QUERY IDENTIFICATION COMPONENT FOR THE PARES PLATFORM Avgoustos Tsinakos Associate Professor Konstantinos Georgakopoulos Adjunct Professor ABSTRACT The current paper describes the implementation of a Query Identification Component (QIC) to improve the Test Composer of the PARES platform (Platform for Adaptive and Reliable Evaluation of Students). QIC objective is to ensure the singularity of the questions inserted to the Test Component in order to avoid duplications. To achieve this, QIC make use of ontology thesaurus, controlled vocabulary and case based reasoning techniques. KEYWORDS Knowledge-based systems, case based reasoning. 1. INTRODUCTION In nowadays asynchronous distance education appears as one of the most appealing studying methods as it combines flexible access of teaching material, self-study techniques, and peer-to-peer collaboration. Though, form the origins of this education method, literature reveals a number of pedagogical drawbacks which are still valid. Indistinct monitoring of student performance/ progress, the inability to record individual learning needs, and the lack of student modeling processes (Wulf, 1996; Pernici and Casati, 1997; Relan and Gillani, 1997; Hunt 1999; Pritichard, 1998; Khan, 1997; Harasim, 1990; Hall, 1997b) are just some of the most commonly reported drawbacks. To remedy some of the above mentioned problems a number of sophisticated platforms have been developed such as WebCT, Blackboard (commercial products) or Moodle, ATutor (open source) which offer many tools and functionalities. Such platforms provide some tools for E-testing purposes, such as multiple choice tests or automate generated questionnaires, though the aspects of E-testing are still under the development phase. Authoring tools and test banks for E-tests are becoming an integral and mandatory part of E-learning platforms. Some sophisticated platforms are reported in literature such as Cadmus (Hage and Aïmeur, 2006), Plateau Exams (http://www.plateau.com/prod/exams.htm) and PARES (Kaburlasos et all, 2004). Though, dealing with E- testing issues a new challenge arises- ensuring that created exams are free of conflicts. In many cases conflict exists in an exam if two or more questions are redundant in content, and/or if a certain question reveals the answer of another question within the same exam. Selecting questions depending on the teacher’s preference cannot guarantee a flawless exam with no conflicts. Such conflicts within an exam might be frequent typically when a teacher is using shared questions authored by others, and especially in the automation of the exam creation process. The current paper describes an implementation called Query Identification Component (QIC) to identify duplication or overlapping questions/ answers in PARES platform so that the produced exam tests to be free of conflicts. IADIS International Conference e-Learning 2008 89