Hello Code4Lib community,
I am looking for practical patterns for representing a very large
web-native textual work as a coherent collection rather than as
hundreds of unrelated pages.
The immediate use case is an open-web New Testament
commentary/reference library. Individual book projects can span 100+
chapter pages, with a master guide, part-level pages, chapter-level
pages, and previous/next relationships. The human-readable side is
working, but the scale raises several library-technology questions
that may be familiar to people managing digital collections.
The core problem is that the “work” is not any single HTML document.
It is the graph connecting those documents.
A reader may arrive through search directly at chapter 73. That
chapter needs to identify the larger work, expose its sequence, give a
reliable route back to the master guide, and remain citable even if
surrounding presentation changes later.
The design rules I have converged on are:
1. Stable chapter identifiers and canonical URLs should be treated as
publication infrastructure rather than cosmetic slugs.
2. Every page should be understandable as a landing page because deep
linking is the normal web entry path.
3. Semantic segmentation should drive page boundaries. Splitting every
N words is technically easy but editorially destructive.
4. Navigation should expose hierarchy as well as sequence: collection
→ part → chapter, plus descriptive previous/next links.
5. The master index functions as a recovery point and collection-level
authority record, not merely a table of contents.
6. Link validation and markup validation are insufficient for a long
corpus. The publication pipeline also needs checks for duplicate
content, broken sequence, missing parent/child relationships,
fragmentary prose, and inconsistent naming.
7. Human-facing structure and machine retrieval often want the same
things: descriptive headings, explicit relationships, stable
identifiers, self-contained local context, and consistent metadata.
I am especially interested in standards or established practices that
can express “these 108 separately addressable HTML documents
collectively constitute one ordered work” in a durable way.
I have been looking at schema.org CreativeWork relationships,
linked-data approaches, collection/item modeling, and the conceptual
lessons of TEI, but I would be interested in what people are actually
using in production library or digital-humanities systems.
The live collection that prompted the question is here, if seeing the
structure helps:
https://douglasvandergraph.com/douglas-vandergraph-master-index/
Are there particular metadata patterns, RDF vocabularies, IIIF-like
approaches for text collections, or other models you would recommend
for representing ordered long-form textual works across many web
resources?
Best,
Douglas Vandergraph
https://www.youtube.com/@douglasvandergraph
Received on Sat Sep 19 2026 - 18:33:18 EDT