Knowledge Organisation I: the Universe of Knowledge and Modes of Formation of Subjects; Classification Theories, Canons and Principles; SKOS, Taxonomies, Folksonomy and Trends; and Mapping Subjects in DDC, UDC and CC

Library classification arranges documents and their surrogates in a helpful sequence, collocating related subjects, so that users can browse and librarians can find, file and shelve. There are three families of scheme. Enumerative schemes list ready-made class numbers for compound subjects (the Library of Congress Classification is the purest example; DDC began that way). Faceted or analytico-synthetic schemes list only simple isolates in facets and let the classifier synthesise class numbers for any compound subject using rules (Ranganathan's Colon Classification; the second edition of Bliss's Bibliographic Classification). Almost-enumerative schemes combine an enumerated base with synthetic devices (UDC, and modern DDC with its tables). E. Wyndham Hulme's principle of literary warrant (1911) holds that a scheme's classes should be based on the literature actually published, not on a philosophical order of knowledge.

Ranganathan's General Theory of Library Classification works in three planes: the idea plane (analysing subjects into facets and deciding their sequence), the verbal plane (expressing the ideas in words, the terminology of the schedules) and the notational plane (expressing them in ordinal numbers). His normative principles descend from basic laws (such as the laws of interpretation, impartiality, symmetry, parsimony, local variation and osmosis), through the five laws of library science, to canons for each plane. The canons of the idea plane are grouped by what they govern: characteristics (differentiation, relevance, ascertainability, permanence), succession of characteristics (concomitance, relevant succession, consistent succession), arrays (exhaustiveness, exclusiveness, helpful sequence, consistent sequence), chains (decreasing extension, modulation) and filiatory sequence (subordinate classes, coordinate classes). Principles for helpful sequence include later-in-time, later-in-evolution, spatial contiguity, quantitative measure, increasing complexity, canonical sequence, literary warrant and alphabetical sequence; principles for facet sequence include the wall-picture principle, the whole-organ principle, the cow-calf principle and the actand-action-actor-tool principle.

Knowledge organisation systems (KOS) is the umbrella term for classification schemes, subject heading lists, thesauri, taxonomies, ontologies and authority files. The Simple Knowledge Organization System (SKOS), a W3C Recommendation of 2009, is an RDF vocabulary for publishing such systems on the web as linked data: each concept gets a URI and is described with labels (skos:prefLabel, skos:altLabel), notes and semantic relations (skos:broader, skos:narrower, skos:related), and concepts can be matched across schemes (skos:exactMatch, skos:closeMatch). The Library of Congress Subject Headings, AGROVOC and many classification schemes are published in SKOS. A taxonomy is a hierarchical classification of concepts or content, usually for a particular organisation or website navigation. A folksonomy is the uncontrolled vocabulary that emerges from users' tags; it is cheap, current and in users' own words, but suffers from synonyms, homonyms, misspellings and no hierarchy.

• Faceted classification has spread far beyond libraries, into e-commerce filters and discovery-system facets, vindicating Ranganathan's analysis; the Classification Research Group (UK, 1952) and ISKO (the International Society for Knowledge Organization, 1989) carried facet theory forward.
• Schemes are now maintained continuously online: DDC through WebDewey, UDC through its Master Reference File, and linked-data versions of both.
• Automatic and machine-assisted classification uses text mining and machine learning to suggest class numbers and subject terms; human review remains necessary.
• Ontologies and SKOS link classification with the semantic web (Unit VIII), and user tagging is combined with controlled vocabularies in catalogues.

Melvil Dewey published the Dewey Decimal Classification anonymously in 1876 as a 44-page pamphlet. It divides knowledge into ten main classes, each into ten divisions and each division into ten sections, with pure Arabic-numeral notation of at least three digits and a decimal point after the third, so that the notation can be extended indefinitely. OCLC has owned the DDC since 1988. The 23rd edition (2011) was the last full print edition; the scheme is now maintained continuously and made available through WebDewey. DDC 23 has six auxiliary tables: T1 standard subdivisions, T2 geographic areas, historical periods and biography, T3 subdivisions for the arts and for individual literatures, T4 subdivisions of individual languages, T5 ethnic and national groups, and T6 languages. Number building combines a base number with table notation, for example 330.9 (economic situation and conditions) plus T2 -54 (India) gives 330.954, the economic conditions of India; and T1 -03 is the standard subdivision for dictionaries and encyclopaedias.

The Universal Decimal Classification was developed by Paul Otlet and Henri La Fontaine from DDC and first published in French in 1905 for their universal bibliography. It is managed by the UDC Consortium (since 1992) through a Master Reference File. Its main classes are 0 to 9, with class 4 left vacant after linguistics was moved to class 8, beside literature, in the 1960s. Its strength is its auxiliary signs, which make it the most widely used faceted-style scheme for documentation. The Colon Classification, Ranganathan's analytico-synthetic scheme, appeared in 1933; the sixth edition (1960) is the one most libraries used, and the seventh and last edition (1987) was edited by M. A. Gopinath. CC analyses every subject into five fundamental categories, Personality, Matter, Energy, Space and Time (PMEST), each introduced by its own connecting symbol, so the notation shows the structure of the subject.