Towards a sustainable high-quality materials data ecosystem: frameworks and strategies

Area of Science:

• Materials Science
• Data Science
• Artificial Intelligence

Background:

• High-quality data is essential for advancing artificial intelligence (AI)-driven materials discovery and design.
• Current materials science data systems often produce low-quality, heterogeneous, and inconsistent data.
• This data fragmentation necessitates significant resources for consolidation and governance.

Purpose of the Study:

• To propose a comprehensive quality-controlled framework for materials data.
• To optimize data quality across the entire research lifecycle: acquisition, management, and utilization.
• To foster a sustainable and robust materials data ecosystem for AI applications.

Main Methods:

• Developing a framework with stage-specific requirements for data quality.
• Implementing an integrated computational/experimental 'data factory' for data acquisition.
• Utilizing an automated toolkit for data quality assessment.

Main Results:

• Concurrent optimization and enhancement of materials data quality throughout the R&D process.
• Establishment of five strategic initiatives to operationalize the quality-controlled framework.
• Embedding quality-governed production directly into scientific inquiry.

Conclusions:

• A systematic approach to data quality management is vital for AI in materials science.
• The proposed framework addresses data deficiencies from acquisition to utilization.
• This initiative promotes reliable data infrastructure for accelerated materials discovery and design.