Friday, February 28, 2:30 p.m.,
N202 LC

Abstract

Automated item generation (AIG) refers to an emerging area, where technology and psychometrics integrate to construct new test items in a time- and cost-effective manner (Gierl & Haladyna, 2012). The introduction of AIG to education has also allowed efficient item development in various large-scale assessments. However, AIG attempts for developing reading comprehension questions that accurately evaluate students’ inferential knowledge were often considered challenging, oftentimes, unsuccessful. This was largely due to the complex and nested analytic structure of the reading passages. This presentation will introduce a new hierarchical topic-modelling approach that can uncover hidden topics and sentiments within a given text. Then, the modelling capacity of the current system will be demonstrated by generating multiple-choice reading inference items automatically using the uncovered topical and sentimental structure of the text.

Gierl, M. J., & Haladyna, T. M. (Eds.). (2012). Automatic item generation: Theory and practice. Routledge.

If you have questions about this seminar, contact Professor Mark Davison, mld@umn.edu.