Data platforms for Data-driven Materrials Science: ChemDX and MatDX

Abstract

Jungho Shin, Yea-Lee Lee, Gyoung S. Na, , Seunghun Jang, Jino Im and Hyunju Chang

Korea Research Institute of Chemical Technology (KRICT)

In the field of materials science, the data-driven approach has played a crucial role in the discovery of novel materials over the past decade. This approach has been significantly accelerated by the emergence of valuable data infrastructures that adhere to FAIR (findable, accessible, interoperable, and reusable) data principles. Notable examples of these infrastructures include NOMAD, OPTIMADE, Materials Project, AFLOW, and OQMD. To effectively utilize these data infrastructures, it is necessary to integrate and classify various types of metadata related to materials properties. This integration and classification process enables the research community to make more accurate predictions using artificial intelligence techniques.

MatDX (Materials Data eXplorer) has been developed with a specific focus on the integration and classification of materials-related metadata based on materials ontologies. These ontologies encompass classes and instances related to material name, composition, compound, structure, property, and applications. MatDX utilizes a data warehouse solution to facilitate integration by connecting multiple databases. The introduction of material tags, derived from the materials ontologies, allows for easy and quick searching and retrieval of detailed information on materials of interest. Additionally, MatDX offers an interactive "Analysis" functionality that visually represents statistically significant relationships within the materials data.

The primary goal of these services is to enable researchers to discover new materials with desired properties based on a vast amount of research data. MatDX encompasses three sub-categories: PubDX, which focuses on published data; ExpDX, which pertains to experimental data; and CalcDX, which deals with calculated data. Researchers can access MatDX through the following web address: http://materials.chemdx.org.

Biography

Data platforms for Data-driven Materrials Science: ChemDX and MatDX