Orion, built by Mozilla Fellow Kostas Stathoulopoulos, is an open-source tool that uses machine learning and interactive data visualizations to spot gaps and opportunities in scientific research

Timely and open access to novel outputs is key to scientific research. It allows scientists to reproduce, test, and build on one another’s work — and ultimately unlock progress.

The most recent example of this is the research into COVID-19. Much of the work was published in open access journals, swiftly reviewed and ultimately improving our understanding of how to slow the spread and treat the disease. Although this rapid increase in scientific publications is evident in other domains too, we might not be reaping the benefits. The tools to parse and combine this newly created knowledge have roughly remained the same for years.

Today, Mozilla Fellow Kostas Stathoulopoulos is launching Orion — an open-source tool to illuminate the science behind the science and accelerate knowledge discovery in the life sciences. Orion enables users to monitor progress in science, visually explore the scientific landscape, and search for relevant publications.

A screenshot from Orion

Kostas is a London, UK-based Mozilla Fellow working at the intersection of machine learning, social science, and policymaking, developing tools that help people make evidence-based decisions. Before joining Mozilla, Kostas was a Principal Data Science Researcher at Nesta.

Says Kostas: “How open is science when you are not able to parse the published work? When using an academic search engine, say for a literature review, we are not interested in what we fully understand. We want to discover the unknown unknowns, work that completes our knowledge mosaic. I believe we can accelerate knowledge discovery and production by developing better search engines and exploration mechanisms. Tools that combine machine learning with interactive interfaces and bring human intelligence and attention into the search process. As a Mozilla Open Science Fellow, I am trying to close the knowledge gap between what exists and what is known to us.”

Kostas worked on Orion with Zac Ioannidis and Lilia Villafuerte.

We can accelerate knowledge discovery and production by developing better search engines and exploration mechanisms.

Kostas Stathoulopoulos

How it works

Orion collects, enriches and analyses scientific publications in the life sciences from Microsoft Academic Graph.

Users can leverage Orion’s views to interact with the data. The Exploration view shows all of the academic publications in a three-dimensional visualization. Every particle is a paper and the distance between them signifies their semantic similarity; the closer two particles are, the more semantically similar. The Metrics view visualizes indicators of scientific progress and how they have changed over time for countries and thematic topics. The Search view enables the users to search for publications by submitting either a keyword or a longer query, for example, a sentence or a paragraph of a blog they read online.

Orion’s views are interconnected; users can identify an interesting country profile in the Metrics view, visualise its publications and filter them in the Exploration view and retrieve those papers and their metadata in the Search view.

Here’s an example of Orion at work: Sarah, an epidemiologist, wants to do a literature review on AI applications in COVID-19 research as part of her new project. At the start of it, Sarah reads a paper on arXiv about modelling the transmission of COVID-19 with machine learning and wants to find relevant research. She queries Orion using the paper’s abstract and retrieves semantically similar publications from bioRxiv along with their metadata. She switches to Orion’s particle view to explore the results and interactively search nearby clusters. She discovers other interesting machine learning applications in healthcare. Sarah looks up the machine learning topic on Orion’s metrics page and finds that the United Kingdom is a leader in it. Then, she uses Orion to retrieve the most cited machine learning papers published by UK-based researchers and contacts some of them in order to collaborate.

Learn more about Orion:

Orion: An open-source tool for the science of science, by Kostas Stathoulopoulos

A walkthrough of Orion’s backend, data and design decisions, by Kostas Stathoulopoulos

Orion’s open source code, by Kostas Stathoulopoulos and Zac Ioannidis

More than ever, we need a movement to ensure the internet remains a force for good. Mozilla Fellows are web activists, open-source researchers and scientists, engineers, and technology policy experts who work on the front lines of that movement. Fellows develop new thinking on how to address emerging threats and challenges facing a healthy internet. Learn more at https://foundation.mozilla.org/fellowships/.