Abstract:

Modern e-Learning systems should support many different functionalities. This paper deals with three of such aspects; first, the user should be allowed to choose among heterogeneous access modalities, on the basis of his preferences and technologies. The second issue concerns interactive learning, which should fit the students’ achievements. Third, it is necessary to design an efficient organization of both assessment and learning contents.
In this paper, an architecture is proposed for implementing such functionalities in an integrated environment. Particular attention is paid to adaptive testing, characterized by the process of tailoring the difficulty or contents of the next question/task or timing aspects on the basis of earlier responses, meant as the core of the whole process. The discussed architecture is a multi-layered one and reflects the above classification of functionalities.
As for fruition modalities (external layer, user-system interface module), new technologies made e-learning evolve into the broader concept of e/m-learning, where tests and learning material can be accessed by fixed or mobile users through heterogeneous network technologies (UMTS, DSL, WiFi, wired/fiber) and devices (mobiles, PDAs, laptops, desktops). This layer is in charge of: (i) managing authentication and access rights; (ii) adapting contents to different devices and connection speeds.
As far as interactive learning modalities are concerned (middle layer, decision module), a testing sub-module and a learning one are provided for. In the proposed approach, the testing phase guides both the evaluation and the choice of further learning steps.
Before describing how such module acts, it is worth defining what testing means in this context and what was meant over the years. Computerized Testing can be classified in four generations:

(i) administering conventional tests by computer
(ii) tailoring the difficulty or contents of the next step or the timing to users’ responses
(iii) using calibrated measures embedded in a curriculum to estimate changes in the user’s achievement and profile as a learner
(iv) producing intelligent scoring, interpretation of individual profiles, and advice to learners and teachers, by means of knowledge bases and inferencing procedures

With respect to the above classification, the proposed architecture belongs to level (iv). As a matter of fact, the adaptive testing module in the middle layer manages access to learning material by means of a Petri Net. On the basis of the testing results, such Petri Net allows to access given learning material with a given difficulty or not and assigns intelligent scoring. The novelty lies in two aspects: (1) use of a Petri Net in the testing module and (2) definition and use of intelligent scoring to make the Petri Net evolve. The Petri Net in the assessment module will control the evolution of intelligent testing and, in consequence, the access to learning contents.
The inputs to the Petri Net are issues and learning levels; the outputs are issues and levels used to define the Net marking. Such outputs are used as inputs to the learning module and can be interpreted as the degrees of difficulty the user was judged to be able to face.
As for information contents (internal layer, database module), they are divided into assessment material and learning material, and hierarchically arranged on the basis of degree of difficulty by means of two LDAP subtrees.

Keywords:

E-Learning, computer-assisted testing, adaptive testing, integrated architectures, Petri Nets.