Frédéric Kaplan holds the Digital Humanities Chair at Ecole Polytechnique Fédérale de Lausanne (EPFL) and directs the EPFL Digital Humanities Lab where he leads projects combining archive digitisation, information modelling and museographic design. He and his team are currently working on the Venice Time Machine, a project in collaboration with Ca’Foscari University to model the evolution and history of Venice over a thousand-year period. Frédéric Kaplan is a graduate of the Ecole Nationale Supérieure des Télécommunications in Paris and holds a PhD in Artificial Intelligence. Previously, he worked as a researcher at the Sony Computer Science Laboratory.
“Thanks to technology, there need be no split between the continent that is the present and the continent of the past. Our role is to digitise centuries-old data and import it into our present-day information systems.”
Digital humanities researcher Frédéric Kaplan is leading the incredible Venice Time Machine project, using archive digitization to model the construction and history of Venice over a thousand-year period.
His job title sure sounds fantastic! Frédéric Kaplan is indeed a “researcher in digital humanities.” Holder of the Digital Humanities Chair at the Ecole Polytechnique Fédérale de Lausanne (EPFL) and with a PhD in Artificial Intelligence, Kaplan set himself the task of recreating “1,000 years in the (virtual) life of a city.” He and his team are currently collaborating with Venice’s Ca’Foscari University on the Venice Time Machine, a project to model the city’s evolution and history. To achieve this, the scientists embarked on digitizing all the archives held by the city of the Doges, from official documents to masterworks of art, including cadastral maps, correspondence, engravings, and more… Their objective: to cross-check the data on computers and eventually extract its meaning. Kaplan came to share some of the findings from this extraordinary venture with the 9th FHH Forum.
Frédéric Kaplan’s very singular job is “to develop temporal depth.” Formerly on the staff at Sony—where he worked on robotics projects—Kaplan has specialized in archive digitization, information modeling and museography at EPFL. As head of the Digital Humanities Lab, he is working with his research team to read through all the archives in Venice. But this is no ordinary read. “The building that houses the city’s historical documents numbers 30 rooms with 80 km of shelving,” he explains. “There’s all sorts of material: wills, ambassadors’ letters, birth certificates, harbor timetables… And as for cadastral maps, there’s a whole forest of them! This ocean of data is the cornerstone of the project. Our challenge is to digitize these texts and images, and then make them readable by computers.” The researchers have had to develop ad hoc scanners that enable them to process large-format books, or geographical drawings, at a steady pace. Using this technology, they hope to complete the full digitization process within two years. Kaplan is optimistic that in the future, an imaging technique called tomography could be adapted to archives so researchers can scan a book without even opening it.
A revolutionary technique
“Once the documents are digitized, we need to be able to read the data and make it searchable,” he continued. To achieve that, he developed a technique that’s simple, yet revolutionary: documents are broken into segments of words or groups of words, which constitute as many images. These are then compared in order for the team to identify analogies. Finally, each defined image is given a meaning validated by the researchers. For instance, once the image of the word “Nobis” written in calligraphy is recognized as signifying Nobis, the thousands of “signs” that are interpreted in the same way are translated in one fell swoop. Indexing of old books by keywords becomes possible.
To process cadastral maps, the parcels are geometricalized using algorithms. For engravings, paintings and photographs, optical cells are capable of setting parameters for the locations so they can be categorized by their similarities. Once each individual building, street, square or other place is defined, the whole city finds itself virtually classified. “It’s already possible, even at this stage, to reconstitute someone’s social network, their circle of friends and professional acquaintances,” Kaplan enthused. “But it gets most interesting when you mix it all up!” The cross-checking of dates, places, names, images, and genealogies suddenly makes it possible to model the entire city at a given moment in time, and over a given period. “It’s like a colossal crossword which, with the help of computers, we’re on our way to solving!”
By interlinking its data, the Venice Time Machine has managed to reconstitute 1,000 years of the city’s history. Computer-generated images make it possible to stroll through the streets and find the signs of stores that really did exist there, or to get an accelerated view of how the different districts were built up. “Naturally, we factor in the vagaries of history, like wars, fires and disease,” says Kaplan. Entirely open source, this extraordinary venture is eagerly awaiting to be taken up in other locations, such as Paris, Budapest, Amsterdam or Jerusalem—to give a little more meaning to digital humanities research.
Click on the video to view the speaker’s interview at the Forum.
See which experts and personalities are invited to the Forum.