Synthetic Data Quality Evaluation Form
Use this form to comprehensively assess the quality of a synthetic dataset. Please provide objective, structured feedback using the evaluation prompts below.
Dataset Name or Reference
*
Intended Use Case Clarity
*
Very clear and well-defined
Mostly clear
Somewhat clear
Unclear or ambiguous
How realistic is the synthetic data compared to real-world data?
*
1
2
3
4
5
Diversity and Representation (e.g., does the data cover all required categories and variations?)
*
Very Limited
1
2
3
4
Highly Diverse
5
1 is Very Limited, 5 is Highly Diverse
Privacy Protection (e.g., risk of re-identification or data leakage)
*
No privacy concerns
Low risk
Moderate risk
High risk
Utility for Intended Analysis or Model Training
*
Not Useful
1
2
3
4
Extremely Useful
5
1 is Not Useful, 5 is Extremely Useful
Labeling Accuracy (if applicable)
Perfectly accurate
Minor errors
Frequent errors
Not labeled or mostly incorrect
Presence of Artifacts or Anomalies
*
None observed
Minor artifacts
Noticeable artifacts
Severe or frequent artifacts
Overall Quality Assessment
*
1
2
3
4
5
Additional Comments or Suggestions
Submit Evaluation
Should be Empty: