Kitware’s Data and Analytics team has worked closely with the Toyota Research Institute (TRI) to develop MaterialNet, a new open source web application. TRI’s mission is to develop automated driving, robotics, and other human amplification technology for Toyota, and one of their main areas of focus is discovering advanced materials. Materials research has a reputation for being time-consuming, with advancements often taking decades. Scientists at TRI are determined to overcome this limitation, and they sought Kitware’s help to develop MaterialNet software to enable materials scientists to visually uncover important relationships among technological materials, contributing to the acceleration of the discovery process.
How MaterialNet Works
Materials science research is predominantly data-driven which has resulted in the prevalence of materials databases. While these databases contain the information researchers need, their volume and complexity make it difficult for researchers to efficiently navigate through the datasets. MaterialNet allows users to visualize aggregates from these databases. This provides researchers with intuitive layouts that help them understand which materials are related and how.
Seeing MaterialNet in Action
MaterialNet uses interactive “maps” of the materials space exposed in large material databases to help researchers to easily focus on a particular research task and visualize the properties of specific materials. As you can see in the figure below, materials are represented with a graph structure that has nodes standing in for materials, and links between them encoding the appropriate relationships of interest (i.e. a chemical similarity or co-occurrence in text). In addition to exploring the topology of material networks, MaterialNet also displays information about each material, highlighting its immediate neighborhood within the graph, and offers several auxiliary tools to help drill down into the details contained within the dataset.
The Future of MaterialNet
Kitware plans to explore several directions to increase the reach and value of MaterialNet. Currently, MaterialNet has been architected to support different types of data, including the materials stability network, a text co-occurrence network, and a materials similarity network. But the tool can be easily extended to display any other type of material network as well. We would like to develop a more powerful and flexible search mode that will extend the researcher’s ability to find materials with very specific properties or ranges of properties. We are also looking into adding new visualization modes and integrating with existing modes and tools so researchers have the ability to view the data in different ways.
MaterialNet is an open source web application, so we encourage you to access it for free here: http://maps.matr.io/ If you think your project may require a more customized solution, please contact us at firstname.lastname@example.org.
If you would like to read more about MaterialNet, Kitware’s Roni Choudhury co-authored a paper that was published in The Journal of Open Source Software called “MaterialNet: A web-based graph explorer for materials science data.”