NOMAD: The FAIR Concept for Big-Data-Driven Materials Science

Abstract: Data is a crucial raw material of this century, and the amount of data that has been created in materials science in recent years and is being created every new day is immense. Without a proper infrastructure that allows for collecting and sharing data (including the original data), the envisioned success of materials science and, in particular, Big-Data driven materials science will be hampered. For the field of computational materials science, the NOMAD (Novel Materials Discovery) Center of Excellence (CoE) has changed the scientific culture towards a comprehensive and FAIR data sharing, opening new avenues for mining Big-Data of materials science. Novel data-analytics concepts and tools turn data into knowledge and help the prediction of new materials or the identification of new properties of already known materials.

• The paper demonstrates that applying FAIR principles transforms data management, enabling effective secondary analysis in materials science.
• It details NOMAD’s comprehensive infrastructure including a global repository, normalized archive, and user-friendly encyclopedia that supports 40 electronic structure codes.
• The study underscores the importance of robust descriptors and advanced analytics, such as machine learning and compressed sensing, for uncovering novel material properties.

Overview of "NOMAD: The FAIR Concept for Big-Data-Driven Materials Science"

The paper "NOMAD: The FAIR Concept for Big-Data-Driven Materials Science" by Claudia Draxl and Matthias Scheffler, delineates the substantial technological shift introduced by the Novel Materials Discovery (NOMAD) Center of Excellence (CoE) in the field of computational materials science. This piece helps contextualize the pivotal role of the FAIR (Findable, Accessible, Interoperable, Re-purposable) principles in harnessing and democratizing data to accelerate scientific progress.

The study underlines the common problem in materials science: the deluge of data generated is often underutilized due to inadequate practices in data sharing and management. The NOMAD CoE addresses this juncture by transforming data into a form that is interoperable, allowing it to be employed for secondary analysis beyond its original intent, notably advancing the prediction and identification of novel materials properties using Big-Data analytics.

Key Contributions

The authors introduce the concept of the fourth paradigm in scientific research, which emphasizes data-driven discovery. This paradigm shift acknowledges that certain characteristics of materials are arduous, if not impossible, to capture entirely within a mathematical formalism but instead could be discernible through pattern recognition in large data sets.

NOMAD is positioned at this frontier by establishing a comprehensive data infrastructure that encompasses:

• NOMAD Repository: A massive, global repository of raw data from computational calculations, supporting approximately 40 different electronic structure codes and ensuring accessibility and long-term data preservation.
• NOMAD Archive: Provides a normalized format for data from diverse sources, facilitating interoperability and compatibility.
• NOMAD Encyclopedia: A user-friendly platform offering visual and detailed access to the computational materials data.
• Visualization and Analytics Tools: Novel visualization techniques, including VR, and advanced analytics tools, like machine learning and compressed sensing, to discover and map novel material properties and phenomena.

Analytical Insights

The authors put forth that the evolution and utilization of data-mining tools offered by NOMAD are contingent on data quality and comprehensiveness. This paper evidences that even with substantial data sets, the exploration of possible materials remains sparse, highlighting the importance of robust descriptors for effective data mining. The study demonstrates that appropriate descriptors—identified through methods such as compressed sensing—are crucial for uncovering significant insights into the properties of materials, such as those observed in two-dimensional quantum spin Hall insulators.

Implications and Future Directions

The presented research has broad implications for both industrial and scientific endeavors, from facilitating the design of new materials and enhancing existing products to propelling theoretical advancements in condensed matter physics. The platform's ability to handle vast and intricate data volumes suggests promising avenues for future experimental data integration.

The paper also alludes to the challenges of scaling these methodologies to experimental data paradigms, emphasizing the necessity of metadata enrichment and the harmonization of data quality across experimental origins. Meeting these challenges will necessitate sustained efforts in developing novel HPC solutions and methodologies.

In closing, the NOMAD initiative sets a precedent for open and efficient data use in scientific disciplines, promoting a cultural shift towards open science while simultaneously tackling the complexities presented by Big Data in materials science. These efforts not only advance the field but also bridge the gaps between computational predictions and practical applications, suggesting a profound potential for NOMAD's methodologies to inform future scientific inquiries.