The worldwide book publishing market is dominated by publishers based in what is known as the Global North. Of the world's top 50 publishers, 82% are based in Europe and North America. This means that access to material outside of this dominant Western-centric worldview is increasingly difficult.
In a recent audit, the library found that a full quarter of the books on resource lists came from just 19 publishers, all of whom were located in the UK or US.
The picture is no better or more diverse when it comes to academic journals.
20-25% of journal articles published worldwide are from the US. Of the 100 scientific journals with the highest rankings, 92% were from the US or UK. The highest-rated non-US/UK journal was 38th.
Systems such as databases and search engines are not free from bias. Bias can be built into them, often unconsciously, be reflecting the priorities and prejudices of those designing them. This is known as algorithmic bias. Many tools and pieces of software rely on machine learning based on data, where the system 'learns' and evolves based on the data it is provided. If the data is not free from bias, that bias will show in how the machine responds.
A paper published in Science found that as a computer taught itself English, based on the publicly available, internet-based data it consumed in the process, it developed prejudices against Black people and women - reflecting how widespread these prejudices are on the internet.
Bias can also present itself via metadata. Many academic resources, including databases and our own library catalogue, rely on subject keywords, subject headings and classification schemes for organisation and discovery purposes. The choice of language in these subject headings can be indicative of perspective or prejudice.
Problematic subject heading terms e.g “illegal aliens” for undocumented people are still routinely applied to book metadata, though this particular heading has been publicly challenged. There are many others. To combat this a library can decide to remove these headings, use their own subject headings to describe works or automatically convert these headings to display different wording if their library system is capable. In the Midlands, the universities' Mercian Metadata Group (of which Derby is a member) will be looking at creating alternative subject headings.
The Dewey Classification scheme, over 100 years old and subject to revision over the years, now actively seeks to adapt itself to become more diverse and inclusive and proactively invites interested parties to comment on their proposals to change Dewey. It can be a case of terminology or where the classification sits within the structure that needs changing. An example of how responsive a classification change can now be is shown when recently Dewey mapped the subject heading “Black lives matter movement” to 323.1196073 African Americans—civil rights.
Lastly, name authority records which record structured information regarding authors have in the past included detail on gender using only she and he, his and her. Now there is detailed guidance produced by the Library of Congress on best practices for recording information about gender in name authority records.
