Publications remain essential because they give scholarship boundaries. They gather evidence, argument, authorship and responsibility into objects that can be reviewed, cited, catalogued and preserved. Yet the intellectual life of knowledge rarely respects those boundaries. A method introduced in one paper may become standard practice elsewhere. A concept can be reformulated across decades. A claim may be corrected while its first publication remains permanently accessible. A distinction may become foundational even though no infrastructure identifies the distinction itself. What persists intellectually is often smaller than a document and longer-lived than any particular version of it. This creates a structural mismatch between how scholarship is administered and how ideas actually develop. One response is to identify several levels separately: the publication, the concept, its successive versions, its important occurrences and the relations connecting it to other objects. These levels should not compete. The publication remains the context of authorship and review; the concept becomes the object whose continuity can be traced across contexts. A version records a historically specific formulation. An occurrence shows where and how that formulation appears. A relation states whether one unit revises, applies, narrows, translates, contests or depends upon another. Such granularity could improve citation as well as retrieval. Scholars often cite an entire paper when the actual dependency concerns one definition, method or distinction. A more precise system could identify the conceptual dependency while preserving the conventional citation of the document from which a particular formulation was taken. It could also help computational systems distinguish a current definition from a historical one, or a genuine conceptual continuation from a lexical coincidence. The obvious risk is identifier inflation. Not every sentence, phrase or argumentative move deserves an independent address. Granularity has to follow intellectual durability rather than technical enthusiasm. The relevant threshold is whether an object has enough coherence to recur, enough significance to be cited or revised, and enough continuity to justify maintenance. The aim is not to atomize scholarship but to expose some of the internal architecture that documents already contain. Future scholarly infrastructures may therefore operate simultaneously at two scales: publications as composed and accountable wholes, and selected concepts as evolving objects whose histories exceed any single publication.
Berners-Lee, T., Hendler, J. and Lassila, O. (2001) ‘The Semantic Web’, Scientific American, 284(5), 34–43.
Briet, S. (1951) Qu’est-ce que la documentation?
Buckland, M. K. (1991) ‘Information as Thing’, JASIS, 42(5), 351–360.
Kuhn, T. et al. (2016) ‘Decentralized Provenance-Aware Publishing with Nanopublications’, PeerJ Computer Science, 2, e78.
Jaradeh, M. Y. et al. (2019) ‘Open Research Knowledge Graph’, K-CAP 2019.