Approach to intelligent adaptive testingan optimized fuzzy logic model

  1. Angel Neira 1
  2. Alfredo Alguero 1
  3. José A. L. Brugos 1
  4. Víctor García 2
  1. 1 Dept. Informática. University of Oviedo. Campus de Viesques, 33201 Gijón,
  2. 2 Dept. Informática. University of Oviedo. Campus de Viesques, Gijón
Book:
Computers and education in the 21st century
  1. Manuel Ortega (ed. lit.)
  2. José Bravo (ed. lit.)

Publisher: Kluwer Academic Publishers

ISBN: 0-306-47532-4 1-280-20019-7

Year of publication: 2000

Pages: 241-249

Type: Book chapter

Abstract

Considering the evaluation of acquired knowledge as a fundamental elementof Computerized Tutoring Systems, the classical Bayesian methods foradaptive testing present several deficiencies, such as arbitrary assignments orprior probabilities, non representativity, the effect of the context and the needto include all the concepts under evaluation. An alternative model of theadaptive test is proposed here based on content-balanced and impreciseinterpretations of the information level and prior distribution of each itemusing Gil's fuzzy relationships instead of Bayesian relationships.