Ingram and Banerjee started their study by quantifying the fame, creativity, and social network of the artists in “Inventing Abstraction.” To determine each artist’s renown, they turned to Google’s database of historical texts in French and English (given that the artists were primarily living in France and the U.S.), and recorded the number of mentions each artist had between 1910 and 1925. They were looking at fame in terms of how well-known the artists were beyond their own social circles, Ingram noted, “and we’re essentially saying [that] how often you show up in the written word is an indicator of that.”
To examine the artists’ social networks, they relied on MoMA’s research, which was based on sources like biographies and artists’ letters to identify relationships. Ingram and Banerjee analyzed the artists’ social circles, which also involved data on each artist’s nationality, gender, age, and location, as well as the media they were using and the schools they attended. (They didn’t look into the artists’ exhibition history or the market for their work, though Banerjee’s future research may include such factors, Ingram said.)
In order to understand the creativity of the artists’ work, they employed two methods. First, they used machine learning to analyze and rate the creativity of thousands of artworks by the relevant artists; the computer program rated how unique works are in comparison to a set of representational artworks from the 19th century. They also asked four art historians to rate artworks by each artist for their creativity, based on factors like originality and innovation. (They found that the scores artists earned from machine learning and art historians were positively correlated.)