Our Research Lead, Tori, shares her reflections on a recent conference in Florence, where she detailed the progress on Data Lifeboat alpha and the value of collecting from social media for heritage practitioners.
Last week I had the pleasure of presenting our paper, Built to Last: Preserving User-Generated Networked Images for the Next Century, to the academic cultural heritage community at the International Conference of Cyber Humanities, held at the Fondazione CR Innovation Centre in Florence, Italy.
This paper focused on the value of social media collecting for cultural heritage researchers, practitioners and academics. Whether uploading heritage projects to create a digital record, in order to maximise reach and impact; collecting community responses to those networked images in the form of comments and tags; or simply appealing to the future historians of our present age, social media images are worth safeguarding. However, recognising the difficulty of securing social media (check out our Mellon research report to find out why), we proposed Data Lifeboat as a preliminary solution. Mirroring the metadata, both technical and social, from Flickr.com, Data Lifeboat is simple, faithful, and cohesive in its mechanism and display.
The feedback was generally inquisitive and positive. I was pleased to hear how many projects were using Flickr as a research resource, including one photogrammetry project at the University of Turin delving into Flickr for photos of Expo ’58 and what became of the buildings in the half-century that followed. Many others attendees were keen to try Data Lifeboat with their own institutional images on Flickr.
A few highlights: Ines Vodopivec delivered a rousing yet thought-provoking keynote on the necessity for institutional collaboration, standard-setting and critical dialogue across subject specialties. Detailing the community research work of AI4LAM, it is clear that application in the G.L.A.M. sector needs to be pioneered by those who understand the contents (sensitive or otherwise) of their in-house collections. We look forward to joining the network at Fantastic Futures in London later this year.
I also particularly enjoyed Federico di Pasqua’s paper, Generative AI for Ancient Insights, which compared emergent Retrieval-Augmented Generation to more generalist Large Language Models (the dominant mode) in cultural heritage. The criticality of localised, specific models is a question not only of better retrieval of information, for interpreting Homeric texts for example, but orients us towards more sustainable practices.
If you missed our presentation at IEEE-CH and would like to know more, you can find our slides here. We’re pleased to share that result of this presentation will be our first (!) published paper on Data Lifeboat… watch this space.
