Computational Reproducibility Validation in Cloud Big Data Pipelines Through Provenance and Machine Learning

Key Points

• The research aims to create and validate a scoring framework for assessing computational reproducibility in cloud environments.
• Developed a three-layer reproducibility scoring framework.
• Utilized provenance metadata to track data lineage.
• Applied machine learning techniques to evaluate reproducibility across cloud platforms.
• The scoring framework effectively assesses reproducibility across varied cloud environments.
• Validation confirms that the framework can reliably reproduce experimental results.