Previous automated single-label classifiers of cognitive presence risked inheriting the subjectivity of manual categorisation from their training data preparation¶

Claim · All claims
Evidence · 1 study · design r2 · q1
Currency · newest study 2023 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q1 i? The article argues that because prior automated methods used single-label classifiers, they risked inheriting the subjectivity of the manual categorisation work done in the training data preparation stage. → Yuanyuan Hu 2023

Evidence¶

Yuanyuan Hu 2023¶

Yuanyuan Hu. (2023). Can Multi-Label Classifiers Help Identify Subjectivity? A Deep Learning Approach to Classifying Cognitive Presence in MOOCs. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-022-00310-5

q1 · i? · design · r2

Theoretical argument in the article's background and motivation, not a tested result. The authors reason that coders in prior studies resolved 'problematic' messages by negotiation to a single category, so "single-label classifiers, which had the risk to inherit the subjectivity" were trained on subjectively consensual labels.

"These automated methods built single-label classifiers, which had the risk to inherit the subjectivity from the manual categorisation work in the training data preparation stage."

Discussion¶

Related Claims¶

• Multi-label classifiers predicted MOOC messages as one label or two adjacent cognitive presence phases only, while expert coders produced non-adjacent and three-phase labels on almost 4% of messages — related