Masked by Trust: Bias in Library Discovery

Product Description

The rise of Google and its integration into nearly every aspect of our lives has pushed libraries to adopt similar “Google-like” search tools, called discovery systems. Because these tools are provided by libraries and search scholarly materials rather than the open web, we often assume they are more “accurate” or “reliable” than their general-purpose peers like Google or Bing. But discovery systems are still software written by people with prejudices and biases, library software vendors are subject to strong commercial pressures that are often hidden behind diffuse collection-development contracts and layers of administration, and they struggle to integrate content from thousands of different vendors and their collective disregard for consistent metadata.
 
Library discovery systems struggle with accuracy, relevance, and human biases, and these shortcomings have the potential to shape the academic research and worldviews of the students and faculty who rely on them. While human bias, commercial interests, and problematic metadata have long affected researchers’ access to information, algorithms in library discovery systems increase the scale of the negative effects on users, while libraries continue to promote their “objective” and “neutral” search tools.

From the Author

Libraries have adopted “Google-like” tools, called discovery systems, to search their collections, presenting scholarly articles, books, archival items, and other library holdings in a single results list, removing the need for the user to search individual databases, one at a time.

Because these tools are provided by libraries and search subscription databases of scholarly materials rather than the open web, we often assume (and libraries actively promote the idea) that they are more trustworthy than their general-purpose peers like Google. Although the content may be more academic, library discovery systems are still software written by people with prejudices and biases. As I will show, library discovery systems struggle with accuracy, relevance, and human biases, and these shortcomings have the potential to shape the academic research of the students and faculty who rely on them.

To date, critical algorithm scholars have focused on commercial search tools and social media platforms, but there has been no study of the influence of algorithms in academic search tools. In 2016, I conducted a small analysis of one relevance algorithm in ProQuest’s Summon. In my study I analyzed 8,000 Summon searches and uncovered a pattern of biased reference results against women, black people, Muslims, the LGBT community, and the mentally ill. “Stress” in the workplace was equated with working women, any search on mental illness implied that this was a “myth,” and a search for information about rape in the United States suggested learning more about “hearsay evidence.” I have continued studying the algorithms of all major discovery systems and have found similar results. Far from “neutral,” these systems must be examined to uncover the claims, beliefs, and prejudices they perpetrate that challenge the values and ideals of libraries.

This book will extend the scholarly discussion surrounding critical algorithmic studies to commercial library discovery tools. It will contribute to the critical examination underlying how library collections are searched, as well as how contemporary academic research is done. In addition, the book will present best practices for software developers, instructional designers, information literacy instructors, as well as library managers for assessing, teaching about, and mitigating bias in library search tools.

This book is primarily aimed at academic librarians and those studying library and information science. While the book in nominally about a technical aspect of librarianship, the book addresses the impacts of bias in discovery systems for public service, instruction, and technical service librarians. Rather tha