Automated information extraction and classification of matrix-based questionnaire data

Abstract

Survey research is an important part of teaching quality evaluation systems in higher education. Although universities more and more often use software with electronic questionnaires for surveying, the results are not always satisfactory due to low response rates or high workload of data processing. In this paper an effective model of interviewing, based on compact matrix questionnaires is presented. Innovative methods for automated extraction, transformation and classification of survey responses are introduced. They were validated on real-life university questionnaire data showing very good performance, especially for closed questions.