Machine learning interatomic potential models organized in knowledge graph
Automatically Building and Updating a Knowledge Graph of MLIP Models
Artificial Intelligence
Summary
Materials scientists use machine learning models to predict how materials behave, but keeping track of all these models is hard because the field changes quickly. The authors created a computer process that automatically reads texts about these models and builds a structured network, called a knowledge graph, linking information about each model. Their method checks for mistakes and keeps the information consistent, making it easy to add new models over time. They demonstrated this by organizing models from a popular materials prediction benchmark.
What this means in practice
- •For materials data engineers: Automatically update and maintain structured data about machine learning models for materials property prediction.
- •For software developers for ai research tools: Build tools that query detailed, standardized information about interatomic potential models to assist researchers.
Authors
Alexis Beer, Liudmyla Klochko, Mathieu d'Aquin
Abstract
Complementing the many efforts in providing semantic representations of concepts, notions, and entities in materials science, we report and illustrate a process by which we can automatically build a knowledge graph of the fast evolving field of machine learning applied to the prediction of material properties, focusing on MLIP (Machine Learning Interatomic Potential). This LLM-based process relies on multiple steps, from information extraction in documents and articles to a validation loop using SHACL constraints to detect and correct errors. It is carried out on a model-by-model basis, focusing on the consistency of representation, therefore enabling an iterative construction where the addition of new models is facilitated. We illustrate the process by showing a few interesting aspects that can be queried from a knowledge graph built from the models listed in the Matbench Discovery leaderboard.