Data&Musée: developing data science for cultural institutions

Data science is a recent scientific discipline that extracts information, analyses and forecasts from large quantities of data. It is already widely used in many industrial fields, notably in energy and transport, but also in the healthcare sector.

However, this discipline has not yet become part of French cultural institutions' practices. While they collect their data individually, no initiative has yet been launched to aggregate and analyze all the data from museums and monuments across the country. And yet, once aggregated, this data would be of real interest to both institutions and visitors, particularly in terms of establishing analyses of the consumption of cultural goods in France, measuring the performance of institutions, and offering visitors relevant recommendations for visiting museums and monuments.

The Data&Musée project, for example, offers a territory for experimentation and reflection on how cultural institutions can analyze data, and how this data can support them in their development. The project leader, the Orpheo group, supplier of visitor assistance systems (audio-guides, multimedia guides, software, etc.) for cultural and tourist sites, is surrounded by companies specializing in data analysis, such as Tech4Team, Kernix, MyOrpheo, as well as researchers. The Centre des Monuments Nationaux, which brings together nearly 100 monuments, and Paris Musées, a group of 14 Paris museums, have positioned themselves as testing grounds for the project.

A unique, open data centralization platform

The Data&Musée project aims to bring museums into the data age, by pooling data from numerous cultural institutions on Teralab, the ITM and GENES Big Data platform. "This platform offers a neutral, secure and sovereign storage hosting space. The data will be hosted on the IMT Lille Douai site in France," explains Antoine Garnier, project manager at IMT. " Teralab is able to host sensitive data in accordance with current standards and regulations. It is already recognized as a reliable player."

Moreover, if the data is too sensitive, it can be anonymized. In this case, the project could call on the Lamane start-up, a product of the IMT Atlantique incubators, which is working on these technical issues.

In addition to the data already collected individually by each institution, such as ticketing and website traffic data, Data&Musée will be adding new sources of data created by visitors, thanks to the introduction of an intelligent guestbook (being developed by partner company GuestViews), social network analysis and an indoor geolocation system.

" Orpheo is looking to enhance visitor paths, but we're not sure whether this should be user-initiated or automatic, " explains Nel Samama, whose research laboratory at Télécom SudParis is working on the issue of geolocation with Orpheo. " Fully automatic flow analysis would require the use of radio or optical techniques, which work well in demonstrations, but are very unpredictable in real use. Including the visitor as an actor in this loop would be an enormous simplification."

The development of indication, prediction and recommendation tools

Based on the analysis of these data, the aim is to develop performance indicators for museums and build tools to personalize the visitor experience.

Working on modeling aesthetic taste from the data collected, Reciproque, a cultural engineering company, and the UNESCO ITEN Chair, also partners in the project, will deduce typical visitor silhouettes and content recommendations corresponding to them. This tool will help to better promote the wealth of French cultural institutions to visitors, and thus boost the tourism sector. Jean-Claude Moissinac, a teacher-researcher at Télécom ParisTech, is working on this aspect of the project, in partnership with Reciproque. " I'm particularly interested in data semantics, or the semantic web ," explains the researcher. "The idea is to index retrieved data in a homogeneous way, then build a graph and link them together. From this, we can deduce groups: it could be works, or users. We use this knowledge to propose itineraries.

For the partner institutions, the project involves setting up an interface to visualize attendance in the region and the seasonality of the public, as well as its segmentation in thematically related institutions. Museum performance indicators will also be developed. The various data collected will be used to develop analytical and predictive models for visits to cultural sites in France, as well as recommendations for action to help establishments define strategies to develop their activities.

With a subscription or membership fee system, this structured data could eventually be passed on to non-producing institutions or third parties with the agreement of the institutions and users: an economic model could then emerge, and Data&Musée could subsist beyond the duration of the project.