Question Answering in STACK Applying String Similarity

Achim Eichhorn; Andreas Helfrich-Schkarbanenko

doi:10.3991/ijet.v17i23.35893

Question Answering in STACK Applying String Similarity

Authors

Achim Eichhorn Esslingen University of Applied Sciences
Andreas Helfrich-Schkarbanenko Karlsruhe University of Applied Sciences

DOI:

https://doi.org/10.3991/ijet.v17i23.35893

Keywords:

online assessment, question answering, STACK, CAS, Damerau-Levenshtein edit distance, string metric, string similarity, e-learning, Natural Language Processing

Abstract

We present a method to evaluate fill-in-the-blank student answers in STACK using a string metric, which is not possible in the current version of STACK. To increase the quality of the evaluation, we use whitelist and blacklist instead of a single teacher answer. The performance of a STACK question equipped with a string metric is quantitatively demonstrated by evaluating its use in mathematics courses.

Downloads

Published

2022-12-08

How to Cite

Eichhorn, A., & Helfrich-Schkarbanenko, A. (2022). Question Answering in STACK Applying String Similarity. International Journal of Emerging Technologies in Learning (iJET), 17(23), pp. 56–63. https://doi.org/10.3991/ijet.v17i23.35893