Semi-Automatic Generation of Metadata for Items in a Question Repository

Abstract

Question repositories are organized collections of assessment questions (items) that serve many purposes. Questions required vary widely along many dimensions such as cognitive level, difficulty level content and question type depending upon the context in which it is used. This paper proposes a software system that semi-automatically generates metadata for items in a question repository. The metadata in this paper corresponds to "cognitive level", "question type", "content" and "difficulty level". The developed system was tested for its usability and accuracy and the results are promising. The accuracy with respect to automatic generation of cognitive level tags was 78% and other tags were generated with more than 90% accuracy. Usability testing has shown that the system is user friendly and useful in multiple ways.