Basic vocabulary for artificial intelligence

Chat botAnd deepAnd big data, etc. How can these concepts be expressed in French?

To answer this question, DataFranca has published a glossary 101 Words of Artificial Intelligence: A Short Guide to Basic Data Science and AI Vocabulary. In addition to translating key terms in the field, this book generalizes concepts for easier understanding.

This project was led, for the science component, by Patrick Drouin, Professor of Translation in the College of Arts and Sciences at the University of Montreal, and Claude Colomb, a Physics and Computer Science graduate from UdeM.

Fits the vocabulary of artificial intelligence in French

Realizing that terms related to artificial intelligence, machine learning, and deep learning were scarce at the French nonprofit DataFranca’s wiki has collected more than 5,500 terms related to these fields of activity. The worldwide francophone community of many scholars, terminologists and translators has written thousands of encyclopedic files that have been submitted for peer review. Contributors to this project, supported by the Fonds de recherche du Québec and the Office québécois de la langue française, include Joshua Bengio, from Mila – the Quebec Institute for Artificial Intelligence – and Hugo Larochelle, from Google.

Lexicon 101 words of artificial intelligence represents the heart of this work. It consists of 101 entries in French and English, presenting the most frequently used terms such as deep learning, decision tree or layer of neurons.

Claude Colomb and Patrick Drouin worked in pairs on this lexicon: the first devoted himself to the technical side, and the second took charge of the generalization and linguistic side. “Claude discussed the concepts, and then I modified them into understandable explanations,” says Patrick Drouin.

required working speed

“It’s important, quickness of reaction, flexibility, because if we don’t go fast enough, the English idiom will prevail, and then it will be very difficult to go back and undo the habits,” says Claude Colombe.

We think of many English computer terms used in French, for example data set. Today’s linguists have difficulty implementing the “dataset” equivalent.

Create new terms in French

Some words have been translated almost verbatim as chatbot, provided by “chatbot”, or deepfake Translated as “deepfake”.

In some cases, metaphors, such as those of expression, were retained mining data, which translates to “data mining”.

In other cases, the researchers chose to make the terminology clearer: the name Transformation It became “autonomic neural network” or “self-awakening network”, which refers to “a high-performance, serial-to-sequence deep neural network architecture. It uses the attention mechanism, more precisely the self-attention, to replace both iterations and convolutions. Moreover, it comes Using interest in machine translation from the groundbreaking work of MILA”, we read in the glossary.

Well developed vocabulary

This vocabulary is subject to change. big data Until now, it was translated as “big data”. Now, when we know that prefix Mega It comes from the Greek meaning “million” and that today we are talking about information that is no longer stored in megabytes [10⁸]but exceeded a terabyte [10⁹]a petabyte [10¹⁵]it no longer makes sense to talk about it big data. So we should talk instead big datasays Patrick Drouin.