We show that the word embedding technique word2vec is mathematically equivalent to the gravity law of mobility, making it ideal for learning dense representations from migration data that can be distributed, re-used, and studied. By treating locations analogously to words and trajectories to sentences,we demonstrate the power of word2vec by applying it to the case of scientists’ migrations, for which it encodes information about culture, geography, and prestige at multiple layers of granularity. Our results lay a theoretical and methodological foundation for the application of neural embeddings to the study of migration.

In the 21st century, humans acquire and exchange knowledge primarily through language usage, with the emergence of information technology breaking down geographical barriers. A study using Wikipedia, a vast dataset, explores how knowledge diffusion occurs. It finds that when population groups have more socio-economic connections, like cultural and linguistic ties, their shared knowledge is more similar. Geographical proximity is no longer crucial for knowledge dissemination, and the dominant channel for information distribution in the 21st century appears to be online.