interactivity-intense assignments to students. By domain-specific features
we mean functionalities that use domain specificities in system design in order to improve learning. By
interactivity-intense assignments we mean the possibility of interacting with the assignment through specific
types of interactivity, e.g. manipulate, construct and support (Sims, 1997, Tang, 2004). Manipulate
interactivity allows the modification of domain-specific objects to simulate behavior and then make
conjectures and tests. Construct interactivity allows the creation of objects before manipulation or simulation.
These types of interactivity are essential to constructivist problem-based learning approaches (Albanese and
Mitchell, 1993, Savery and Duffy 1995), which we value the most. Additionally, the support type of
interactivity, which feedbacks students on their performance, provides many benefits to teachers and students
(Tang, 2004). The feedback could be “right or wrong” information (Isotani and Brandão, 2008) or could be a
more sophisticated dialog, such as those provided by Intelligent Tutoring Systems (Aleven et al., 2008).
A literature review shows several approaches to facilitate code reuse during the development of
educational software. In the context of the ESCOT Project (Roschelle et al., 1998, Roschelle, Digiano,
Chung, Repenning, Tager and Treinen, 2000), the task of authoring educational components was divided
among teams of developers and teachers. Therefore, code reuse was based on the search for environments
and authoring tools for assembling software components, which resulted in using a JavaBeans (Oracle, 2012)
component repository to create small applications to a broad extent in middle and high-school curricula.
Problems related to the development of educational software and their usage were reported by Spalter and
van Dam (2003). Among others, they have identified problems related to team organization, the lack of
programming skills and interoperability among components. Moreover they have pointed out issues related
to the growth of intellectual property and problems of software usage for pedagogical purposes in schools. To
overcome such problems, Spalter and van Dam (2003) have proposed a family of educational software called
Exploratories. Each Exploratories application was built to support a specific knowledge domain, such as the
circulatory system in Biology, and was (recursively) composed of code components in order to address
interoperability and flexibility issues. However, the goal of independent sub-applications available on the
web, where a regular teacher would assemble them for classroom use, was still far away.
In the same direction of using components and repositories to foster reuse, Bote, Hernandez, Dimitriadis,
Asensio, Gomez and Vega (2004) worked with IMS Learning Design – IMS-LD (IMS, 2003) specification to
make components interoperable and grid-based techniques to build a family of collaborative educational
systems to provide domain-specific assignments to a computer architecture course. They adopted an
approach in which the teacher could enter his or her students’ data and tailor the tool by using grid scripting
to assemble IMS-LD components. On the contrary, our approach tries to reduce the amount of teacher work
by removing the need to learn complex technologies, such as grid scripting.
An alternative method to foster reuse is to design a development process to create flexible components
that can be reused in different educational contexts (Douglas, 2001, Muzio, Heins and Mundell, 2002, Boyle,
2003, Polsani, 2003). Thus, the whole system is reusable instead of just parts of the code. Generally they are
supported by repositories of Learning Objects (LO) and other educational systems (Richards, Mcgreal,
Hatala and Friesen, 2002, Nash, 2005). Pankratius, Stucky and Vossen (2005) proposed the reengineering of
existing disconnected LO using Aspect Oriented Programming – AOP to reuse code during LO development.
AOP was used to describe original LO and how to use them, or parts of them, to create new LO. These
approaches solve many reuse problems, provide benefits to programmers and designers, but they are far from
the teacher’s everyday work.
The example of LO used by Pankratius et al. (2005) is related to digital information products. This kind of
LO fits in the group of educational systems that intend to organize and present content. A significant effort
has been spent on developing techniques to increase the quality of digital information products and to reduce

the effort to develop them by reuse. These products do not have the same educational goals or computational
structure as assignment-based educational software such as iLM. Nevertheless, some contributions from
these works can be extended to the context of developing assignment-based systems, such as methods to
foster code reuse and lessons learned from facing difficulties with managing people and setting requirements.
IMS-LD specification was also used to create a family of similar tools using a model-driven generative
approach (Dodero and Díez, 2006). A feature model was used to list the available features that were mapped
into models through IMS-LD files, which could be shared as LO. Additionally, the same specification was
used during the instantiation of a framework for reengineering presentation of educational software (Choquet
and Corbière, 2006). This framework models learning scenarios that can get existing systems to create new
and more organized ones.
AOP was also adopted to provide reuse of content-intense systems (Ateyeh and Lockermann, 2006). In
this work, AOP was applied to woven aspects similarly to those of a SPL, with domain engineering and
course engineering (which has the role of application engineering in a SPL) detached. Domain engineering
has been modeled using an ontology, and in each step there were content and didactics components. A
courseware authoring tool was developed to produce LO, such as LMS modules and digital information
products. Oberweis, Pankratius and Stucky (2007) proposed the full use of SPL to the creation and
maintenance of digital information products. In the case of e-Learning Systems, Ahmed and Zualkernan
(2011) reported a SPL that is structured to create systems that model entire courses on a given subject, using
components repository and a process based on different views of assets.
In addition to code reuse, there are different educational software design proposals that consider
contributions from other aspects, for instance, from content experts, teachers, instructional designers,
institution practices, curriculum and usability (Mayes and Fowler, 1999, Douglas, 2001, Muzio et al., 2002,
Sampson and Karampiperis, 2006). Furthermore, the related literature also highlights other aspects. An
educational system should not be user-centered, in which the “user” has a domain knowledge and uses it with
the system, but learner-centered, in which the “user” aims to develop a domain knowledge by using it
(Quintana, Krajcik, Soloway and Norris, 2002). This implies deep graphical user interface and content
structural changes. Interactivity is very important to educational software, and different types of interactivity
must be considered to foster different types of learning (Tang, 2004), on which the whole system is based.
Also, the tool must be in line with teacher’s instructional practices, and not establish new ones (Hinostroza,
Rehbein, Mellar and Preston, 2000), since this could alter the system most basic requirements.
Finally, the literature presents many uses of software engineering techniques to enhance educational
software development. These approaches contributed to our research with lessons learned and challenges to
be faced. They propose the development of educational software only intended to present content while they
do not support interactive assignments which is a key feature for such systems. Nevertheless, it is possible to
learn from these approaches how to support code, architecture and process reuse. Unfortunately, few of them
consistently report pros and cons, evaluation methods, or differences in development contexts (systems size,
team size, team experience), limiting knowledge capitalization (Tchounikine, 2011).
In order to contribute to the field, our proposal adopts the SPL approach to deal with code, architecture
and process reuse during the development of a family of educational software that is intended to provide
domain-specific assignments with different types of interactivity, e.g. manipulate and construct interactivity
(Sims, 1997). Such a family is described in the next section.
3. THE SOFTWARE FAMILY OF INTERACTIVE LEARNING MODULES
Our research group develops educational software intended to improve the learning and teaching processes
mainly by facilitating teachers’ work and providing interactive assignments to students. These systems,
called iLM, were developed following some guidelines to be compatible with a communication protocol of
LMS (Rodrigues et al., 2010), but they were also conceived as single applications.
Prior to this work, five systems were built within the iLM family: iGeom, an Interactive Geometry System
(Isotani and Brandão, 2008); iGraf, a system to study mathematical functions and graphics (do Prado, 2006);
iCG is a system that emulates a computer with an embedded compiler; iComb, an educational counting

system (Eisenmann and Brandão, 2009); and iVProg, a visual programming system (Kamiya and Brandão,
2009). Fig. 1 shows some of their screens.
Although their development had followed some guidelines, they did not follow a systematic development
process. For instance, iGeom has been developed since 2000. Currently, with five developers working on
iGeom, it is at version 5 and it has 46,000 lines of code. The first deployment of iComb occured in 2007 and
it had 7500 lines of code written by two programmers. At present, the second version is being developed by a
third developer. Therefore, their maintainability has become a hard task and the adoption of software
engineering techniques is imperative for the group research and development success.
Fig. 1: Screenshots from iLM: (a) iGeom, (b) iGraf, (c) iComb and (d) iVProg.
In order to decide what software engineering technique to adopt, a systematic analysis of existing iLM
was conducted and some commonalities among them were observed so that we could characterize them as a
family of similar software products. Thus, we decided to apply a SPL approach for the development of new
iLM and the refactoring of existing ones.
During the systematic analysis of iLM, their core features were identified (Dalmon et al., 2011a):
 Run in web-browsers – by running in web-browsers, iLM are platform independent and can be used in
classrooms as well as in distance education.
 Communication with LMS – under a simple protocol, iLM communicate with LMS in order to (a) open
assignments files and (b) send the student’s assignment solution back to the LMS.
 Authoring tools – to allow the creation of assignments and the choice of a pedagogical approach.
Usually, educational software adopts an underlying pedagogical approach that may limit the systems
flexibility and content reuse for teachers.
 Interactivity-intense assignments – systems of the iLM family provide interactivity-intense assignments
since users can create or modify domain-specific objects within assignments to simulate a behavior and
then make conjectures and tests.
Besides the core features, we had identified some optional features exploited by the learning and teaching
processes. Although these features were included in the SPL core, we kept them optional.

 Automatic checking of assignments – as an additional interactivity feature, feedback is one of the most
important features in educational software (Hentea et al., 2003). In assignment-based systems, such as
iLM, an automatic assessment feature can help both teachers and students. By using an iLM with such a
feature embedded in an LMS, it is possible to store all the students’ results, and students can profit by
checking whether their solution is correct. This raises students’ motivation and release teachers from the
task of checking their assignments (Isotani and Brandão, 2008).
 Intelligent tutoring – Intelligent Tutoring Systems – ITS bring many benefits to educational practices
(Koerdinger and Corbett, 2006). By using ITS features during the authoring of iLM assignments,
teachers can increase the amount and quality of feedback interactivity provided by the system. In spite of
being suggested, this feature is under development for an existing iLM (Dalmon, Brandão and Brandão,
2011b).
After specifying and defining the iLM software family, we detail the scope of the SPL approach in the
next section.
4. DOMAIN ENGINEERING METHOD FOR ILM
This section presents the method used to develop the Domain Engineering of a Software Product Line (SPL)
for iLM in order to allow a discussion of its influence on the development process in next sections. A SPL
consists of methods and tools to define and to guide the creation of iLM, fostering code, architecture and
process reuse (Clements and Northrop, 2001). It is composed of two parts, Domain Engineering in which
code, architecture and process are developed for reuse, and Application Engineering, which makes use of
code, architecture and process from Domain Engineering to develop iLM. The main goals of applying this
technique for developing iLM are to reduce the effort during systems creation and maintenance, to improve
systems quality and to provide systematic processes for design, implementing, testing and documenting.
The SPL approach affects four aspects of development: business, architecture, process and organization
(Linden, Schmidt and Rommes, 2007). Business sets the main goals of the products that will be created by
the SPL. Architecture defines the systems code structure, and where code reuse is more evident. Process
establishes the methods, activities and steps along the development; and organization sets the tasks
attribution to developers. Our focus lies on the architecture and on process aspects. The business aspect is
defined as the approach taken by the research group, which is mainly to contribute to the community by
providing open source tools, while the main goals of the iLM family is helping teachers and students. The
organization aspect is defined by delegating tasks to graduate students and programmers with scholarships.
Currently, the group consists of seven people, one for each iLM, one for the SPL, and two newcomers as
additional programmers for iGeom and iVProg.
The method applied to the Domain Engineering for iLM is a slightly modified PLUS method (Gomaa,
2004). It was chosen mainly considering the team’s experience in UML-based software design and the
availability of documentation, as there is no SPL expert in the team. PLUS states that the features of the
aimed systems must be listed and classified as core (compulsory), variant (that varies from a range of
possibilities) and optional (that can be chosen or not). Therefore, during application engineering, developers
must choose which possibility of each variant feature and whether the system will have the optional features.
Later, during the analysis step, classes must also be classified as core, variant and optional, which must
reflect the previous features classification.
The change in PLUS was made in order to adapt the method to the iLM domain, because of the variability
encountered among existing systems. Features are categorized as core, alternative or optional, but classes are
only categorized as core or variant. This is due to the fact that in our systems each class only implements one
feature, thus they can only be of types core or variant (mainly user interface and domain model classes), no
matter whether the feature is core or optional. This simplifies implementation of individual components.
Additionally, when core and optional features have only alternative behaviors, such as user interfaces, they
are not considered alternatively in order to simplify the design and to separate user interface implementation
from the other classes.

Moreover, the Spiral model of software development process (Bohem, 1986) was integrated to the
Domain Engineering process, to define how the PLUS method would be applied. Fig. 2 depicts a schematic
view of the used process. First, we defined the Domain Engineering scope, by describing the system family,
the core requirements, core and optional features that will be provided as the preliminary design. Then, an
iterative development step, including software analysis, design, implementation and testing, is performed to
build the first Domain Engineering prototype. With this prototype, it is possible to start the development of
iLM in Application Engineering. Whenever application engineering sets new requirements and features for
Domain Engineering, it restarts an iterative development step, in order to provide new prototypes, or releases.
Fig. 2: Schematic view of the process used: SPL with PLUS method and Spiral models.
In this context, the PLUS method utilization started by listing the non-functional requirements of existing
systems, then a feature analysis was made, categorizing the features into core or optional. Later, some desired
features were added and categorized. The next step was use cases description for all features. During the
analysis phase, UML class diagrams were created, categorizing classes into core or variant for each feature,
followed by UML sequence diagrams for each use case. During design, groups of classes within features
were organized into components, before implementation and testing.
An application framework was chosen for the Domain Engineering implementation. In this case,
application engineering consists of the framework instantiation and specialization. Variability management at
code level is performed with two types of variability points in the framework for building iLM: (a) groups of
components that form an optional feature must be chosen, or not, to be part of the system; and (b) variant
classes within components that define specific behaviors of the system by inheritance. Therefore, during
application engineering, developers must choose which features the system will have, and the behavior of
each of them by inheriting abstract classes of the framework.
In the next section, we describe the development of Domain Engineering, represented by the first version
of the framework, and the detailed methods for application engineering.
5. DEVELOPMENT OF DOMAIN ENGINEERING FOR ILM
This section presents the development of Domain Engineering for building iLM. According to the Spiral
process and PLUS method described in the previous section, it was conducted in four steps: requirements
analysis, feature analysis, use case analysis and an iterative step of design, implementation and testing,
resulting in the framework prototype. The four steps will be detailed next.
5.1. Requirements Analysis
During the preliminary design and the requirements phase, regarding the framework development, the
analysis of desired and existing requirements resulted in the list below. Application engineering considers
additional requirements, which are described later.
 Documentation – a Domain Engineering framework must be well documented so that it is possible to
maintain it during domain application and to use it easily during engineering application.
 Extensibility – the framework for building iLM must be capable of absorbing new features and specific
features of iLM as the family of systems grows.

 Maintainability – existing iLM have poor maintainability. One of the main goals of using Domain
Engineering approach is to improve the code quality, which is key in this work.
 Open source – our project will follow an open source approach in order to provide free tools to
educational institutions and to allow a community to develop iLM.
 Platform compatibility – they must be compatible with any operational system in order to be used by any
educational institution.
 Resource constraints – some educational institutions in developing countries have outdated
computational resources; hence, iLM must be lightweight and demand low processing power.
 Reusability – as Domain Engineering, the framework most important requirement is code reusability.
 Testability – to ensure maintainability and reusability, the framework code must be highly testable.
Existing iLM explicit requirements were only platform compatibility, resource constraints and open
source. The additional requirements defined in this research mainly reflect systematic reusability in a
software life cycle longer than two years.
5.2. Feature Analysis
The features of the iLM family were defined by mapping functionalities of existing iLM and some additional
ones. Fig. 3 shows a feature model with the core and optional features described in section 3, divided into
two levels of sub-features.
Fig. 3: Feature model of the framework for building iLM.
This diagram follows the FODA, Feature Oriented Domain Analysis (Kang, 1990) notation, in which an
arc between features means alternative features and a circle means optional feature. An iLM has all
mandatory features (which are provided by the framework), one of the alternative features (which is the iLM
specific domain) and may have one or more optional features. A brief description of the sub-features is
presented.
 Domain Operations – contains all domain-specific functionalities that must be defined during application
engineering. It is an alternative feature; each iLM thus has a different implementation, such as geometry in
iGeom and counting in iComb. This sub-feature implements domain-specific manipulate and construct
types of interactivity.
 Assignment Management – the management of assignments is performed by this feature. It allows the user
to author and edit assignments as a teacher, and to solve an assignment as a student. The modeling of
assignments is domain independent, which allows this feature to be mandatory and provided by the
framework.
 Communication Management – provides the functionality for opening and saving files and for
communicating assignment data through the iLM protocol. These tasks can be entirely implemented
during Domain Engineering, thus it will be independent of the specific data being transferred.

 Configuration Management – this feature manages the system configuration, with parameters such as
language and teacher or student versions (to enable or disable assignment authoring). Configurable
behaviors specified during application engineering can also use this feature.
 Functionalities over Assignments – this feature has operations over assignments that are being solved by
the student or authored by the teacher. Examples are automatic checking of the student’s solution (which
is provided by the framework, even though considered as optional) and exporting to images. Other
functionalities may be included, such as exporting assignments as Learning Objects standard packages.
 Functionalities over Domain Operations – this feature manages the user’s domain operations during an
assignment. It provides domain-independent functionalities that facilitate or improve manipulate and
construct types of interactivity. For instance, it has sub-features of undo and redo, domain operations
history, domain object list (which are mandatory) and others, such as scripts and intelligent tutoring.
There are many feature interactions among mandatory features, which are omitted in Fig. 3 for simplicity,
and all of them depend on the existence of the Domain Operations feature. For instance, authoring and
editing assignments (from Assignment Management feature) depend on reading and writing files and on the
configuration of the teacher or student version. Since the framework handles all interactions among
mandatory features, during Application Engineering, developers of iLM do not need to consider them. In the
case of optional features, they can depend on each other (as Intelligent Tutoring depends on Tutor
Authoring), but no mandatory or alternative feature depends on them.
All subsequent development adopting the PLUS method is highly dependent on this feature division. Next
in the paper, feature means one of the groups of sub-features listed above. Use case analysis and each step
during prototype development is also divided into these sub-features.
5.3. Use Case Analysis
For designing the framework, the use case analysis was conducted for each mandatory or core feature (since
it will not provide alternative or optional features). As its result, five use case diagrams were created. Each
use case, depending on its sub-feature, is classified as core or optional.
Fig. 4 shows, as an example, a combined use case diagram for the Assignment Management and
Functionalities over Assignments features. Core sub-features are depicted in bold and optional sub-features
in italic. We briefly describe its use cases below:
Fig. 4: Use case diagram for the Assignment Management and Functionalities over Assignments features.
 Assignment Solution – the user playing the role of student opens the iLM with an assignment to be
solved; the assignment file is provided by the LMS server which embeds the iLM.
 Intelligent Tutoring – in the case of an iLM and an assignment with intelligent tutoring features, when
the file is opened, the tutor is also executed. After finishing the assignment, it sends tutoring data to the
LMS server.

 Automatic Checking – when the student wants to check whether his or her solution is correct, if
available, the automatic checking is launched. It sends the assignment result to the LMS which embeds
the iLM. The result can be stored for further analysis.
 Assignment Authoring – the user opens the iLM playing the role of teacher. All features for assignment
authoring are available to this role. The assignment file can be stored in the LMS server.
 Tutor Authoring – if the iLM has intelligent tutoring features, while authoring an assignment, the user
can set the tutor behavior.
 Assignment Editing – as a teacher, the user can open assignment files to edit from the LMS server page.
The available features are the same as in the Assignment Authoring use case.
 Assignment Export – the teacher may also, when available, export the assignment authored to standard
specifications such as IMS-LD (IMS, 2003) or SCORM (Advanced Distributed Learning, 2004).
According to the adopted development process, having finished the first iteration of software analysis for
Domain Engineering, the next step is to develop its first version, as described in the next section.
5.4. Framework development
After the use case analysis, the application framework development was composed of an iterative phase of
software analysis, design, implementation and testing. The Domain Engineering of a SPL intends to provide
code, architecture and process for reuse. The framework provides the code and architecture, while the method
for its use provides a systematic reusable process for developing iLM. This section describes the iterative
development steps, and then the results from each of the development iterations, including a discussion about
the influence of the used method and some lessons learned.
Software analysis takes data from requirements and creates a coarse structure of the software. In this case,
each use case diagram was analyzed individually in order to generate separated UML class diagrams. We
define code components that had a one-to-one relation with features. Variability management was considered
by choosing which classes had variant behaviors and which were constant. All diagrams had variant classes
only on user interfaces, except the one modeling the Domain Operations feature.
In the last iterations, variability management had little impact on implementation, as the architecture got
more concise and analysis was reviewed mainly to solve problems faced during implementation or to add
more flexible or simpler project choices. The PLUS method was of great importance during conception but
not during refinement. This may be due to the relative small size of the project and the number of people
involved.
The design phase transforms the coarse architecture from analysis into a well defined set of
communicating components. This was done using design patterns (Gamma, Helm, Johnson and Vlissides,
1995 and Gomaa, 2003) in the first iterations and refactoring techniques later (Fowler, Beck, Brant, Opdyke
and Roberts, 1999 and Kerievsky, 2004). The main product of this phase, obtained in the last iterations, was a
component architecture that reflected feature variability in a good manner. This was allowed by the PLUS
method and refined design decisions incrementally, which were used to restrain variability management only
to the analysis and design phases, making it more transparent during implementation. Consequently,
implementation could deal with each component individually, no matter how it would be used during
application engineering. It is important to highlight that this design level was not attained until the last
iterations, and was a result of several implementation efforts and iterations.
The implementation phase generates code from the designed architecture, and was a key phase to identify
when to start new development iterations. Testing phase was conducted along implementation with a small
time offset. Because of design decisions, considerations over variability management reduced over time,
together with the difficulty in programming.
Development took four main iterations, most of them dedicated to design the framework. At the first
iteration, design was under specifying, which hindered implementation. In the second iteration, design
activities took longer but the result was an over specifying design, which also impaired other phases. By the

third iteration, design was mature and most of the implementation was achieved. Finally, during the fourth
iteration, some design flaws were corrected and the method for using the framework was simplified. For
testing the framework, an iLM was created to serve as an example. This iLM-example has no pedagogical
objectives and its functionalities were defined in order to instantiate the flexible points of the framework.
The proposed method to use the developed framework also matured throughout the iterations by getting
more and more simple. Here, the PLUS method aided to model variability points which were implemented as
hotspots (Fayad and Schimidt, 1997), such that the instantiation of different of these flexible points were
similar. The framework started with many hotspots with no relation to each other and, as design was refined,
hotspots were merged and were organized into three groups: domain model, domain user interface and plug-
ins of functionalities over assignments and over domain operations. To document this method, we developed
three manuals, in ascending order of technical depth: (a) How to instantiate the framework; (b) How to
extend the framework with plug-ins; and (c) How the framework works internally. The most used was (a),
which has a step-by-step nature, providing a systematic process for developing an iLM. Moreover,
maintenance and evolution issues were considered throughout the manuals and code documentation.
Based on our experience in applying our process to a set of different learning systems, we have identified
the most important characteristics of the framework, namely iterative design and implementation, a powerful
method, good design decisions and improved by feedbacks from developers. Iterative design and
implementation made explicit errors early, reducing the amount of rework and turning the fixing task easier.
A powerful modeling method, such as PLUS, guided the early design, was over-used sometimes, and then
turned almost transparent in later iterations with the help of good design decisions, which made
implementation simpler. Lastly, documenting and the users’ opinion made visible some design that could be
enhanced.
5.5. Framework current state
The application framework that implements the Domain Engineering has a component architecture, as
shown in Fig. 5. This architecture can be divided into two groups: (a) the user interface, with the two
components in the upper part of the diagram; and (b) the features, with six components, one for each feature
in the feature diagram. Core components are shown in bold font, the variant components in italic and
underlined, and the optional ones in comic sans font. The framework currently provides all core components,
some abstract classes for the variant components, and some examples of components for the optional ones.
Fig. 5: Application framework’s component architecture.
Each of the components is detailed as follows:
 DomainUserInterface – is composed of abstract classes that are Java Swing components, which must be
inherited and specialized to provide domain specific behavior. The framework also provides
functionalities to create buttons and to trigger domain operations.

 BaseUserInterface – this is an empty frame with basic and domain-independent functionalities that is
responsible for initializing and for showing the DomainUserInterface component. Another responsibility
is the management of each module user interfaces (the plug-ins).
 DomainModel – the component that provides the model on which the domain operations must be
created. It is a group of abstract classes that model domain operations, objects and functionalities.
 AssignmentControl – a component for the management of assignments, which is also responsible for
managing plug-ins. The main control class connects assignments with plug-ins, with the communication
component and with the domain model.
 SystemControl – is responsible for initializing the whole system and for managing the configuration of
features. It also provides the public interface that is used by LMS for external communication.
 AssignmentModule – is the component responsible for setting the plug-in architecture to modules that
provide functionalities over domain operations. In addition to the provided architecture, the framework
has three modules: undo/redo, history of domain operations and a list of domain objects. These modules
can be used as examples or basis for the development of new plug-ins.
 SystemModule – is the component responsible for setting the plug-in architecture to modules that provide
functionalities over assignments. The framework also provides an example of system module: the
automatic checking module.
 Communication – this component has the functionalities related to communicating with other systems,
such as saving and reading files, opening assignments and sending evaluations to LMS.
The current state of the framework has around 5,000 lines of code within 50 classes and 12 interfaces.
Each component is a Java Package. All the features were been implemented, except for some details of
teacher and student versions and other configuration parameters. The code is available as open source at
http://ccsl.ime.usp.br/redmine/projects/ima/files.
Regarding the framework evolution and extension, the method proposes the inclusion of plug-in based
functionalities, as AssignmentModule and SystemModule components. A manual was created to document
the method in detail for this development. The proposed systematic method for using the framework to
develop an iLM is presented in next section.
5.6. Method for instantiating the Framework
Application Engineering is the process of using the products of Domain Engineering to develop systems
of the family, in this case, iLM. This section details the method of instantiating the application framework
and using the systematic process provided by Domain Engineering.
First, one must know which specific domain the iLM will provide to teachers and students. Among the
iLM features, the application framework provided by Domain Engineering does not have the domain
operations and can be extended by functionalities over domain operations and assignments. Hence, in order
to develop an iLM, in this phase one must define, within the chosen domain, what the domain operations are
and if it will use plug-in based functionalities.
Examples of domain operations, in existing iLM, are: in iGeom, creating points, lines, circumferences,
moving points, editing objects, reflecting objects to an axe; and in iGraf, creating functions, curves, points,
editing these objects, animating them, etc. Moreover, there are two types of plug-in based functionalities:
domain-specific and domain-independent. Domain-independent plug-ins, modules such as undo/redo and
automatic assessment, are considered framework extensions as they can be developed during Domain
Engineering. Nevertheless, it is possible that a certain domain needs specific functionalities, such as specific
intelligent tutoring and specific export engines (e. g., exporting iVProg programs to programming
languages). For this reason, domain specific plug-ins must be developed during Application Engineering.
Different cookbooks are proposed for each alternative or optional component in Fig. 5. In the case of
DomainModel component, each domain specific operation must be modeled as a class inheriting
DomainAction class, which is a Command design pattern (Gamma et al., 1995). These operations manipulate

objects of classes inheriting DomainObject class. The receiver of DomainAction objects is always the
Singleton concrete DomainModel object, the abstract class of which is also provided by the framework.
DomainModel implements all operations, it is the core of the iLM domain.
In the case of plug-ins, the modules of AssignmentModule and SystemModule components,
implementation is different for each kind. If they implement the AssignmentModule interface, they are
Observers of DomainActions and/or DomainObjects, so they must do something when an operation is
triggered and/or an object is created or destroyed. If they implement SystemModule, its functionality
manipulates an Assignment object, which has data of all DomainActions and DomainObjects, but also of
propositions, other modules, etc. These modules also have access to communication functionalities provided
by the framework.
DomainUserInterface is the remaining component. Its design can be produced as a standalone domain-
specific user interface, which must be put together using the Strategy pattern. In the end, the connection of
this user interface with the framework is made by inheriting the DomainGUI class. This class has some
methods that were designed to connect the interaction mechanisms of the user interface with the
DomainAction objects. Thus, for instance, when a button is pressed in DomainGUI, a DomainAction object is
triggered, which calls a DomainModel method to execute the functionality.
Another step in designing DomainUserInterface component is the user interface for authoring
assignments, which is also based on the Strategy pattern. The AuthoringGUI abstract class provided by the
framework must be inherited to allow the final user, usually a teacher, to author domain-specific assignments
for this iLM. This step is simply allowed by inheriting AuthoringGUI class and by implementing its abstract
methods, which are specific for each datum of the Assignment object.
In summary, the suggested method for Application Engineering using the framework provided by Domain
Engineering consists of: defining domain operations and objects, which are implemented as DomainAction
or as plug-ins, and designing two user interfaces, the domain and the authoring one. In the end, the complete
iLM consists of the framework, the chosen plug-ins, the objects of the domain model, and the specific user
interfaces. This method is described in detail in the manuals provided with the framework and is also
available (in Portuguese) with its code at http://ccsl.ime.usp.br/redmine/projects/ima/files.
6. RESULTS AND DISCUSSION
As mentioned in Section 1, the evaluation of complete results concerning an SPL and its products is a long
term task. Nevertheless, since an SPL can be divided into Domain Engineering and Application Engineering,
some intermediate results can be analyzed during each of the aforementioned phases. In fact, Domain
Engineering may be analyzed through the work products that will be generated by following its associated
process, the use of its underlying architecture and the provided code. Moreover, such analysis may be
provided by the description of Application Engineering for the iLM SPL, or even part of it, associated with
some metrics to evaluate what is already done.
Therefore, we analyze process, architecture and code reuse, and an estimative on quality and
maintainability for each system, following some pre-established criteria. Such criteria consider that some
existing iLM would be adapted to the proposed Domain Engineering in order to achieve the long term goals
of improving quality for maintainability and evolution, and that some iLM would be built from scratch. Thus,
process reuse would be analyzed by interviewing the developers before and after the use of Domain
Engineering. Architecture reuse would be analyzed by comparing the new structure of the system with the
previous one. Code reuse would be analyzed by evaluating how much of the code (%) could be discarded
from the old system. Quality and maintenance could be estimated by requirements and maintainability
analysis.
In this section, we describe how Application Engineering for the existing iLM is being conducted. We
present current products of the development process and we discuss the corresponding results by highlighting
some lessons learned. Currently, each iLM is being rebuilt or adapted in order to work with the framework.
Also, a new iLM to support Genetics teaching and learning is in its early stage of development, and it has
used the proposed method from the very beginning. Depending on the iLM specificities and developers’

profiles, their adaptation to the proposed Domain Engineering follows a different path towards the integration
to the framework. The current stage of adopting the Domain Engineering to the existing iLM is described as
follows.
iGeom is being adapted through disentangling the code, mainly between the user interface and the
geometry model. The separation of concerns was made in order to make the system architecture compatible
with that of the framework. Since iGeom is the oldest and the largest (46,000 lines of code), maintenance
issues were common and justified the effort of restructuring. Two programmers with professional experience
in programming are working on iGeom. They are refactoring the older version of iGeom by applying our
Domain Engineering approach, good practices of programming and professional tools. Thus they have
enhanced the development process and its products. Currently, iGeom is being integrated to the framework
code.
With iGraf and iComb, adaptation is similar to iGeom, but in smaller scales. iGraf has around 15,000 lines
of code (one third of iGeom) and was being restructured even before the Domain Engineering development
due to maintenance issues. As the framework reached stability, this restructuring was adapted to follow its
design. This task is performed by one developer with only academic experience in programming and it is
now being finished in order to be integrated to the framework. On the other hand, the adaptation of iComb,
with half the size of iGraf, also started with a non-experienced programmer and was delayed due to
administration issues. The current stage is still at the beginning of separation of concerns and domain model
specification.
iVProg is being rebuilt from scratch rather than adapted. Reasons for this includes the fact that iVProg
was created based on Alice (Carnegie Mellon University, 2012), a well established educational software, that
was simplified and adapted to be an iLM for 2 years. The current version of iVProg has around 37,000 lines
of code, most of them as Alice’s legacy, which makes it very hard to maintain and to modify. In the last three
months, one programmer with no professional experience is developing the new iVProg version, rebuilding it
from scratch using Domain Engineering for the family. The domain model and the domain user interface of
this new version are ready, now with good and clean design as guided by Domain Engineering. There was
gain in productivity not only due to the use of Domain Engineering but also due to the reuse of the domain-
specific feature design. The integration of the new iVProg to the framework code has just started.
In this context, process reuse analysis resulting from interviews with the developers shows that they came
from ad-hoc development, in which requirements were defined at any time and features were designed
directly on the code, to a guided development stated by the process provided by Domain Engineering. After
the beginning of the restructuration process, they knew beforehand what to do, and it was possible to know
where they were going in the long term. Consequently, development was more organized and satisfactory.
As far as code reuse is concerned, we can only estimate it since the refactoring of our learning systems is
not yet completed. Still, architecture reuse has had the most important impact on the developers’ work. The
percentage of code reuse decreases with the system size; iGeom has around 35% of code for domain
independent features and system structure, while iComb has over 45%. iVProg, due to its legacy code
structure, cannot be analyzed this way. The modularity of these systems increases when their architecture are
restructured in order to be compatible with the framework, and the responsibilities of each component in the
overall architecture gets clearer. A key point stated by iGeom and iVProg programmers is that with the
provided architecture, they do not need to spend time thinking of how the system code should be organized,
and this organization does not need to change over time due to including specific features.
The system quality may be evaluated through the requirements that were fulfilled. Since requirements that
are common to the iLM family are provided by Domain Engineering, system quality that relies on them can
be estimated by testing the framework. These tests are currently being conducted to assure this quality.
Moreover, architecture reuse can affect the quality of domain-specific features but, since they are
implemented with a previously tested structure, the tasks of specification, design and debugging are
benefited. Maintainability is mainly dependent on documentation, architecture and code quality. On the one
hand, for common structure and features, they are provided directly by Domain Engineering, being reviewed
by all programmers that use it, raising its quality iteratively. On the other hand, they are guided by Domain

Engineering, which also allows the iterative revision. These considerations, added to the fact that
restructuring leads to better quality systems from the initial stage of development, also contributes to making
maintenance easier.
7. CONCLUSIONS
Interactive Learning Modules (iLM) have the potential to improve teaching and learning quality by providing
adequate technology that meets teachers’ and students’ needs in both classroom and distance education
settings. However, the development of this software family is complex and time-consuming, mostly because
the lack of methods and processes that help its development. To overcome this problem, we proposed a
Domain Engineering within a Software Product Line (SPL) approach to define and to guide the iLM
development process.
To characterize the iLM software family, we conducted an initial analysis of five iLM developed to
enhance learning in different domains (Geometry – iGeom, Mathematical Functions – iGraf, Counting –
iComb and Programming – iCG and iVProg). This initial result helped to identify common and desired
features that are essential to keep the functional consistency among them and it provided the foundations to
build an application framework. With this framework, Domain Engineering provides documentation to foster
process, architecture and code reuse during the development of iLM.
The main contribution of this work is the centralization of knowledge about the iLM family as the result
of Domain Engineering, in the proposed application framework. Before that, knowledge was spreaded among
existing systems, now detailed specifications and code can be shared among all iLM with the support of our
framework. This is important in order to increase standardization among common functionalities of existing
and new iLM and to facilitate the understanding of the systems developed in our research group by others.
The proposed Domain Engineering provides code, architecture and process reuse, and code reuse usually gets
more attention. Besides the importance of code reuse, our results show that architecture and process reuse
have great influence on systems development, mostly if they were ad-hoc implemented before.
Research on evaluating the use of software engineering methods and techniques for the development of
educational software is at its early stages. This work intends to disseminate the idea of using and evaluating
these methods and techniques by showing preliminary results of the evaluation of applying a SPL approach
to develop or to refactor iLM. In the long term, we expect to provide continuously better evaluation of the
software engineering techniques used, while improving the educational software development process.
Future work consists of finishing the adaptation and rebuilding of existing iLM. After that, we could better
analyze code reuse provided by Domain Engineering. Also, new iLM are being created from scratch using
Domain Engineering, which can generate new development results as well as new data for studying its
influence. As a long term goal, maintenance and evolution of existing iLM will be monitored in order to
evaluate maintenance gains.
8. ACKNOWLEDGMENTS
Danilo L. Dalmon is supported by FAPESP under grant 2010/06805-2. This work was partially supported by
FAPESP (2011/10926-2) and CNPq (550449/2011-6).
9. REFERENCES
ADVANCED DISTRIBUTED LEARNING, (2004): Shareable Content Object Reference Model 2004 Specification,
www.adlnet.gov/capabilities/scorm accessed in December 2011.
ALBANESE, M. A. and MITCHELL, S., (1993): Problem-based learning: A review of literature on its outcomes and
implementation issues, Academic Medicine, vol. 68, pp.52-81.
ALEVEN, V., MCLAREN, B. M., SEWALL, J., and KOERDINGER, K. R., Example-tracing tutors: A new paradigm
for intelligent tutoring systems. International Artificial Intelligence in Education Society, 2008, pp. 105–154.
AHMED, F., and ZUALKERNAN, I. A., (2011): A Software Product Line Methodology for Development of E-
Learning System, International Journal of Computer Science and Emerging Technologies, Vol. 2, pp.285-295.
ATEYEH, K., and LOCKERMANN, P. C., (2006): Reuse- and Aspect-Oriented Courseware Development,
Educational Technology and Society.

BOEHM, B., (1986): A Spiral Model of Software Development and Enhancement, ACM SIGSOFT Software
Engineering Notes, vol. 11, pp.14-24.
BOTE, M. L., HERNANDEZ, D. L., DIMITRIADIS, Y.A., ASENSIO, J. I. P., GOMEZ, E. S., VEGA, G. and
VAQUERO, L. M. G., (2004): Towards Reusability and Tailorability in Collaborative Learning Systems using
IMS-LD and Grid Services, International Journal on Advanced Technology for Learning, vol. 1, pp.129-138.
BOYLE, T., (2003): Design Principles for Authoring Dynamic Reusable Learning Objects, Australian Journal of
Educational Technology, vol. 19, pp. 46-58.
CARNEGIE MELLON UNIVERSITY, (2012): Alice – An educational software that teaches students computer
programming in a 3D environment. www.alice.org. Accessed 19-June-2012.
CHOQUET, C. and CORBIÈRE, A., (2006): Reengineering Framework for Systems in Education, Educational and
Technology, Vol. 9, No 4, pp. 228-241.
CLEMENTS, P. and NORTHROP, L., (2001): Software Product Lines: Practices and Patterns (The SEI Series in
Software Engineering), Addison-Wesley Professional.
DALMON, D., TANBELLINI, M., EISENMANN, A., NASCIMENTO, M., RODRIGUES, P., DO PRADO, R.,
KAMIYA, R., ISOTANI, S., BRANDÃO, A. and BRANDÃO, L., (2011): Interactive Learning Modules in
Engineering Education and as a Motivational Tool for Middle and High School Students, Proceedings of
International Symposium on Engineering Education – IGIP.
DALMON, D. L., ISOTANI, S., BRANDÃO, A. A. F. and BRANDÃO, L. O., (2011): Work in Progress - A
Framework for Building Interactive Learning Modules, Proceedings of Frontiers in Education.
DALMON, D. L., ISOTANI, S., BRANDÃO, A. A. F. and BRANDÃO, L. O., (2011): Work in Progress - Enhancing
Interactive Geometry Systems with Intelligent Tutoring Features, Proceedings of Frontiers in Education, 2011.
DO PRADO, R. and BRANDÃO, L.O., (2006): iGraf: Módulo de Aprendizagem para Ensino de Função na Web, XIV
Brazilian Symposium on Informatics in Education (in Portuguese).
DODERO, J. M. and DIEZ, D., (2006): Model-Driven Instructional Engineering to Generate Adaptable Learning
Materials, Proceedings of International Conference on Advanced Learning Technologies, pp. 1188-1189.
DOUGLAS, I., (2001): Instructional Design Based on Reusable Learning Objects: Applying Lessons of Object-
Oriented Software Engineering to Learning Systems Design, Proceedings of Frontiers in Education.
EISENMANN, A. L. K. and BRANDÃO, L. O., (2009): iComb: um sistema para o ensino e aprendizagem de
combinatória em ambiente Web, XXBrazilian Symposium on Informatics in Education (in Portuguese).
FAYAD, M., SCHIMIDT, D. C., (1997): Object-oriented application frameworks. Communications of the ACM,
vol. 40, pp. 32-38.
FOWLER, M., BECK, K., BRANT, J., OPDYKE, W. and ROBERTS, D., (1999): Refactoring: Improving the Design
of Existing Code, Addison-Wesley Professional.
GAMMA, E., HELM, R., JOHNSON, R., and VLISSIDES, J. (1995): Design Patterns: Elements of Reusable Object-
Oriented Software. Addison-Wesley, Reading, MA.
GOMAA, H., (2004): Designing Software Product Lines with UML: From Use Cases to Pattern-Based Software
Architectures, Addison-Wesley Professional.
HARA, N., KLING, R., (1999): Students' Frustrations with a Web-Based Distance Education Course. First Monday:
Journal on the Internet, 4(12). http://131.193.153.231/www/issues/issue4_12/hara/index.html
HENTEA, M., SHEA, M. J., and PENNINGTON, L., A (2003): A perspective on fulfilling the expectations of distance
education, Proceedings of Conference on Information technology Curriculum, pp.160–167.
HINOSTROZA, E., REHBEIN, L. E., MELLAR, H. and PRESTON, C., (2000): Developing Educational Software: a
Professional Tool Perspective, Education and Information Technologies, vol. 5, pp.103-117.
IMS (2003): IMS Learning Design Specification, Boston: The IMS Global Learning Consortium, Final Specification.
ISOTANI, S. and BRANDÃO, L. O., (2008): An algorithm for automatic checking of exercises in a dynamic geometry
system: iGeom, Computers and Education, vol. 51, pp. 1283-1303.
KANG, K.; COHEN, S.; HESS, J.; NOWAK, W.; PETERSON, S. (1990): Feature-Oriented Domain
Analysis (FODA) – Feasibility Study (Technical Report). Software Engineering Institute, Carnegie
Mellon University. CMU/SEI-90-TR-21.
KAMIYA, R. H., BRANDÃO, L. O., (2009): iVProg - um sistema para introdução à Programação através de um
modelo Visual na Internet, XXBrazilian Symposium on Informatics in Education (in Portuguese).
KERIEVSKY, J., (2004): Refactoring to Patterns, Addison-Wesley Professional.
KOERDINGER, K. R. and CORBETT, A. T., (2006): Cognitive Tutors: Technology bringing learning science to the
classroom. In K. Sawyer (Ed.) The Cambridge Handbook of the Learning Sciences. Cambridge University Press.

KORTENKAMP, U and RICHTER-GEBERT, J., (2004): Using automatic theorem proving to improve the usability of
geometry software, Proceedings of MathUI (Mathematical User Interfaces).
LINDEN, F.J. van der, SCHMIDT, K. and ROMMES, E., (2007): Software Product Lines in Action: The Best
Industrial Practice in Product Line Engineering, Springer-Verlag New York, Inc. Secaucus, USA.
MAYES, J.T. and FOWLER, C.J., (1999): Learning technology and usability: a framework for understanding
courseware, Interacting with Computers, vol. 11, pp. 485–497.
MOR, Y. and WINTERS, N., (2007): Design Approaches in Technology-Enhanced Learning, Interactive Learning
Environments, vol. 15, pp.61–75.
MUZIO, J. A., HEINS, T. and MUNDELL, R. (2002): Experiences with reusable E-learning objects From theory to
practice, Internet and Higher Education, vol. 5, pp. 21–34.
NASH, S. S., (2005): Learning Objects, Learning Object Repositories, and Learning Theory: Preliminary Best
Practices for Online Courses, Interdisciplinary Journal of Knowledge and Learning Objects, vol. 1, 2005.
NICOLSON, R.I and SCOTT, P.J., (1986): Computers and Education: the software production problem, British
Journal of Educational Technology, 17: 26–35. DOI: 10.1111/j.1467-8535.1986.tb00494.
OBERWEIS, A., PANKRATIUS, V. and STUCKY, W., (2007): Product Lines for Digital Information Products”,
Information Systems, vol. 32, pp. 909-939.
ORACLE, (2012): Java SE Technologies. http://www.oracle.com/technetwork/java/javase/documentation/spec-
136004.html. Accessed 19-June-2012.
PANKRATIUS, V., STUCKY, W. and VOSSEN, G., (2005): Aspect-oriented reengineering of e-learning courseware,
The Learning Organization, vol. 12, pp. 457-470.
POLSANI, P. R., (2003): Use and Abuse of Reusable Learning Objects, Journal of Digital Information, vol. 3.
QUINTANA, C., KRAJCIK, J., SOLOWAY, E. and NORRIS, C., (2002): A framework for understanding the
development of educational software. In The human-computer interaction handbook, Julie A. Jacko and Andrew
Sears (Eds.). L. Erlbaum Associates Inc., USA, pp. 823-834.
RICHARDS, G., MCGREAL, R., HATALA, M. and FRIESEN, N, (2002): The Evolution of Learning Object
Repository Technologies: Portals for On-line Objects for Learning, Journal of Distance Education, vol. 17, pp.
67-79.
RODRIGUES, P. A., BRANDÃO, L. O. and BRANDÃO, A. A. F., (2010): Interactive Assignment: a Moodle
component to enrich the learning process, Proceedings of Frontiers in Education, pp. T4F1-T4F6.
ROSCHELLE, J., KAPUT, J., STROUP, W., and KAHN, T. M., (1998): Scaleable Integration of Educational
Software: Exploring the Promise of Component Architectures, Journal of Interactive Media in Education.
ROSCHELLE, J., DIGIANO, C., CHUNG, M., REPENNING, A., TAGER, S., and TREINEN, M., (2000):
Reusability and Interoperability of Tools for Mathematics Learning: Lessons from the ESCOT Project,
Proceedings of Intelligent Systems and Applications, pp. 664-669.
RUNESON, P., HÖST, M., (2009): Guidelines for conducting and reporting case study research in software
engineering, Empirical Software Engineering, DOI 10.1007/s10664-008-9102-8, pp. 131-164.
RUTHVEN, K., HENNESSY, S., and DEANEY, R., (2007): Constructions of dynamic geometry: A study of the
interpretative flexibility of educational system in classroom practice, Computers and Education, vol. 51, pp. 297-
317.
SAMPSON, D. G. and KARAMPIPERIS, P., (2006): Towards Next Generation Activity-based Learning Systems,
International Journal on e-Learning, vol. 5, pp. 129-149.
SAVERY, J. R. and DUFFY, T. M., (1995): Problem Based Learning: An instructional model and its constructivist
framework, In Constructivist Learning Environments: Case Studies in Instructional Design, Ed. Wilson, B.
SIMS, R., (1997): Interactivity: a forgotten art?, Computers in Human Behavior, vol. 13, pp. 157-180.
SPALTER, A. M. and VAN DAM, A., (2003): Problems using Components in Educational Software, Computers and
Graphics, Vol. 27, pp. 329-337.
TANG, B. C., (2004): Interactive e-learning activities to engage learners – A simple classification. Proceedings of
World Conference on Educational Multimedia, Hypermedia and Telecommunications, pp. 4092-4097.
TCHOUNIKINE, P., (2011): Computer Science and Educational Software Design, DOI 10.1007/978-3-642-
20003-8_1, Springer-Verlag Berlin Heidelberg.

A Domain Engineering For Interactive Learning Modules

More Related Content

Similar to A Domain Engineering For Interactive Learning Modules (20)

More from Nathan Mathis (20)

Recently uploaded (20)

A Domain Engineering For Interactive Learning Modules