On Wikipedia, barely one page out of six concerns a female personality… To remedy this inequality, a French student has developed an algorithm capable of automatically generating biographies of women, still too absent from the largest encyclopedia in the world.
Dn the context of her doctoral project, Angela Fan, a computer science student at the University of Lorraine, started from an observation: today, only 19% of biographies in French published on Wikipedia concern women. “There are still many women whose stories and achievements do not reach future generations,” she explains. in a press release of the Facebook Artificial Intelligence Research (FAIR) Meta.
It is precisely to reverse this trend that the student, in partnership with FAIR in Paris and the National Institute for Research in Digital Sciences and Technologies (INRIA), has developed a super-intelligent algorithm. Its goal : compile information found on the internet to draw sketches of Wikipedia-style biographies, then send them automatically to the online encyclopedia so that they can be validated and enriched by its editors. Among the content already online, this page dedicated to the American zoologist Libbie Henrietta Hyman.
15% of female editors at Wikipedia
But how can you be sure of the data delivered by the software? It’s all about programming, Angela Fan will answer you: the AI, particularly structured, has “trained” beforehand on no less than 677,000 Wikipedia biographies… After having integrated the way of writing and learned to recognize reliable information, she goes fishing for information: date and place of birth, student life, professional background… The algorithm will classify this information, generate a short text as well only references and then make them known to the contributors of the site.
But the problem is not only with the contents of the online encyclopedia. Only 15% of the site’s editors define themselves as women. “The gender bias, in terms of the number of female contributors, will have a direct impact on the topics that are covered on Wikipedia. So-called feminine subjects are treated less, and subjects relating to LGBTIQ, feminism, even less! And then, there is also a bias with regard to the representation of racialized people,” precise to France Culture the economist Natacha Rault, at the origin of the project The pagelesslaunched to combat the invisibilization of women and feminist content on Wikipedia.
More intimate details in women’s biographies
Other sexist biases must be taken into account, according to the computer science student: “when women are represented, their biographies are more likely to include details of their personal or intimate life. In 2015, the word “divorced” appeared four times more often in the biographies of women than in those of men.
Similarly, Angela Fan aims to fighting social prejudice with artificial intelligence, to increase the visibility and improve the quality of biographies of transgender and non-binary personalities. “Ultimately, we hope that our techniques can be used as a working basis for human Wikipedia authors,” concludes the researcher.
> #MeTooBots, artificial intelligence to detect workplace violence
> Twitter: a competition highlights the sexist and racist biases of the platform
We would like to say thanks to the writer of this article for this awesome content
How artificial intelligence will make Wikipedia more inclusive
Check out our social media profiles and other pages that are related to them.https://www.ai-magazine.com/related-pages/