Designing Idea-Native Discovery for Public Libraries: Beyond Container-Centric Search

Library discovery systems are container-centric: they retrieve documents. A reader who arrives with a question about a theme — exile, the social contract, the tragedy of the commons — must translate it into a query about containers, and then reconstruct the theme by reading across whatever the query returned. The collection holds the relationship; the catalogue does not. This paper specifies idea-native discovery: a paradigm in which ideas are represented as first-class objects, with typed relationships linking them to each other and to source material, so that a collection can be navigated at the level of meaning rather than only at the level of documents. It gives four design principles — local neighbourhoods rather than global graphs, explicit typing, progressive disclosure, and intentional path-based expansion — and a reference architecture separating extraction, representation, storage and presentation. ⭠The paper's most consequential claim is practical rather than architectural. Because ideas and thematic relationships are not copyrightable expressions, an idea layer stores no protected text and generates no substitute summaries — which is what makes this deployable in a public institution where retrieval-augmented generation over licensed collections is not. The design is offered as a specification for institutions under exactly those constraints: interpretable, auditable, and pointing always back to the authoritative source rather than standing in for it.