R&D insights at lightning speed: introducing the Map of Science

The Map of Science collects and organizes the world’s research literature, making it easy to pinpoint key trends, hotspots, and concepts in global science and technology.

What are the hottest topics in Chinese AI research? It's a simple question to ask––less simple to answer. You could start by Googling around, looking at recent news coverage, or browsing an academic database. Maybe you're lucky enough to know a subject matter expert who can steer you in the right direction. Still, it's hard to know if any of these approaches will give you much to go on (much less the full picture), or how you'll know if you picked the right place to start. And who has the time for open-ended research in the first place?

Enter the Map of Science.

Animation of the Map of Science in use.

The Map of Science collects and organizes the world’s research literature, revealing key trends, hotspots, and concepts in global science and technology. Building on cutting-edge work by CSET’s data team, the Map makes it easy to pinpoint important technological trends, as well as the countries, authors, institutions, and funders driving progress. The Map is powered by our Merged Academic Corpus, which contains detailed information on over 270 million scholarly articles from around the world.

In this post, we’ll show how the Map can help with that Chinese AI question. We’ll start by opening sciencemap.eto.tech to its main view, showing a visual mapping of the global literature alongside a filter pane. Each dot on the Map is a research cluster: a group of articles that cite each other frequently. That’s typically because they share other things in common, like topics or languages (for example).

The default Map of Science view, showing the cluster visualization (center), filters (left), and details for the cluster in focus (right).
The default Map of Science view, showing the cluster visualization (center), filters (left), and details for the cluster in focus (right).

We’ll apply some filters to quickly winnow down to the relevant research only. The Map covers all topics and technologies, but it does have some special features related to AI and similar topics. Let’s raise the threshold for AI relevance, which will select only clusters with a lot of AI research articles.

Animation: selecting various thresholds in the Map's filter pane.

Next, filter further by Chinese author affiliation. We’ll also limit the view to groupings of research that are growing especially fast.

Animation: selecting various thresholds in the Map's filter pane.

From hundreds of millions of articles down to a handful of clusters––not bad for a start...

Animation: view of the overall Map interface; selecting various thresholds in the Map's filter pane; a smaller group of filtered clusters is now visible.

To begin figuring out what these results mean, hover over the remaining dots for quick summaries, or switch to list view to skim multiple clusters at the same time.

Animation: hovering over clusters, then switching to the list view and scrolling down

This summary data is the tip of the iceberg - the Map includes detailed information on the research in each cluster, including its growth rate, citations, top subject and concepts, and leading countries. You can also browse the top articles, sources, authors, institutions, and funders in the cluster, and see how its research is being used by industry.

To access this information, click on any cluster that intrigues you. It looks like the research in this particular cluster is focused on hyperspectral image processing, with a large share of Chinese authors and rapid growth in the past few years. A lot of the authors work at the Chinese Academy of Sciences and a few other Chinese universities.

Animation: scrolling through the detail view for a specific cluster and highlighting various metadata fields.

In just a few minutes, we’ve moved from the broad, even overwhelming question of “what’s going on in Chinese AI?” to detailed trends and indicators for further exploration. That’s the Map’s goal: to get you to the best possible “starting points” for further analysis and action.

To start using the Map of Science, visit sciencemap.eto.tech, or read the documentation for a thorough introduction to the tool and its features. In future blog posts, we’ll show off more of the Map’s capabilities and explore the unique datasets that power it.