Tuesday, 8 June 2021

Book discusses bias in algorithms, searching, and library "discovery" systems

The beSpacific blog today featured a post about a book titled "Masked by Trust: Bias in Library Discovery" by librarian Matthew Reidsma, who works at Grand Valley State University Libraries. He starts by discussing how most librarians develop, on their own, a favorite search that they use to evaluate "the dizzying array of search tools that are a part of modern librarianship." But as he studied the way variations in search terms can pull up results that can vary significantly, he decided to look more closely at the inner workings of search and the sorts of biases that can happen.

For this research, I investigated four library discovery systems, with varying degrees of access to each: Ex Libris’ Summon and Primo, OCLC’s WorldCat Discovery, and EBSCO’s EDS...Because these systems bear a remarkable resemblance to one another, it is easy for users to assume that they will all give the same results for similar searches. But in fact, the combination of di#erent centralized indexes, varied collection development practices at subscribing institutions, and competing relevancy algorithms means that it is unlikely that these systems will return the same results for the same searches.

The book is full of great examples and provides many useful  citations to other articles. It is written in plain language without a lot of technical terminology, and is available both from Library Juice Press and as a PDF download under a Creative Commons license.  

Hat tip: Sabrina Pacifici

No comments: