Skip to content

Multilingual embedding-based automated classification of self-regulated-learning think-aloud utterances

Research Method · All research methods
Evidence · 1 claim (1 for) · 1 study (1 associational), q2 · 0 of 1 report an effect size · 1 claim rests on one study

Description

The pipeline vectorizes English and German think-aloud utterances with OpenAI's text-embedding-3-small, "a pre- trained sentence embedding model that converts textual input into a high-dimensional vector with a length of 1,536". A single-hidden-layer ReLU neural network (28 intermediate units, sigmoid output, Adam optimizer, learning rate 0.01, 30 epochs, batch size 10) performs binary classification of each SRL category, evaluated with student-level 5-fold cross-validation using AUC. All analysis code is released in a public GitHub repository.

Accounts

  • Multilingual SRL detection pipeline using OpenAI text-embedding-3-small embeddings with a single-hidden-layer neural network classifier: The pipeline vectorizes English and German think-aloud utterances with OpenAI's text-embedding-3-small, "a pre- trained sentence embedding model that converts textual input into a high-dimensional vector with a length of 1,536". A single-hidden-layer ReLU neural network (28 intermediate units, sigmoid output, Adam optimizer, learning rate 0.01, 30 epochs, batch size 10) performs binary classification of each SRL category, evaluated with student-level 5-fold cross-validation using AUC. All analysis code is released in a public GitHub repository. (Conrad Borchers et al. (2025))

Claims

-

Key Sources

  • Conrad Borchers, Jiayi Zhang, Hendrik Fleischer, Sascha Schanze, Vincent Aleven, Ryan S. Baker. (2025). Large Language Models Generalize SRL Prediction to New Languages Within But Not Between Domains. Journal of Educational Data Mining, Volume 17, No 2. https://github.com/pcla-code/EDM24_SRL-detectors-for-think-aloud
Page metadata
Field Value
type research-method
id multilingual-embedding-based-automated-classification-of-self-regulated-learning-think-alo
title Multilingual embedding-based automated classification of self-regulated-learning think-aloud utterances
description The pipeline vectorizes English and German think-aloud utterances with OpenAI's text-embedding-3-small, "a pre- trained sentence embedding model that converts textual input into a high-dimensional vector with a length of 1,536".
status review
moved_from element/text-embedding-3-small-srl-detection-pipeline
generated by process:sweep-kinds on 2026-10-10
trust unverified — no one has confirmed this page's content
sources 1 — conrad-borchers-2025

Frontmatter is the page's machine-readable record — see the Schema & Guide.