During the last three decades, Cold Spring Harbor Laboratory Meetings recorded videos of some 25,000 of hours of scientific talks at key meetings.
The CSHL Archives include these recordings. The collection documents decades’ worth of findings and key achievements in biology, medical research, and biotechnology. However, copyrights prevent us from placing the full videos on the Internet.
We have successfully completed our one-year pilot project to develop AI methods for recognizing specialized scientific terminology; created a historically sophisticated AI-powered chatbot to retrieve and synthesize information from the transcripts. The chatbot not only answers questions about the conference talks but locates where in the transcripts the answer to the query was found.
We are now working with a collection of Retrovirus meetings videos covering 1996 to 2024 period. For this project we are developing a semi- automated workflow to transcribe talks and extract metadata for inclusion in archival finding aids. Our goal is to retrieve specialized answers related to the Retroviruses field which play a significant role in the fields of biotechnology, medicine, and evolutionary biology.
By granting researchers, educators, and historians analytic tools and unprecedented access to decades of scientific dialogue, the chatbot will be a powerful resource for understanding and leveraging the science shared at CSHL’s world famous meetings.
I will describe our methodology, share problems we’ve troubleshooted, as well as discuss ongoing challenges and describe our plans for further development.
Hopefully, other archives can use our project as a model for their special collections with a copyright limitation

The archival object’s future is largely determined by its catalogue entry. How it is described often shapes how researchers usually first encounter this object and take it further to new spaces of meaning-making. It is a truism, but it bears repeating that the production of knowledge out of the archive depends substantially on the structure of that archive and its catalogue. For long, this has been under scrutiny as we rightfully examine the biases of the archive, from sourcing to appraisal, description to access. Researchers and archivists have, for generations, offered context to the archive, from marginalia on card catalogues to reparative description on digital catalogues. While these are critical, these annotations occur in a somewhat disjoint environment, removed from the core technical structure of the archive.

What is needed is a linked space to connect events (the processes that give rise to the archival), the database (the structure of the archival object), and narrative (the stories that emerge as a result of the presence of an archival object). At previous ICHST symposia, we discussed our early work in this space, starting from a narrative that draws on multiple interpretations of the same archival object (ICHST 2017), and a template for a consortium of science archives (ICHST 2021) with three meshed spaces: discovery, archival interpretation, and narrative. In this symposium, we extend this work and present one of the first prototypes of an annotation tool developed within the framework of archival description.

Ojas Kadu

The goal of this paper is to propose ways of doing historical analysis of the development of academic disciplines by combining data from subject bibliographies with social/institutional data found in various guides and disciplinary publications. The central case study will be the discipline of history of science (and, to a certain extent, science studies more broadly) as reflected through publications produced primarily by the History of Science Society. The core of the analysis will be based on data form the IsisCB database as well as the legacy data in the digitized version of the Isis Cumulative Bibliography, which together contain over 100 years of bibliographical work in the field of history of science. In addition to the bibliographical data, we propose the collection and digitization of other data found in guides to the discipline (published by the History of Science Society and other academic societies). By importing these and similar materials into the IsisCB database, we plan to build a social/institutional framework for studying the discipline and its academic production over the past century. The result should provide a means of studying the discipline of the history of science in ways that complement and advance standard histories of the field found in historiographical essays, biographical materials, and the like.