Data Label Quality Issue

A Data Label Quality Issue is a data quality issue where data labels do not correctly represent data instances.

• AKA: Label Quality Problem, Annotation Quality Issue.
• Context:
  • It can typically affect Data Label Quality Training Data.
  • It can typically impact Data Label Quality Model Performance.
  • It can typically arise from Data Label Quality Human Annotation.
  • It can typically require Data Label Quality Correction.
  • ...
  • It can often introduce Data Label Quality Noise.
  • It can often reduce Data Label Quality Model Accuracy.
  • ...
  • It can range from being a Minor Data Label Quality Issue to being a Critical Data Label Quality Issue, depending on its data label quality severity.
  • It can range from being a Random Data Label Quality Issue to being a Systematic Data Label Quality Issue, depending on its data label quality pattern.
  • It can range from being a Local Data Label Quality Issue to being a Widespread Data Label Quality Issue, depending on its data label quality extent.
  • It can range from being a Detectable Data Label Quality Issue to being a Hidden Data Label Quality Issue, depending on its data label quality visibility.
  • ...
• Example(s):
  • Classification Label Quality Issues.
  • Boundary Label Quality Issues.
  • Missing Label Quality Issues.
  • Inconsistent Label Quality Issues.
  • ...
• Counter-Example(s):
  • High-Quality Label, which accurately represents data instances.
  • Label Enhancement, which improves label quality.
• See: Data Quality Issue, Label Noise, Data Annotation, Training Data, Quality Control, Performance Degradation Issue, Information Extraction Strategy.

---