Quality Prediction of Open Educational Resources A Metadata-based Approach

TL;DR

This paper presents a metadata-based model for predicting the quality of Open Educational Resources, achieving high accuracy and aiding in automatic quality assessment to improve search and recommendation systems.

Contribution

It introduces a novel metadata scoring and prediction model for OER quality assessment, leveraging large-scale metadata analysis.

Findings

Achieved an F1 score of 94.6% in OER quality prediction.

Demonstrated the effectiveness of metadata analysis for automatic quality control.

Provided insights into metadata quality's impact on OER evaluation.

Abstract

In the recent decade, online learning environments have accumulated millions of Open Educational Resources (OERs). However, for learners, finding relevant and high quality OERs is a complicated and time-consuming activity. Furthermore, metadata play a key role in offering high quality services such as recommendation and search. Metadata can also be used for automatic OER quality control as, in the light of the continuously increasing number of OERs, manual quality control is getting more and more difficult. In this work, we collected the metadata of 8,887 OERs to perform an exploratory data analysis to observe the effect of quality control on metadata quality. Subsequently, we propose an OER metadata scoring model, and build a metadata-based prediction model to anticipate the quality of OERs. Based on our data and model, we were able to detect high-quality OERs with the F1 score of 94.6%.