How AI offers new opportunities for the librarian
Digital transformation has also reached libraries. Previously, reviewers wrote more than fifteen thousand short reviews of new books each year, which were part of the purchase information. Based on a large dataset and this text, libraries decided whether they wanted to include a book in their collection. Since the beginning of 2022, artificial intelligence has been used for metadata and organizing the books. “We are breaking away from ‘book review’ and replacing it with description,” explains Nina Nannini, director of NBD Biblion (the library supplier).
NBD Biblion previously ran into the limitation that the old work process was not scalable. A selection was made upstream of the books offered to the libraries. Costs were high, lead times were long, and quality was variable. To deal with this problem, NBD Biblion (since 2019) decided to work together on a solution. The choice fell on Bookarang. Nannini: “Bookarang is a company that has developed a program capable of doing text analysis using artificial intelligence. The computer scans and analyzes a book file from start to finish and the program extracts about six hundred data points. This concerns factual elements such as title, number of pages, author, publisher, keywords and ISBN number.
In addition to this factual data, the computer is also capable of describing more subjective things. Nannini: “Then you have to think about genre, mood, language use and reading level. The text analysis is then checked by annotators. They are people who check the data produced by the computers and provide feedback if necessary. Subsequently, bibliographers also read the text to see if it is correct from a bibliographic point of view.
“AI offers many opportunities. Books are easier to compare to each other, there is no more personal preference in purchasing information and this leads to a faster and more efficient way of working.
“Using the algorithms, we can give reading advice based on someone’s reading level.”
Find a book
AI was previously only used to write text analytics, but it seems to be useful for many other applications. “Before, you walked into the library and looked for a book that you couldn’t remember the title. For example, a book about a frog that lives in the swamp and has a blue house. If you presented this to a library staff member, chances were they couldn’t help you further. Bookarang’s technology now allows us to search for that specific book based on that information. »
Give reading tips
The data model is also used to compile lists of themes. In libraries, the collection is often linked to topical themes. Like Ukraine, saving energy, etc. With this, you will find the relevant titles in no time. Even better, it can be used for reading promotion. Nannini: “Using algorithms, we can give reading advice based on someone’s reading level. If you can’t read very well, it’s motivating when someone recommends a book that you can read. We conducted research with students in pre-vocational secondary education. We asked them the question: who do you want advice on what books to read from: teachers or classmates? They trusted neither one nor the other, but what the computer advised them. The younger generation has great faith in today’s technology.
Focus on content
The system only deals with the content of a book, Nannini points out. “That’s the big difference compared to, say, Netflix or Amazon, which give behavioral advice. People should be told about books that you might never expect, but are right for you. And to make sure all the books come out of the closet, because maybe you want to read some older books that you have no idea exist.
Skepticism and distrust
Nannini acknowledges that skepticism and distrust were high when AI made its appearance. “There wasn’t a lot of knowledge in the book business about AI. Libraries were also initially skeptical and wanted to examine and see the data before they could believe it. We really started this initiative from the idealism that we want to help everyone find the right book. We want to contribute to the promotion of reading, because things are not going well with the development of language in the Netherlands. »
Not as a replacement
This development raises the question of whether the librarian will become redundant. Nannini: “It was never designed to replace the employee, but to support the employee because he has less time. This employee has been given many additional duties over the past few years. For example, he or she is the point of contact for the government’s digital inquiry point, where information is to be provided on 21st century skills and healthy living. Such tools offer only one solution. As an employee, if you enjoy reading historical novels, you may not know everything about thrillers or other genres. With this catalog you can always inform and advise people more widely.
“The younger generation has great faith in today’s technology.”
AI is now proving its value for libraries and it can be of great value for businesses as well, says Nannini. “Using AI to automate processes is one thing. What’s even more interesting is that you create many new options with it. It’s worth getting AI techs to think along with someone who knows the problems of a work process. And think very carefully in advance about the ethical framework in which you want to apply and test AI. This looks very different for a library than for an online store. »
She concludes: “The business community can do a lot, but it starts with identifying a question or problem for which data and text analysis can offer a solution. We specialize in books, which is a very specific niche. Europe is lagging behind in AI. So take a good look at the examples from China, the United States, Israel or India in your own sector.
Follow Executive Finance on LinkedIn!
“Bacon trailblazer. Certified coffee maven. Zombie lover. Tv specialist. Freelance communicator.”