The Bite-sized ILI had three sessions: Augmented Libraries and Librarians, User Engagement in a Pandemic and Machine Learning for Libraries and Archives.
Augmented Libraries and Librarians
The day began with Martin Hamilton talking about the augmented nature of librarianship and librarians. Hamilton is a digital innovation consultant, advising organisations on how best to get the most out of emerging technologies such as artificial intelligence. He ran the Innovation Lab at Jisc and has experience as an internet researcher. In his view, AI is all around us. Likening it to an “extra brain”, he explained that AI was really an alphabet soup of technologies. Pattern matching can detect and classify objects; machine learning can reveal how much of his plays Shakespeare actually wrote; and AlphaFold can solve protein structures.
It can, however, go wrong, mislabelling photographs in racist terms and seeing a turtle as a gun. It’s important for information professionals to keep on top of these technologies and understand what they are capable of and where they are likely to fail, particularly since large companies such as Google tend to sweep problems under the carpet.
Eric Kokke has been with GO | School of Information in the Netherlands since 2013 and has a keen interest in training and educating information professionals for the future. He sees a shift from traditional skills to interpersonal ones. Communication is important to digital thinking, information literacy, and augmented librarianship. He did note an English language bias in most AI projects, as well as a bias toward prestigious journals, which usually publish only in English.
User Engagement in a Pandemic
Switching gears from AI technology Susana Cardoso, Chief of Client Services at the International Labour Organisation in Switzerland, described how her library and her staff coped with the sudden shift to working from home. They had to rethink user engagement since they had no access to their physical collection. WhatsApp helped them connect with users. Two consequences of working virtually was an increase in meetings—over 1,000!—and longer working hours. With 48 offices around the world, the ILO is truly a 24/7 operation.
Cardoso was particularly concerned about the health and wellbeing of the staff. Every morning, staff checked into a survey portal that simply asked how they were feeling. Navigating the new normal is draining, emotionally, on them, particularly since all are ex-pats, whose families are not in Switzerland.
The pandemic has forced the ILO librarians to learn new things and unlearn/relearn others. Echoing Kokke, Cardoso said that soft skills, such as empathy and compassion, became even more essential. In a virtual environment, where you can’t see body language, it’s very easy to say something that will be misunderstood. She found it reassuring to realize that she couldn’t control everything.The future for the ILO library involves opening up to physical access without an information desk in a new space in Geneva. The Research Information Hub will be key to this endeavour. Cardoso is excited about meeting users where they are rather than only inside library space. It opens up new possibilities for user engagement.
Helle Lauridsen has been involved with Danish libraries and eresources for a number of years and is now Product Manager for the Intelligent Material Management System (IMMS), Lyngsoe Systems. She continued the theme of the future of library space by noting that most libraries in Denmark have now reopened and users are coming back but user patterns are different: There are more holds and less circulation from browsing.
She used the Copenhagen Public Library’s use of evidence-based planning to determine the best use of space. At the Østerbro Library, books are displayed more in the manner of bookstores than tradition libraries, which increased circulation dramatically.
Machine Learning for Libraries and Archives
Turning to machine learning, Bohyun Kim, Chief Technology Officer & Professor, University of Rhode Island, and a columnist for Online Searcher magazine, described the possibilities of the technology while noting that it is not yet widely adopted in libraries and archives. Machine learning can expedite the processing of archival materials although challenges exist because of a scarcity of training datasets and the overall difficulty of establishing “ground truth”.
The image classifier in use for the Frick Collection, developed by the Frick Art Reference Library, Stanford University, Cornell University, and the University of Toronto, is a fine example of machine learning successfully labelling hundreds of thousands of images. Notre Dame University used natural language processing to generate MARC summary fields for its collection of over 5,500 Catholic pamphlets in PDF format. Machine learning also plays a role in full content extraction for archival materials.
Looking ahead, Kim anticipates some exploration of pre-trained off-the-shelf models, such as AWS Rekognition (aws.amazon.com/recognition) and Google Cloud Vision (cloud.google.com/vision). Libraries and archives need to continue their digitization efforts. She wonders, though, if more reliance on AI technologies will decrease the interactions among human experts, researchers, students, and the public who are seeking knowledge as they rely more on technology at the expense of the personal aspects of information seeking.
Future Internet Librarian Internationals
We are hopeful that we can meet once again for Internet Librarian International in person in London in 2022. This Bite-sized ILI was very successful, so maybe we’ll be able to do another virtual event soon. Meanwhile, we have this ILI365 newsletter to continue the community spirit of ILI. And don’t forget our LinkedIn Group as well.