Bringing Library and Museum Resources Together: How Can Artificial Intelligence Help (based on the Ivan Tsvetaev’s Book Collection Project)?

A few years ago at the Pushkin State Museum of Fine Arts Research Library we began a long-term digitalization program of the most valuable and/or fragile parts of our book collection. For a start, we decided to divide this process into a number of smaller programs, each dedicated to a different collection. One of our first and obvious choices to digitize (and thus make visible and fully accessible) was the collection of our museum’s founder and first director Ivan Tsvetaev (1847-1913), that consists of more than 850 volumes and vividly highlights his scientific interests, academic ties and hobbies. While digitizing Tsvetaev’s books we discovered quite a large number of dedicatory inscriptions (nearly 250) — a very useful complementary source of information on scientific and civil circle of our founder. All the inscriptions were scanned, decoded and translated into Russian, where necessary. Each of the inscription images was also enriched with brief biographies of the inscriptions’ authors (donators). First of all (and that was the most obvious step) we joined our various digital resources into a common semantic space — a digital collection — using available technical tools that were in our disposal. By compiling descriptions of all units related to the collection in our catalog and linking them with our unified authority data system (such as subject and name indexes), we managed to create a very comfortable in use semantic fragment, that better describes Ivan Tvetaev’s personality and the history of our museum’s creation through the prism of his personal interests and professional contacts. Then we started contemplating the way of possible connection of this collection to other complementary museum resources, that are kept in other departments and described in other IT-systems. One of the possible approaches to bring all these uncoordinated sources together is applying artificial intelligence technology. We found this way intriguing and rather promising. Our paper is aimed to explain the possible way of data relationship to design a dialogue between the voice assistant and the user while searching through library, museum, and possibly external resources (based on the example of a certain digital collection).