Organized, in collaboration with the ANR projects CultureIA and ScientIA, the workshop at the conference AI2S2 is structured around two main proposals: AI scientometric landscapes: four maps presenting different views on the scientific literature about artificial intelligence. Search me an AI: a simple and playful tool to brainstorm about AI-related keywords. Both are based on data made available by the scientometrics database OpenAlex. The tool allows you to play with this database and explore its records and keywords, while the maps showcase what is possible to do with these data.

Click here for a high-resolution PDF of the maps   (to download, right click and save the target file)
 

Maps description

OpenAlex Concepts

The map shows OpenAlex “concepts”, i.e. the subject categories used by the database to categorize its records (cf. docs.openalex.org/api-entities/concepts). Two concepts are connected if they are related or linked by a parent-child connection in the OpenAlex classification.

The nodes representing the concepts are spatialized through a force-vector layout and the contour lines in the map represent the density of nodes in different parts of the network, thus highlighting its clusters and structural holes. Only the labels of the most generic concepts are shown.

Artificial Intelligence Related Concepts

The map displays the “concepts” associated with the OpenAlex that contain the expression "artificial intelligence" in their title or abstract. Concepts are connected by their co-occurrence in the publications of the database with a strength proportional to their pointwise mutual information (PMI). The map displays all concepts with more than 50 and less than than 10.000 occurrences.

The nodes representing the concepts are spatialized through a force-vector layout and the contour lines in the map represent the density of nodes in different parts of the network, thus highlighting its clusters and structural holes. Only the labels of the concepts with more than 1.000 occurrences are shown.

AI Communities and Evolution

The circular tree-map in the figure represents the network of the fifteen disciplinary specialties that allow to partition the ensemble of the publications related to AI available in the Microsoft Academic Knowledge Graph and OpenAlex databases (i.e., whose titles or abstracts contain AI-related keywords). The partition is based on the co-occurrence of keywords in the publications.

The area charts show the annual evolution of the number of publications in each disciplinary specialty.

The two figures are extracted from the paper: Gargiulo, F., S. Fontaine, M. Dubois, and P. Tubaro. 2022. “A Meso-Scale Cartography of the AI Ecosystem,” no. 2004: 1–14.

Artificial Intelligence in Neurosciences

The map in the figure represents the knowledge landscape of neuroscience, captured in titles and abstracts provided by a set of neuroscientific publications extracted from the Microsoft Academic Knowledge Graph database. This map is divided into nine topical clusters delimited by dashed curves. Inside each cluster figures a density plot based on the local density of neuroscientific publications. The small white points covering this map are the publications containing AI-related keywords in their titles or abstracts.

The ridge plot shows the temporal evolution of the number of publications of each cluster (height of the line in the chart) and the share of AI-related publications (color of area under the curves).

This is currently work in progress

Search me an AI