Addressing Systemic Bias in Academic Search Applications - A Community-based Approach

In late-2020, members of Ex Libris and its international user communities convened a collaborative group of discovery, metadata, and subject experts to identify and address systemic bias within its large article index as well the local catalogs of its membership - users of Alma, Primo and Summon. The goal was to address systemic racial and gender bias within discovery and create a venue where these issues and initiatives could be collected, openly discussed and addressed.

In the OCLC Research report on the findings of a 2017 RLP survey on equity, diversity, and inclusion (https://www.oclc.org/research/areas/community-catalysts/rlp-edi.html), Karen Smith-Yoshimura reported that over 50% of respondents from libraries, archives and museums identified ongoing initiatives within their local metadata systems to address their institution's EDI goals within their catalogs, their digitized collections, or within their ongoing descriptive practices. Unfortunately, systematic examination has not occurred within the large centralized article indexes our patrons use for their research where vendors contribute metadata and content for their databases.

While Ex Libris has a number of controls in place to address algorithmic and systematic bias within its products, additional enrichment and improvements can be introduced through the use of linked open data, multiple subject taxonomies, and various result weighting strategies to address content and metadata bias. Changing incoming metadata from publishers retrospectively can be a massive undertaking that any one part of the content supply chain can attempt, but those interventions may introduce their own biases when trying to correct something farther upstream.

Rather than trying to address one area of the supply chain, this approach brings publishers, librarians and the discovery provider to a virtual table to openly discuss bias and interventions. In making each contributor’s EDI metadata initiatives known, it is hoped the participants learn from one another, align their efforts, and work together

The presenters will discuss the charge and provide a progress report of this group. The group will show before and after searches and enhancements made through the discovery and content services teams, address how they identified problems and how they measured success.
The session will conclude with best practice recommendations for contributors, publishers, and institutions to include their content and metadata within the centralized article index and Publisher Zone as well as act as a call for interested participants.