Grapple Summary

The Generic Responsive Adaptive Personalized Learning Environment (GRAPPLE) project aims at delivering to learners a technology-enhanced learning environment that guides them through a life-long learning experience, automatically adapting to personal preferences, prior knowledge, skills and competences, learning goals and the personal or social context in which the learning takes place. The same environment can be used/accessed at home, in school, at work or on the move (using mobile/handheld devices).

GRAPPLE will include authoring tools that enable educators to provide adaptive learning material to the learners, including adaptive interactive components (visualizations, simulations, virtual reality). Authoring includes creating or importing content, assigning or extracting meaning from that content, designing learning activities and defining pedagogical properties of and adaptation strategies for the content and activities.

To ensure the wide adoption of adaptation in technology-enhanced learning environments, GRAPPLE will work with Open Source and commercial learning management system (LMS) developers to incorporate the generic GRAPPLE functionality in LMSs. Evaluation experiments in higher education and in industry will be performed to verify the usability of the GRAPPLE environment (for authoring and delivery) and to verify the benefits of using adaptive environments for the learning outcome. Apart from stimulating the use of adaptive learning environments by making it available to every organization using a (popular) LMS, the GRAPPLE consortium will also organize training/evaluation events to help higher education institutes and companies with the adaptive learning design needed to create adaptive learning material, and to receive usability feedback which the project will use to improve the user interfaces.

A distributed user modelling service architecture will help end-users to stay in control of their user profile while at the same time allowing them to use the profile to get personalized access to learning applications offered through different LMSs by different organizations.

First Year

GRAPPLE has just completed its first year of collaborative research and development. The year, in which requirements analysis was performed, followed by design and specifications. In addition to what was planned some early prototypes were developed as well.

Through a serious of interviews with potential users in academia and industry requirements for adaptive learning systems were gathered. Three perspectives were considered: that of learners, tutors and authors. As an example to guide the interviews an adaptive business English course was created (using material from the BBC) and implemented in the most commonly used open source adaptive learning environment at the moment, named AHA! (version 3) developed at Eindhoven University of Technology. Some important results of the requirements analysis are:

•  Adaptation should be based on the learner’s knowledge as well as learning goals.
• Tests to determine the learner’s knowledge are considered the most important (reliable) measure of that knowledge.
• Adaptive content and learning activity selection and adaptive navigation support are considered the most valuable adaptation techniques.
• Authoring support (through wizards and/or templates) is needed to reduce the complexity of authoring adaptive course material.

Using these requirements three main lines of activity were started: the work on authoring tools, on user (learner) model services and on the adaptation engine. This was complemented by the design of a common infrastructure aimed at bringing these three (sets of) components together.

The figure below shows the main components that are involved in the authoring process, leading up to delivery in the GRAPPLE Adaptive Learning Environment (GALE), i.e. the engine that provides serves the adaptive courses.

Figure 1: Information flow in the authoring process.

When designing a course an author will first design a domain model (DM) consisting of concepts, services and perhaps other elements such as conceptual elements from a virtual reality or a simulated dialog domain. Apart from semantic relationships that may exist between elements of DM there may be relationships that are of a pedagogic nature and that are used in the adaptation/personalization process. These are of certain types, which we call concept relationship types (CRTs) and are instantiated (i.e. used to connect actual concepts and services) in the conceptual adaptation model (CAM). The adaptation model may also use application-independent properties that are known about the learner, stored in the user model (UM). The authoring part of GRAPPLE thus mainly consists of an editor for DM, CRTs and the CAM.

In order to use these models in GALE a two-stage compiler is being developed. First the CAM is translated to an adaptation-engine-independent language GAL: the GRAPPLE Adaptation Language. This defines the concrete adaptation that is desired, but thus so using abstract constructs like concepts, attributes and queries (to retrieve information from the user model for instance). GAL “programs” can in principle be compiled into concrete adaptation rules for any adaptation engine. Within GRAPPLE we develop a compiler only for the GRAPPLE engine GALE.

In order to allow early experimentation in GRAPPLE work on the adaptation engine (to be used as a stand-alone adaptive learning environment as well as in combination with an LMS) was not limited to the specification level but a first version of the adaptation engine was built as well. Although based on ideas from the Adaptive Hypermedia Architecture (AHA!) developed at the Eindhoven University of Technology the new adaptation engine GALE is a completely new, modular and extensive adaptation engine. Its main components: user modelling service, domain/adaptation model service and adaptation engine, communicate with each other through an event bus that is accessible directly by other Grapple components. The figure below shows the overall structure of GALE:

Figure 2: Core GALE architecture

GALE can in principle perform adaptation to any XML format, using a highly configurable (and even adaptive) pipeline of processing modules.

Within GRAPPLE information about users exists in different places: GALE stores information that is needed in order to perform the actual adaptation. An LMS may store personal information, course registration, grades and other information. Learners will in the future increasingly obtain their learning experience through different sources, and as a result have information about themselves and their learning progress stored in different places.

In GRAPPLE an ontology-based user model format will be used. Information about users can be retrieved from (and uploaded to) different sources through the GRAPPLE User Modelling Framework, of which the architecture has been defined. Both for the GRAPPLE User Modelling Framework and for the communication between LMSs and GALE it is necessary to reach agreement on “who is who”. The use of the Shibboleth single sign-on framework is being considered as a platform to provide the single sign-on facilities and proper user identification and authentication within the entire GRAPPLE infrastructure. First prototypes of combinations of a single LMS and GALE have already been experimented with (without Shibboleth still). The LMSs that can be used together with GALE, providing one integrated presentation, are Moodle, Sakai, Claroline, CLIX and LearneXact.