Automatic summary of teacher’s error feedback based on a taxonomy

Abstract : This paper presents an algorithm that computes the most common error types reported by teachers on students’ lab reports in LabBook system. These common error types aim to improve students’ learning by helping teachers to take necessary corrective actions. Computing the most common error types is not obvious because of their taxonomic structure. Thus, using frequencies of error types leads to select the uppermost type as the most common one. For this reason, two other parameters are taken into account in addition to the type frequency: a type generality level and a number of type subtypes. To define a computing algorithm, the most common error types are formalised with three rules where each rule uses one parameter. The algorithm proposed is based on ranking function that respects the three rules. It assigns a score to an error type by multiplying its frequency with a weight function based on information content. The feature that provides common error types to teachers was implemented using semantic web technologies. The results of a qualitative study conducted with teachers showed that experienced teachers used and combined the algorithm rules to select the most common error types that can help them take corrective actions efficiently.
Complete list of metadatas
Contributor : Cyrille Desmoulins <>
Submitted on : Tuesday, March 5, 2019 - 10:46:04 AM
Last modification on : Friday, October 25, 2019 - 1:30:27 AM


  • HAL Id : hal-02057185, version 1



Hakim Mokeddem, Cyrille Desmoulins, Nadine Mandran, Rachid Chalal. Automatic summary of teacher’s error feedback based on a taxonomy. International Journal of Technology Enhanced Learning, 2018, 10 (1/2), pp.22-43. ⟨hal-02057185⟩



Record views