The issue of rewarding partially correct answers has been addressed by many authors (Guzman, E. & Conejo, R., 2004, Gardner-Medwin, A.R. 1995, Huffman, D, Goldberg, F., & Michlin, M. 2003). Intelligent systems have been designed to assign scores related to the importance of missing or incorrect part of an answer. Such systems are meant to facilitate the process of knowledge assessment. While trying to be efficient in evaluating students’ responses these systems operate with the answers to a single question addressing learning a new term, understanding a new concept or mastering a new skill. However, experimental practice shows that asking several questions about the same item results in inconsistent and/or incomplete feedback, i.e. some of the answers are correct while others are partially correct or even incorrect. A large number of computer based systems and thus automated assessment systems lack the ability to reason with inconsistent information. Such a situation occurs when, f. ex. a student answers to two questions about one item and one of the answers is correct and the other one is incorrect or missing. Reasoning by applying classical logic cannot solve the problem because the presence of contradiction leads to trivialization, i. e. anything follows from ‘correct and incorrect’ and thus all inconsistencies are treated as equally bad (Priest, 2001). In this paper we discuss how to assess students’ understanding of new terms and concepts, shortly after they have been introduced in a subject. Application of many-valued logic allows the system to give meaningful responses in the presence of inconsistencies. Decision making rules, an intelligent agent is applying for assessing students’ understanding of new terms and concepts are presented. Such rules distinguish between students’ hesitation in the process of giving an answer and lack of knowledge. We propose use of the generalized Lukasiewicz’s logic in a Web-based assessment system as a way of resolving problems with inconsistent and/or incomplete input.
Education Research Using Web-Based Assessment Systems
