Knowledge organization system (KOS),concept system, orconcept scheme is the generic term used inknowledge organization (KO) for the selection of concepts with an indication of selected semantic relations.[1] Despite their differences in type, coverage, and application, allKOS aim to support the organization of knowledge and information to facilitate theirmanagement andretrieval.
KOS vary in complexity from simple sorted lists to complex relational networks. They represent both structural and functional features, and serve to eliminate ambiguity, control synonyms, establish relationships, and present properties.[2] From their origins inlibrary and information science (LIS), KOS have been applied to other domains and disciplines within science and industry, although scholarly research and debate remain primarily within the KO field. Challenges of KOS include ambiguity of terminology, repercussions of biased systems, and potential obsolescence.
KOS can be expressed inRDF andRDFS as per theSimple Knowledge Organization System (SKOS) recommendation byW3C, which aims to enable the sharing and linking of KOS viathe Web.[3]
One of the largest collections of KOS is theBARTOC registry.[4]
While different schema of KOS have been proposed, most are generally arranged in terms of the complexity of their construction and maintenance.[5][6][2] Some scholars argue that organizing KOS on a spectrum oversimplifies the shared characteristics among them, and may even result in a non-ideal structure being chosen.[7][8]
The following types are not exhaustive, and are often not mutually-exclusive in practice.
Term lists are the least structured form of KOS. They includelists,glossaries,dictionaries, andsynonym rings.[5][2]Authority files andgazetteers may also be considered term lists,[5] however other scholars categorize them anddirectories as "metadata-like models".[2]
Examples include theUnion List of Artist Names name authority file and theGeoNames gazetteer.
KOS that emphasize specific (and often hierarchical) structures includesubject headings,taxonomies, categorization schema, andclassification schema &systems.[5][2]
Despite inconsistent use of the terms "categorization" and "classification" in some literature,[8]categorization is generally loosely-assembled grouping schema and may include attributes that are not mutually exclusive (or havingfuzzy boundaries), whileclassification is related to the arrangement of non-overlapping and mutually-exclusive classes.[2][9]
Classification schema may be universal (such asDewey Decimal Classification andInformation Coding Classification) or domain-specific (such as theNational Library of Medicine Classification).[2]
The types of KOS with greatest complexity and which utilize connections between concepts includethesauri,[note 1]semantic networks, andontologies.[2][7]
One of the most prominent examples of a semantic network isWordNet.
Certain structures proposed to be considered types of KOS—but are not consistently included in schema—includefolksonomies,[7]topic maps,[7][8] web directory structures,[10] publication organization systems,[5] andbibliometric maps.[8]
Some KOS organize other KOS themselves—for instance,PeriodO is a gazetteer ofperiodization categories.[11]
Some early KOS were developed as a support system forabstracting and indexing services to be used by specially-trained searchers.[5] With the growth of information digitization, usability became increasingly accessible, and more complex structures were developed.[5]
Prominent examples of KOS outside of LIS includeorganism taxonomy in biology,[1] theperiodic table of elements in chemistry,[1]SIC andNAICS classification systems for industry & business,[5] andAGROVOC agricultural controlled vocabulary.[12][13]
The study and design of KOS is an ongoing topic of discussion among KO scholars.
[There is] a serious lack of vocabulary control in the literature on controlled vocabulary.
— Bella Hass Weinberg, 1998[8]
Inconsistency of terminology within the study of KOS is a common issue.[8] For instance, "ontology" is used for both a specific type of KOS as well as a generic term for any KOS. The terms "taxonomy", "classification", and "categorization" are also sometimes used interchangeably.
As knowledge can be historically and culturally biased, scholars have also discussed how KOS themselves can perpetuate harmful practices or stereotypes.[14][15] For example, a number of concerns and criticisms about the classification of mental disorders in theDiagnostic and Statistical Manual of Mental Disorders have been raised,[16] contributing to ongoing revisions.
Ethical and intentional design approaches have been proposed for multi-perspective KOS in efforts to mitigate bias and other harmful practices.[14]
The possible obsolescence of the thesaurus and other simpler KOS has been the topic of debate, especially in the face of increasingly complex ontologies, the growing usage of "Google-like retrieval systems", and the move of KO theory and research away from LIS and towardcomputer science.[17] Supporters of thesauri argue its continued usefulness for metadata enrichment, vocabulary mapping, and web services,[18] as well as its usage in specific domains such as corporateintranets and digital image libraries.[19]