This paper explores how domain-specific bibliographic metadata can act as a tool for cultural transformation within libraries and digital humanities infrastructures, with a particular emphasis on queer literature and its publics. Drawing on the Swedish project Queerlit and on ongoing international collaborations around the Linked Open Data thesaurus Homosaurus, the paper demonstrates how participatory metadata practices can foster inclusive and ethical resources for the co-construction of queer literary history and cultural memory.

In the project Queerlit, a bibliographic database containing Swedish queer fiction was created. Furthermore, a domain-specific thesaurus (QLIT) tailored to LGBTQI+ themes, identities, and narrative motifs was developed for use in the database. Identifying and indexing queer motifs has long been vital both for scholarship and for queer readers seeking recognition, connection, and community in literature (Cifor and Rawson 2022). Fiction –unlike historical documentation – does not carry a truth claim, yet it plays a central role in shaping queer cultural memory. Historically, queer communities have engaged in what McKinney (2020) terms information activism, producing subject guides, reading lists, and community-based catalogues to counter barriers to access. Today, grassroots digitization practices on blogs, forums, and social media continue this tradition: readers collectively tag, describe, and circulate queer fiction, generating vernacular knowledge structures parallel to institutional metadata.

The Queerlit project built on these practices, positioning community-generated knowledge as a foundational layer for scholarly and infrastructural development. The QLIT thesaurus builds on the Homosaurus, a LOD thesaurus of over 3,600 interconnected LGBTQ+ concepts, originally developed in the Netherlands and now maintained by the Digital Transgender Archive in the United States. Designed for interoperability across archives, libraries, and digital humanities projects, it offers a shared semantic framework that can bridge local collections and global knowledge infrastructures. A recent initiative aims to make the Homosaurus multilingual, with translations underway in Swedish, Dutch, German, French, and Japanese. The Swedish translation, initially developed within Queerlit and expanded within the Gothenburg Research Infrastructure in Digital Humanities (GRIDH), has followed a deliberately collaborative and community-based workflow.

This workflow combines online consultations, in-person workshops, and outreach to organizations, activists, students, librarians, and researchers. Expertise from cultural domains such as racialized and diasporic identities, and medical discourse, and vernaculars of sexual practices has been integrated into translation choices, emphasising both linguistic precision and community legitimacy. As part of this process, large language models such as ChatGPT were used as assistive tools for generating preliminary drafts and identifying terminological inconsistencies, without replacing human judgment or community authority.

By bringing together domain-specific metadata construction, community-driven translation, and interoperable LOD infrastructures, this paper argues that what we call transformative metadata practices not only enhance discovery and accessibility but also create hybrid spaces where institutional and community knowledge interconnect, generating more inclusive and socially responsive infrastructures. Transformative metadata refers to metadata practices that do not simply describe materials but intentionally revise and expand existing metadata systems, drawing on community knowledge and more accurately reflecting lived experience.