The Secret Life of UW Libraries Catalogers and Metadata Specialists
Student Spotlight Series #5: María Fernanda (Fer) Palomares Carranco
By Crystal Yragui, Metadata Librarian & Interim Co-Head, Metadata & Cataloging Initiatives Unit · University of Washington Libraries
You may be surprised to hear that before you can access a library resource when you need it, a lot of work must be done to get that resource into the UW Libraries catalog. A whole department of librarians, staff, and students are quietly working away behind the scenes to get new resources into the catalog and to find innovative ways to enhance the Libraries’ metadata*. This series of blog posts will highlight our brilliant student employees and the work they do to make your tasks of searching, identifying, selecting, and obtaining library resources easier and more effective.
|
Mini Glossary: |
What is metadata?Simply put, metadata is data that provides information about other resources. Metadata makes finding, identifying, selecting, and obtaining resources for any purpose easier. It can tell you what the content is about, what type of thing a resource is, help explain its origin, nature, and lineage, and provide useful access information. |
What is Wikidata?Wikidata provides deeply structured and reliable, human-checked metadata which drives not only its sister projects funded by the Wikimedia Foundation but all over the Semantic Web (the global internet of linked data). Wikidata is one of the largest open source triple stores (or linked data graphs) available for training AI tools, and is ingested to feed commercial web tools such as Google Knowledge Boxes and more. |
What is Linked Data?“Linked Data is a machine-readable, standard format for storing information. In linked data, each element of information to be linked is given a unique identifier, called a Uniform Resource Identifier (URI). Connections between these elements are encoded using the Resource Description Framework (RDF), in which information is encoded in “triples” which include pairs of objects and the relationship between those objects. These triples are what create the interconnectedness of the data on the web, allowing for computers to extract meaningful connections across the internet, and thus integrate information from multiple disparate sources.” — Network of the National Library of Medicine |
María Fernanda (Fer) Palomares Carranco came to the University of Washington iSchool’s Master of Library and Information Science program with a Bachelor of Arts in Sociology and Mexican American and Latino/a Studies from the University of Texas at Austin. Fer joined the University of Washington Libraries Cataloging and Metadata Services Department for Directed Fieldwork with the goal of making library metadata and related workflows more equitable and accessible.
Fer’s work at UW Libraries addresses the broader challenges faced by cataloging and metadata departments in academic libraries across the world at this moment who are working to address inconsistencies and inefficiencies in library metadata, prioritizing ethical descriptive practices, and creatively addressing resource challenges.
Fer’s fieldwork focused on two distinct projects:
- Assessing complex, multi-step workflows for metadata application profiles (input forms for describing digital images held by the Libraries), streamlining them into more democratized and easy-to-use spreadsheets.
- Working alongside international Wikidata teams to evaluate and implement new data modeling practices to address inconsistencies and inaccuracies in the ways personal pronouns are modeled and shared publicly in Wikidata.
Responding to Staffing Challenges: Improving Workflows and Efficiencies
Fer’s work aided the Libraries in responding to staffing reductions in the Cataloging and Metadata Services Department. She developed new processes to streamline workflows, enabling more library staff to use and update the ContentDM metadata application profiles (MAPs) directly. Rather than relying on multiple documents and scripts, Fer helped her advisor, Metadata Librarian Crystal Yragui, to strip these MAPs down to simple shared spreadsheets, uncovering several inconsistencies and reducing redundancies.
At the end of her directed fieldwork, Fer presented a prototype for the new proposed MAP to the University of Washington Libraries Metadata Implementation Group (MIG) for approval. Her work was approved, and is now informing an ongoing project to simplify the rest of the MAPs.
Improving Accuracy in Data Modeling for Personal Pronoun Information
The University of Washington provides useful background information on using correct pronouns here. The relationship between personal pronouns and other aspects of personal identity is not a simple thing to quantify because it is highly individual and unpredictable as far as machines are concerned. Personal pronouns cannot be reliably determined based on other data points for human beings, such as our given names, sex assigned at birth, gender identity, or appearance. Many people use one set of personal pronouns in one environment, and another set in another. Others are happy to use any personal pronouns. Some people use neopronouns, while others do not. Some want their personal pronouns to be publicly known, and others prefer to keep theirs private.
For this project, Fer jumped in with both feet to tackle serious issues with data modeling and practices in Wikidata for personal pronouns. She learned the Wikidata model from the ground up and helped an international, cross-institutional team to engage the global Wikidata community with ideas for a new data model and best practices for personal pronouns. This work was challenging, and involved putting together multiple complex proposals to detail ethical and technical shortfalls with current modeling and agile solutions that would not only function for the Wikidata graph, but would serve the needs of linguistic user groups all over the world.
The model the team set out to repair was fraught with issues, such as confusion of personal pronouns with other aspects of identity such as gender identity, sex, titles, and honorifics. Personal pronouns had also been modeled inconsistently as Wikidata objects, being classed as exemplars of both linguistic objects called lexemes and as more general objects called items. These objects are shaped distinctly in Wikidata and cannot be used and queried in the same ways. Data modeling inconsistency makes data difficult to manage, query, and clean up for users and data managers.
In Wikidata at the time of Fer’s directed fieldwork, there was an absence of universally-understood best practices regarding the need for references when stating that someone used a particular pronoun, or regulations about using bots to infer peoples’ personal pronouns based on other irrelevant data. This led to widespread assumptions about personal pronouns for items about human beings being recorded in Wikidata and then ingested downstream into the semantic web (read: search engine knowledge boxes like the ones you see in Google). The incorrect connection in Wikidata between personal pronouns and gender identity was leading to automated assignment of both personal pronouns and gender identity values for people, often based on nothing more than a person’s first name. More complex issues of outing (revealing someone’s private sexual or gender identity without their consent) and misgendering (referring to someone in a way that does not reflect their gender identity) were also addressed by the group’s proposals.
Thanks in part to Fer’s hard work, the second proposal put forth by the group as a Wikidata Request for Comment has now reached consensus and is ready to be implemented in Wikidata. The solutions proposed there will begin to solve many of these problems.
So, what does it all mean? Why does it matter?
The impact of these changes, when implemented, will result in more accurate and ethical data and processes, for example:
- A male-identifying person named “Sandy” who uses “he/him/himself” for example, who bots assign the pronoun “she” and then assign the gender “female”, will be corrected to reflect the accurate pronoun. Gender references based on pronouns used are no longer valid.
- Personal pronoun sets will reflect a consistent data model “under the hood”. This makes searching for peoples’ pronouns in Wikidata neater and more accurate, which has positive implications not only for Wikidata but for all web services which are based on its free, open, structured data (for example, Google Knowledge Graphs).
- Restrictions on bot editing combat misgendering by bots making assumptions based on personal pronouns, and best practices provide human editors with concrete guidelines surrounding privacy and standards for references for personal pronoun statements. This will significantly cut down on outing, misgendering, and other harms surrounding personal privacy for the people being described in Wikidata.
With both of these projects, Fer’s Directed Fieldwork demonstrated her readiness to step into the world of professional metadata librarianship, and we are so grateful for the time she spent working with us.
###
Welcome, Huskies! UW Libraries is hosting a wide variety of Dawg Daze events to introduce you to our amazing Library resources and programs! Explore our spaces, meet our staff and connect with other students throughout the week including tours, board game events, karaoke, and more! See all Libraries Dawg Daze events below, and find them in your
9/19 KARAOKE FRIDAY at the Tateuchi East Asia Library 1-4pm:
9/19 Tabletop Game Night at Odegaard Library 5:30-10pm: 
9/19 Sounds from the Vaults with UW Ethnomusicology Archives 1-5pm: Join curator DJ Vallier as he spins discs from UW’s historic and subterranean 
Odegaard Library Tours – 9/18, 9/19, 9/22 and 9/23: 
Preserving UW and Olympic History: Conserving the 1936 Rowing Uniforms:
Mini Exhibit: It’s a Dawg’s Life
UW Excellence: UW Libraries’ Adam Schiff, who has been awarded the 
May 28- 
May 20 – 135 Years of Open Access to Federal Government Information
May 21– Teaching With Large Language Models: Hackathon: Large Language Models (LLMs) have recently been top of mind for many in higher education, and UW has hosted a variety of great talks and events centered on LLMs and teaching. The Teaching With LLMs Hackathon picks up where those events leave off: it provides a good stretch of time where instructors can individually or collectively revise their teaching materials with respect to LLMs. 
Some examples student achievements of this year’s class include: 