Publikation

Gender-Biased Adaptations in Educational Adaptive Hypermedia

Erica Melis, Carsten Ullrich

SEKI Technical Report SR-04-05 2004.

Abstrakt

Studies show that there is a statistically significant gender difference regarding computer usage. For instance, a majority of female users expect less success from an interaction with a computer and are more likely to blame themselves in case something goes wrong. As a result, women considerably less often use a computer as a tool than men do. Now, if women do not use a computer as much as men do, they will not profit as much from the advantages of computational systems as they potentially could. That's why this paper reviews work about the causes of this problem, investigates an extended model, and develops possible preventions and reactions to it. Appropriate adaptivity cannot, however, be based on mere gender information which represents a bias only and may imply cliches and discrimination. Adapting a system solely depending on the sex of the user could have the effect all users -- men and women -- will feel discriminated by facing a system that embodies a cliche. We think that an adaptive system with an appropriate user model can help to avoid a cliche-based treatment: the sex of a student is only used to initialize the user model and as soon as more detailed information about the aptitudes of the user is gained from his/her interaction, the initial values of the user model are refined. Moreover, the user should be in control about the behavior of the system at any time.

Projekte

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence