At EPFL researchers are working together with colleagues from Italy on the construction of a time machine that will allow you to immerse yourself in the Venice of the past. The team, headed by Frédéric Kaplan, has already digitised two million documents and images for this purpose. Specially developed search engines are bringing this digital heritage to life.

Prof. Frédéric Kaplan, Head of the Venice Time Machine project (left) and his research employee Isabella di Lenardo. (Photo: Kellenberger Kaminski Photographie)

“Recently, I was standing on the Rialto Bridge in Venice in front of one of the tourist shops and couldn’t help thinking that in 1740, in this exact spot, Francesco Raspi’s bottega was selling clothes, just as his grandson of the same name did decades later”, explains Frédéric Kaplan. “Being able to walk around the city and feel a connection with the spirit, life and achievements of a person from the past makes quite an impression on you.” Soon, however, this is something that anyone may be able to experience.

Frédéric Kaplan is Professor of Digital Humanities at EPFL and head of the “Venice Time Machine” project. As part of this project, over 20 collaborators are digitising historical documents at the Venice city archives, at the Fondazione Giorgio Cini and the public libraries in Venice. The city archive alone covers a period of 1,000 years and fills 80 kilometres of shelves. In less than two years the team has already managed to scan two million documents, maps and images. For some of these it has used a specially developed scanner with a rotatable, two-metre-wide table that can be fed with several A3 documents simultaneously. This is the fastest machine of its kind in the world and can process up to 8,000 images a day. “We are dealing here with ‘dark matter’ – documents that hardly anyone has studied before us”, says the computer scientist.

Prof. Frédéric Kaplan (left) and his research employee Isabella di Lenardo. (Photo: Kellenberger Kaminski Photographie)

Search engines are making this digitised historical material accessible. A system known as “Replica” analyses images and finds pictures of St Mark’s Square, for example, even if these have been captured from different perspectives, or recognises scenes in different styles, but of the same genre. Frédéric Kaplan demonstrates how, in a matter of seconds, “Replica” finds all paintings containing a previously selected element – such as the infant at the feet of the Madonna – and regroups the paintings on the basis of this information. This is made possible by means of a technique known as deep learning, which has revolutionised data processing in recent years.

Linking historical sources

A second search engine known as “Linked Books” finds links between archived documents, books or articles. With the help of a robotic scanner that automatically turns the pages of books, the researchers have digitised 3,000 volumes about Venice. This means that all the bibliographic entries in which a document is cited or, conversely, all cited documents in a secondary source can be found. Once historians have tested these search engines, they will be made publicly accessible in 2018.

One of the most difficult tasks is the automatic recognition of old handwriting. As part of a European project known as “READ” the researchers have developed a technique that uses deep learning to recognise whole words. This means that documents are analysed and their key characteristics – information such as names and places – are linked graphically. These documents form the core of a social network to which further documents can be added almost without limit – a Facebook of the past that allows the lives of countless tradesmen, merchants and officials to be reconstructed.

“It’s all about bringing history into the digital world”, says Frédéric Kaplan.

“Our time machine will change research by facilitating and democratising access to historical sources”, says Kaplan. It is especially important that young people in particular realise there was also a world before Google. Nevertheless, there has also been some criticism, with certain people warning about the possibility of digital data being manipulated or expressing fears that finding out about our ancestors could have a negative impact on our own lives. “That’s the biggest challenge”, according to Kaplan, “namely, how our society deals with this development.” Similar projects are already being planned in Paris, Budapest and Jerusalem. In addition, a consortium made up of 15 other institutions in addition to EPFL is applying to the European Commission to be selected as the next flagship programme with a view to constructing a European time machine. “It’s all about bringing history into the digital world”, the inventor of the “Venice Time Machine” sums up.