Headless Materials Research Management Systems

A headless content management system (CMS) is one that is API-first, decoupled from any particular presentation layer on the front-end. Such a service can support multiple channels for working with and presenting content, given the type of user and their context. Strapi is one example of an open-source headless CMS.

Consider a so-called materials research management system (MRMS). I am specializing this to materials science, rather than say a generalized “scientific research” management system. Furthermore, I am specializing this to research: not procurement, not product development, not manufacturing, not distribution, not sales.

However, I am also not drilling further in to, say, “research data”. I am including all activities that pertain to research, which may include reference to, for example, procurement activity. Other concerns of research activity include domain knowledge/ontologies, hypotheses, experiments, analyses, reports, and presentations. Finally, I reuse the “management system” part from prior art like content management systems (CMSes) and relational database management systems (RDBMSes) that provide useful interfaces for governance.

So, what are some examples of such systems? Citrine Informatics advertises an online platform that uses an open-source data model they developed called Graphical Expression of Materials Data (GEMD). The GEMD model distinguishes materials, processes, and measurements, and distinguishes between intent (specification) and realization (e.g. actual conditions observed for a run of an experiment). For example, a process of procurement can be said to produce a material, another process can prepare a sample of a material as an ingredient for a measurement run, etc. It is clear that the Citrine platform is intended as a system for materials research management.

Another example of a MRMS is the Materials Project’s contributions framework, MPContribs. At this time, MPContribs does not provide many affordances for explicit modeling of research assets and activities – unlike GEMD, the semantics of submissions are mostly implicit and open-ended. The emphasis of MPContribs is empowering users to submit generic reports and custom presentations of processed research results in a way that links them to existing Materials Project materials. MPContribs is a great example of a headless MRMS, as it serves both customized landing pages for each submitted project as well as inset “cards” on pre-existing material detail pages.

Finally, I’d like to point out Toyota Research Institute’s computational autonomy for materials discovery (CAMD) system, an open-source framework that explicitly models and manages research concerns like hypotheses, experiments, and analyses. One designs sequential learning campaigns to deploy software agents that evaluate candidate materials in some search space, given relevant seed data, and that iteratively dispatch hypotheses to an experimental facility (whether a physical laboratory or e.g. a density functional theory (DFT) program). CAMD is also a type of headless MRMS that tracks hypotheses and experiments as well as provenance regarding decision processes.

One reason I’m particularly interested in headless materials research management systems is that research activities are distributed, not centralized. Systems that accommodate a variety of contributions from a variety of stakeholders, particularly across not only space but time (employees move on, graduate students graduate, …), can flourish. The World Wide Web (WWW) does this for documents – HTTP is headless, and web browsers render a variety of markup for a variety of folks. I’m excited about what we can do for data and metadata with HTTP, i.e. the Semantic Web.