Using bayesian networks in computerized adaptive tests

• E. Millán ^[1] ; M. Trella ^[1] ; J.L Pérez de la Cruz ^[1] ; R. Conejo ^[1]
  1. [1] Departamento de Lenguajes y Ciencias de la Computación Facultad de Informática, Málaga,
• Localización: Computers and education in the 21st century / Manuel Ortega Cantero (ed. lit.), José Bravo Rodríguez (ed. lit.), 2000, ISBN 1-280-20019-7, págs. 217-228
• Idioma: inglés
• Enlaces
  • Texto Completo Libro
• Resumen

  • In this paper we propose the use of Bayesian Networks as a theoreticalframework for Computerized Adaptive Tests. To this end, we develop theBayesian Network that supports the Adaptive Testing Algorithm, that is, wedefine what variables should be taken into account, what kind of relationshipsshould be established among them, and what are the required parameters. Asparameter specification is one of the most difficult problems when usingBayesian Networks, we suggest the use of several simplifications. By usingsuch simplifications, the required conditional probabilities can be obtained in arelatively simple way