Knowledge Organization

Knowledge organization is the part of information science that decides how documents, records, and ideas should be described, named, grouped, and related so that someone can find them later.

15 frameworks, from 1841 to today · Information Science · Social Sciences

still a working view ended · each leads to its entry below

Overview

Knowledge organization is the part of information science that decides how documents, records, and ideas should be described, named, grouped, and related so that someone can find them later1. Its practitioners build catalogues, classification schemes, subject heading lists, thesauri, metadata schemas, and, most recently, formal ontologies and linked-data vocabularies23. The stakes are practical and philosophical at once: every scheme that puts a book under one heading rather than another, or declares that two concepts are "related," is making a claim about how knowledge hangs together—and that claim will shape what a reader can and cannot find. The history of the field is a long argument about who gets to make those claims, on what evidence, and for whose benefit.

Cataloguing Rules and the First Bibliographic Codes

The modern field begins in the library. In the nineteenth century, as national libraries amassed millions of printed volumes, the problem was consistency: two cataloguers describing the same book differently made the catalogue useless. Antonio Panizzi's rules for the British Museum catalogue, drawn up in the 1840s, and Charles Cutter's Rules for a Printed Dictionary Catalogue (1876) for American libraries, set out how to choose a main entry, how to transcribe a title, and how to form a subject heading. Cutter's "objects" of a catalogue—to let a user find a known item, and to let them find everything by an author or on a subject—remained the field's charter for a century.

These were enumerative systems: the cataloguer lists every subject that might be asked for and assigns a fixed heading. The great enumerative classification, the Dewey Decimal Classification (1876), arranged all knowledge into ten classes and their subdivisions, with notation (the code of numbers or letters that stands for a class) designed so that a shelf address could be read directly. The Library of Congress Classification, developed from the 1890s, and the Universal Decimal Classification, begun by Paul Otlet and Henri La Fontaine in the 1890s, followed45. Enumerative schemes were powerful because they were simple to apply and to browse, but they aged badly: new subjects had no ready place, and the scheme's view of knowledge was frozen at the moment of its making.

The Documentation Movement and the Birth of the Abstract

Around 1900, Otlet and La Fontaine in Brussels pushed beyond the library. Their Répertoire Bibliographique Universel aimed to index the world's published output on cards, and their Universal Decimal Classification added a faceted apparatus—auxiliary signs for place, time, and form—to the Dewey structure65. The Documentation Movement they founded treated the journal article, the report, and the patent, not just the book, as the unit to be organized, and it invented the abstract as a standard product: a short surrogate that could be filed, indexed, and searched. The movement's institutes and journals spread the idea that scientific literature was a resource to be managed systematically, and it set the agenda that information science inherited after the Second World War.

Faceted Classification and the Analytic Turn

The decisive break with enumeration came from S. R. Ranganathan, a Indian librarian who published his Colon Classification in 1933 and his Prolegomena to Library Classification in 1937. Ranganathan argued that a subject is not a single pre-existing slot but a compound built from a small number of facets—his five categories were Personality, Matter, Energy, Space, and Time17. A book on, say, the design of hospital buildings in India would be analyzed into its personality (hospital), matter (building), energy (design), space (India), and time, then synthesized into a class number by a fixed formula7. The scheme was faceted because the cataloguer builds the class at the moment of cataloguing rather than looking it up. This made new subjects expressible and made the underlying structure of a subject visible. The Classification Research Group in London, active from the 1950s, carried the faceted approach into general classification and influenced the design of later schemes and of thesauri.

Subject Headings, Thesauri, and Controlled Vocabulary

Alongside classification ran the tradition of verbal subject access. Library of Congress Subject Headings, developed from the late nineteenth century, gave cataloguers a controlled list of authorized terms with cross-references ("see" and "see also") to guide users from variant wording to the chosen form. The principle is controlled vocabulary: a fixed set of terms, with rules for which term is preferred and how terms relate, so that everyone describing a document uses the same words and everyone searching can find them12.

After the Second World War, the explosion of scientific and technical literature, and the arrival of computers, forced a redesign. In the 1950s and 1960s, coordinate indexing and post-coordination let indexers assign several separate terms to a document and combine them at search time, rather than pre-building a compound heading. The thesaurus emerged as the tool for this: a controlled vocabulary organized by broader, narrower, and related terms, with scope notes. The best-known early example, Chemical Abstracts' thesaurus work and the Engineers Joint Council thesaurus of the early 1960s, established the format; UNESCO's guidelines and the ISO standards of the 1970s codified thesaurus construction as a craft with rules for term selection, hierarchy, and reciprocity of relationships. The thesaurus became the standard bridge between the language of documents and the language of queries.

Machine-Readable Cataloguing and Cooperative Networks

The next turning point was technical. In the late 1960s, the Library of Congress developed the MARC format (Machine-Readable Cataloguing), which encoded each cataloguing element—author, title, subject, imprint—in tagged fields so that records could be exchanged between institutions by computer8. MARC made possible the shared cataloguing networks: OCLC in the United States, founded in 1967, and its counterparts elsewhere, let libraries copy and contribute records instead of cataloguing every book from scratch. The effect was a vast, standardized, machine-searchable bibliographic commons. It also entrenched the Anglo-American cataloguing codes, revised repeatedly (AACR2 appeared in 1978), as the international norm, and it made the catalogue record itself an object of data management rather than a local artifact.

Bibliometric Organization and the Citation Index

A parallel strand treated the literature as a network rather than a shelf. Eugene Garfield's Science Citation Index, launched in 1963, indexed the references in each article, so that a reader could move forward in time from a known paper to everything that cited it. Citation indexing is a form of knowledge organization: it organizes documents by their intellectual links rather than by assigned subject terms. Bibliometric analysis—counting citations, mapping co-citation clusters, measuring the growth and dispersion of literatures—grew from this base and gave the field quantitative tools for describing how knowledge is structured and used. It also introduced a lasting tension: citation-based organization reflects what authors chose to cite, which is not the same as what a subject scheme would declare relevant.

Evaluation and the Retrieval Turn

Once systems could be searched by machine, the question became whether they worked. The Cranfield experiments of the 1950s and 1960s, led by Cyril Cleverdon, introduced controlled test collections and the measures of recall (the proportion of relevant documents found) and precision (the proportion of found documents that are relevant)1910. This evaluation paradigm gave knowledge organization an empirical standard: a scheme or index was good if it retrieved relevant material and suppressed the rest. The measures were imperfect—recall is hard to estimate in a large collection—but they made the design of vocabularies and classifications testable, and they tied the field to information retrieval, its close sibling.

The Cognitive and Domain Turns

In the 1970s and 1980s, critics argued that schemes built on literary warrant and library practice were not neutral maps of knowledge. The domain-analytic approach, associated above all with Birger Hjørland from the 1990s, held that the concepts a scheme should use are determined by the practices and discourses of a particular knowledge domain, not by a universal logic. A classification for musicology should be built from how musicologists actually work, not from a general philosophy of knowledge. This reframed classification as a social and empirical activity, and it connected knowledge organization to the sociology of science.

A related critical strand asked whose knowledge a scheme encodes. Scholars pointed out that subject headings and classifications have historically marginalized or misrepresented groups—for example, headings that treated homosexuality as a disease, or schemes that filed women's studies as a subdivision of sociology. The critical approach treats every vocabulary as a political artifact and asks who was consulted, who was omitted, and what assumptions the structure carries. It has changed practice: major schemes have revised offensive headings, and the field now routinely debates the ethics of description.

Metadata Standards and the Shift to the Web

The arrival of digital documents and the web moved the field from the library catalogue to metadata: structured description attached to any digital object. The Dublin Core, a small set of fifteen elements proposed in 1995, was an early attempt to describe web resources simply enough for anyone to use. Richer standards followed for particular communities—MARC for libraries, EAD for archives, and later schema families for datasets and cultural heritage. The underlying idea is the same as cataloguing: agree on fields, vocabularies, and rules so that descriptions can be shared and understood across systems.

Entity-Relationship Models and Formal Ontologies

The most recent turn is from records to entities and relationships. The Functional Requirements for Bibliographic Records (FRBR), published by IFLA in 1998, modeled the bibliographic universe as entities—work, expression, manifestation, item—linked by relationships, rather than as flat catalogue records. This entity-relationship approach underlies the successor cataloguing standard, RDA (Resource Description and Access), and the library linked-data models built on it. In parallel, formal ontologies—machine-readable specifications of concepts and their logical relations—moved knowledge organization toward the semantic web1112. The Simple Knowledge Organization System (SKOS), a W3C standard from 2009, gave thesauri and classification schemes a way to publish their structure as linked data, so that a concept in one vocabulary could be explicitly connected to a concept in another. The ambition is a web of interoperable vocabularies rather than a single universal scheme.

The Field Today

Contemporary knowledge organization works across all these inheritances at once. Library cataloguing still runs on MARC and RDA, now migrating toward linked-data models; thesauri and controlled vocabularies are published as SKOS and reused in discovery systems; faceted structures, descended from Ranganathan, underlie the filters of nearly every modern search interface; and bibliometric and citation-based organization feeds research assessment and discovery tools. Domain analysis and critical perspectives have made the field self-conscious about the politics of description, while formal ontology and linked data have made it newly technical. The open questions are the old ones in new dress: how to keep vocabularies current as knowledge changes, how to reconcile many local schemes into interoperable systems without flattening their differences, and how to build organization that serves the people who use it rather than the institutions that maintain it.

References

  1. Knowledge organization (introduction, Facet analytic approaches, Traditional approaches, The information retrieval tradition (IR)), Wikipedia
  2. Controlled vocabulary, Wikipedia
  3. Simple Knowledge Organization System, Wikipedia
  4. Library of Congress Classification, Wikipedia
  5. Universal Decimal Classification (History), Wikipedia
  6. Paul Otlet, Wikipedia
  7. Colon classification, Wikipedia
  8. MARC standards (introduction, History), Wikipedia
  9. Evaluation measures (information retrieval) (Background), Wikipedia
  10. Cranfield experiments, Wikipedia
  11. Ontology (information science), Wikipedia
  12. Web Ontology Language, Wikipedia

The frameworks of Knowledge Organization

Each framework is a way of working the field has used: when it held, and the question it answers.

  1. 1841-2010

    Cataloguing Codes and Rules

    How should a book be described so catalogues agree?

    Fly to its tower in Nousfarer

  2. 1876-1990

    Enumerative Classification

    How can all knowledge be arranged in fixed classes?

    Fly to its tower in Nousfarer

  3. 1895-1960

    Documentation Movement

    How can the world's scientific literature be indexed and abstracted?

    Fly to its tower in Nousfarer

  4. 1898-Present

    Controlled Vocabulary and Thesaurus Construction

    Which words should describe a document so it can be found?

    Fly to its tower in Nousfarer

  5. 1898-Present

    Subject Heading Systems

    Which authorized terms should stand for a subject?

    Fly to its tower in Nousfarer

  6. 1933-Present

    Faceted Classification

    How can a compound subject be built from its parts?

    Fly to its tower in Nousfarer

  7. 1957-Present

    Cranfield experiments

    Do indexing languages actually improve retrieval?

    Fly to its tower in Nousfarer

  8. 1957-1990

    Information Retrieval Evaluation

    How should retrieval systems be measured and compared?

    Fly to its tower in Nousfarer

  9. 1963-Present

    Citation Indexing and Bibliometrics

    How do citations map the structure of knowledge?

    Fly to its tower in Nousfarer

  10. 1967-Present
  11. 1970-Present

    Critical Knowledge Organization

    Whose knowledge does a scheme encode and exclude?

    Fly to its tower in Nousfarer

  12. 1995-Present

    Domain Analysis

    How do a domain's own practices determine its concepts?

    Fly to its tower in Nousfarer

  13. 1995-Present

    Metadata standard

    What standard fields should describe a digital resource?

    Fly to its tower in Nousfarer

  14. 1998-Present

    Functional Requirements for Bibliographic Records

    What entities and relationships does a bibliographic record model?

    Fly to its tower in Nousfarer

  15. 2001-Present

    Formal Ontology and Linked Data

    How can concepts be defined formally and linked on the web?

    Fly to its tower in Nousfarer

How the ideas connect

Each arrow runs from an earlier idea to the later one it shaped. The list below says what passed between them.

Cataloguing Codes and RulesEnumerative ClassificationDocumentation MovementControlled Vocabulary and Thesaurus C…Subject Heading SystemsFaceted ClassificationCranfield experimentsInformation Retrieval EvaluationCitation Indexing and BibliometricsMachine-Readable Cataloguing and Coop…Critical Knowledge OrganizationDomain AnalysisMetadata standardFunctional Requirements for Bibliogra…Formal Ontology and Linked Data

Beyond the field

Where Knowledge Organization's ideas came from in other fields, and what they led to there: the strongest 12 of 65 links.

A city in Nousfarer: a red-carpeted stair climbing terraces to a crowned tower on an acropolis, more towers round it, forest and snowy mountains beyond.
A city in Nousfarer, the game played in the browser.

Knowledge Organization is a planet in Nousfarer

Its city has a tower for each of these 15 frameworks. Land, walk in, and climb one concept at a time.

Land on Knowledge Organization

Related fields

Fields its ideas are linked to

More in Information Science

See where Knowledge Organization sits on the Map of Human Knowledge