A Platform for Managing Lossless Data for Machine Learning and Sharing Experimental Data

The study of data-oriented understanding of materials data, as represented by structures, properties, mechanisms, and protocols, is known as materials informatics1 Automated material design, extensive data analysis, and expedited tests using robots have all been employed in the field to promote the development of materials for energy- and environment-related applications.

The introduction of materials informatics, a strongly data-reliant subject focusing on materials data, including synthesis techniques, characteristics, processes, and structures, was a game-changing advance in this area. Artificial intelligence (AI), which permits extensive, automated data analysis, material design, and experimentation that can aid in identifying valuable materials, has greatly benefited it.

Unfortunately, data loss frequently results from back-and-forth data sharing within the scientific community. This is because most material databases and research papers place a greater emphasis on structure-property interactions than on crucial details such as crucial experimental techniques.

To address these problems, a group of researchers created a platform for managing laboratory data that explains the connections between properties, structures, and experimental procedures. This electronic lab notebook represents observed events and associated environmental parameters as knowledge graphs.

Research, released in the journal npj Computational Materials on August 17, 2022, was based on the idea that knowledge graphs may accurately explain experimental data. The group used an AI-based method to automatically create tables from these knowledge graphs and publish them to a public repository. This procedure was added to ensure lossless data transmission and to give the scientific community a better understanding of the experimental setup.

The team employed this platform to investigate superionic conductivity in organic lithium (Li+)-ion electrolytes to show the platform’s usefulness. In the computerized laboratory notebook, they entered daily raw data from more than 500 successful and unsuccessful tests. The data conversion module then automatically converted the knowledge graph data into datasets that computers can learn from and examined the connection between experimental procedures and outcomes. An ideal room temperature ionic conductivity of 104–103 S/cm and a Li+ transference number of up to 0.8 were achieved thanks to the analysis, which identified the critical factors.

The new data platform makes it possible to efficiently record and store routine experimental events as graphs, which are subsequently converted into data tables to make room for additional AI-based research. Credit goes to Waseda University’s Kan Hatakeyama-Sato.

Real-time applications are a platform that “will be able to contribute to the creation of safer and high-capacity batteries with increased performance.”

This study ensures that all information, including experimental results and raw measurement data, is made publically accessible, in addition to offering a solid foundation for data-driven research.

The researcher explains its long-term repercussions: “Researchers from all around the world may discover innovative functional materials more quickly if they shared raw experimental data. This strategy can hasten the development of energy-related gadgets, such as solar cells and next-generation batteries.”

This Article is written as a research summary article by Marktechpost Staff based on the research paper 'Exploration of organic superionic glassy conductors by process and materials informatics with lossless graph database'. All Credit For This Research Goes To Researchers on This Project. Check out the paper and reference article.

Please Don't Forget To Join Our ML Subreddit