Skip to content

Eight-stage machine-learning pipeline for automatic scoring with stage-specific bias vulnerabilities

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 article describes a machine-learning pipeline for automatic short-answer scoring comprising eight stages: data collection, scoring/labeling, preprocessing, feature engineering, model selection, data split, model training, and model evaluation. For each stage it names a characteristic bias vulnerability, e.g. unrepresentative samples create "representation bias", human labeling creates "annotation bias", and unstratified data splits can induce "sampling bias". It is used to explain where unfairness can enter automatic scoring systems before it surfaces at model evaluation.

Accounts

  • Eight-stage machine-learning pipeline for automatic scoring with stage-specific bias vulnerabilities: The article describes a machine-learning pipeline for automatic short-answer scoring comprising eight stages: data collection, scoring/labeling, preprocessing, feature engineering, model selection, data split, model training, and model evaluation. For each stage it names a characteristic bias vulnerability, e.g. unrepresentative samples create "representation bias", human labeling creates "annotation bias", and unstratified data splits can induce "sampling bias". It is used to explain where unfairness can enter automatic scoring systems before it surfaces at model evaluation. (Nico Andersen et al. (2025))

Claims

-

Key Sources

  • Nico Andersen, Julia Mang, Frank Goldhammer, Fabian Zehner. (2025). Algorithmic Fairness in Automatic Short Answer Scoring. International Journal of Artificial Intelligence in Education. https://doi.org/10.1007/s40593-025-00495-5
Page metadata
Field Value
type research-method
id eight-stage-machine-learning-pipeline-for-automatic-scoring-with-stage-specific-bias-vulne
title Eight-stage machine-learning pipeline for automatic scoring with stage-specific bias vulnerabilities
description The article describes a machine-learning pipeline for automatic short-answer scoring comprising eight stages: data collection, scoring/labeling, preprocessing, feature engineering, model selection, data split, model training, and model evaluation.
status review
generated by process:settle-candidates on 2026-10-27
trust unverified — no one has confirmed this page's content

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