A Context-Aware Service Oriented Framework for
finding, recommending and inserting learning objects
Xavier Ochoa2, Stefaan Ternier1, Gonzalo Parra2, and Erik Duval1

1
Dept. Computerwetenschappen, Katholieke Universiteit Leuven,
Celestijnenlaan 200A, B-3001, Heverlee, Belgium
{Stefaan.Ternier, Erik.Duval}@cs.kuleuven.be

2
Information Technology Center,
Escuela Superior Politécnica del Litoral, Ecuador
{ xavier, gparra}@cti.espol.edu.ec
Abstract. In this poster, we will propose a framework for finding,
recommending and inserting learning objects in a digital repository level,
exploiting the user context that is captured from the Learning Management
System (LMS). The framework we propose builds on top of the ARIADNE
service oriented architecture for learning object repositories, abstracting from
the technicalities of low level metadata and resource management. As a case
study, the framework has been applied in a university learning management
system (SIDWeb). The intent is to exploit contextual information from the
learning management system in digital repositories.
Keywords: Service Oriented Architectures, Learning Management Systems,
Recommender services
1 Introduction
Learning Object Repositories are digital libraries with a special focus on storing
resources that can serve in a learning scenario. As creating learning resources is
expensive in time and hence in costs, it is vitally important that reuse of these
materials is as easy as possible. Recently, a lot of effort has been put in standardizing
both the metadata and services for describing and managing a learning object.
Learning Object Metadata enables tools to attach properties on Learning Objects in a
standardized way so that it does not matter where the metadata was produced,
facilitating the reuse of the metadata. Standardization efforts like the Simple Query
Interface [1] or the Open Archives Initiative for Metadata Harvesting [2] provide
means to query or harvest repositories in a standardized manner so that searching
materials over the boundaries of different repositories results in a critical mass of
learning content that is now available.
While most of the technical barriers for the sharing of learning objects have been
eliminated, the whole process of sharing, that is publishing the metadata in a
repository and finding relevant objects, is outside the workflow of the intended users.
LORs are stand-alone entities, usually separated from the normal environment

(usually a Learning Management System) where the instructor uploads learning
materials for his/her students. In this paper we will present an architecture that
enables eLearning users to share their content automatically each time that they add it
to their environments and to find new materials while they are working, rather than
have to search for them. With this approach, we hope to create a scenario of
publishing and finding resources that fits better in the workflow of an LMS user as
he/she might already find a reusable piece of content before realizing the need for it.
As an example of this deep integration we will glue together an existing Learning
Object Repository (ARIADNE[3]) with a run-of-the-mill Learning Management
System (SIDWeb[4]).
2. Use cases
To extend the capabilities of SIDWeb (and other LMSs), we envision different
scenarios where the work of the instructor or learner could be improved by the use or
more sophisticated (tailored) learning object services. Those scenarios are
summarized in the following use cases:
• Instructor searching for new material
• Instructor is creating content
• Instructor is inserting a new object and gets recommended similar
complementary objects.
• Student is reading content and wants to explore more material
• Teachers/students want to tag/annotate a learning object

A major ingredient to be able to provide this use cases is capturing the context. The
paper doesn’t aim to present a general way to deal with context, but will present an
adhoc representation of the format of context as it is available in SIDWeb and most
LMSs.
3. Core Services
In this chapter, we will present the core services necessary to provide
implementations for the use cases presented above. Some of these services have
already been standardized. The advantage of such a standardisation is that a repository
that implements them can easily be plugged in a framework that consumes these
services. These core services do not only make repositories more pluggable, they also
intent to make them interoperable. Using these core services, repositories can be dealt
with in a more generic way.
With core services we aim to describe indivisible units of interactions that provide
some basic interaction. That means services that do not invoke other services in the
background. The needs drawn from our use cases will require 2 kinds of services:
• Repository: User authentication and configuration management, insertion
and searching for material

• Third Party Services: Metadata generation, keyword extraction, tracking
services, etc.
In the following, we will go in more on detail on each of these services. For each
service, a programming language-agnostic signature of the methods involved will be
provided as well as a description of the context in which this service is to be used.
Apart from that, we will outline in what specials cases exceptions can occur.
3.1 Repository
3.1.1 Session Management (SM)

This first core service allows for making abstraction from access and security issues.
A session will be identified by a token and can last either for a fixed period or forever.
Session Management services provide means to request and terminate tokens that are
used to invoke the other core services.
3.1.2 Query Service (SQI)
For the query service, we will take over the methods provided by the Simple Query
Interface. In this standard, the most important method is the following.
3.1.3 Insert service (IS)
This service allows for inserting and updating resources together with a description
into a repository. The insertResource method enables shipping a resource together
with the metadata to a repository.
3.1.4 Annotate service (AS)
Apart from the insert service that provides support for submitting the learning object
and its metadata, there is also a need to submit user related information to the
repository. The following service is meant to submit metadata about the usage of the
object rather than the object itself. This information can be a comment written by a
user, but it can also be a label that the user or the system wants to attach to the object.

3.2 Third party services
3.2.1 Translate Service (TS)
This service will translate a word or a sentence into another given language, indicated
by the parameter toLanguage. This specification of this service is agnostic about the
way it is implemented. It can work with a small fixed set of keywords or can use a
full-fledged dictionary in the background. Context is an optional parameter that meant
to can solve ambiguity.
3.2.2 Keyword Generator Service (KGS)
This service will extract the keywords from a piece of free text. The array that is
returned has the length specified by the amountOfKeyword parameter.
3.2.3 Automatic Generation of Metadata (AMG)

This service performs automatic metadata extraction from the content and context of
learning objects. More information about this service could be found at [5]. The
most important function of this service is:
3.2.4 Tracking Service (TS)

With this service, applications can send data about the actions of users to a server.
This information can help later with recommending materials to that user.
4. Tailored Services
With tailored services, we target services that consume other services and are hence
not atomic. The tailored services that will be described in this section are tailored to
needs outlined in the use cases. Using these tailored services offers 2 main advantages
over using core service.
• Just like core service abstract from the way resources and metadata are
managed by a repository, these services will hide the specific
implementation of a service. In that respect the implementation of these
services is not meant to be unaltered. The API that is offered by this service
should be kept stable.
• These services are designed so that they fit the needs and specificities of a
LMS and hide the implementation details. It is thus easier to integrating them
into an LMS.

Fig. 1. An LMS can either directly access the core services or manipulate them through the
tailored services layer.
4.1 Example Enhanced Search (EES)
This service recommends new objects based on an already existing resource. An LMS
invokes this service with a metadata instance and the context, describing e.g. the user
of the course in which the object should be used. A valid metadata instance might be
e.g. one that only contains the identifier (of an instance available in the digital
library). This service will return a list of objects that is recommended.
4.2 Context Enhanced Search (CES)
With this service the context in which a query is executed can be taken into account.
Context information can be used here to provide better results and as input for a
ranking algorithm. This service will first feed the context as text to the Keyword
Generator Service, which will return a list of keywords. These keywords will next
serve to enhance the query that will be sent to SQI. In practice, as most of the SQI
implementations provide a keyword based query format, the query that will be sent to
SQI will be a list of search terms.

4.3 Automatic Insert Service (AIS)
As a learning management systems focuses on how learning objects are used, we need
a service that fits this approach. With this service, an LMS can submit a learning
object together with information describing the context in which the object is to be
used. With such a service being invoked in the background, a learning management
system no longer needs to provide the metadata describing a learning object
manually.
4.4 Enhanced Annotation Service (EAS)

This service will help to enrich the annotations made by the user, including into the
annotation contextual information that facilitate the understanding of it by other users.
6. Future work
This paper presents an analysis of the requirements of an LMS and derives from
these requirements context-aware services. Future work will focus on filling the gaps
that this paper does not address:
• A query language for SQI.
Keyword based query languages are easy to implement and to use by an
end user. A context enhanced search service could however benefit from a
richer query language. As an example, it could use such a query language
to limit queries to a given age span or discipline, using the information
available in the context
• Generalize Context.
This paper uses as an example the contextual information that SIDWeb
can deliver. In order for the framework to become useful in other learning
environments, the schema that describes the contextual information the
can be captured should be further enriched.
References
1. Simon, B., Massart, D., van Assche, F., Ternier, S., Duval, E., Brantner, S., Olmedilla, D.,
Miklós, Z.: A Simple Query Interface for Interoperable Learning Repositories. Workshop on
Interoperability of Web-Based Educational Systems in conjunction with 14th International
World Wide Web Conference (WWW'05). Chiba, Japan (2005)
2. OAI. The Open Archives Initiative. http://www.openarchives.org
3. Ariadne. http://www.ariadne-eu.org
4. SIDWeb, Information Technology Center, ESPOL. http://sidweb.espol.edu.ec
5. Ochoa, X., Cardinaels, K., Meire, M., Duval, E., Frameworks for the Automatic Indexation
of Learning Management Systems Content into Learning Object Repositories, Proceedings
of EDMedia 2005, Montreal, Canada, (2005) 1407-1414