| About this Abstract |
| Meeting |
MS&T24: Materials Science & Technology
|
| Symposium
|
Materials Informatics for Images and Multi-dimensional Datasets
|
| Presentation Title |
Advancing AI-Driven Analysis of Synchrotron Data via FAIR Practices, Ontology and Knowledge Graphs
|
| Author(s) |
Mohommad Redad Mehdi, Finley Holt, Weiqi Yue, Alexander C.H. Bradley, Balashanmuga Priyan Rajamohan, Erika I. Barcelos, Daniel Savage, Hemant Sharma, Matthew A. Willard, Frank Ernst, Pawan K. Tripathi, Roger H. French |
| On-Site Speaker (Planned) |
Mohommad Redad Mehdi |
| Abstract Scope |
In this work, we define ontologies which is a formal representation of key concepts, properties, and relationships in a particular domain, providing a shared vocabulary for both humans and AI models for the large scale synchrotron high energy X-ray diffraction (HEXRD) datasets. Leveraging this foundation, we integrate knowledge graphs to establish the connection between these concepts and actual synchrotron data. This approach results in structured knowledge representation conforming to FAIR (Findable, Accessible, Interoperable, Reusable) data standards. These ontologies are used as a precursor for knowledge graphs that are stored in our Janusgraph database. The resulting knowledge graph allows us to execute complex graph computations on the dataset, perform ML techniques like link prediction and graph learning, and even accelerate data-driven discoveries. The improved accessibility and contextual information of data will empower AI-driven analysis of massive-scale synchrotron data. |