Abstract

An important limitation of learning object repositories is that they frequently provide incomplete or imperfect information to describe the resources that they index. A form of dealing with this limitation is to categorize the learning objects in a taxonomy that allows main themes to be identified that cover each of these resources. In this paper, we will explore two techniques to categorize learning objects: language models and logistic regression models. Experiments show that the logistic regression method obtains results with high precision. © 2010 IEEE.