International standards for KOS supporting information retrieval – ISO 25964 Part 2: Thesauri and interoperability with other Vocabularies

Presentation Description

This talk is a short report on ISO 25964 progress and outcomes focusing on Part 2: Thesauri and interoperability with other vocabularies. Part 2 of ISO 25964 is essentially a primer on how to map concepts to a thesaurus. This presentation was part of the panel “International standards for KOS supporting information retrieval and machine learning assessment”  at the 2026 International ISKO Conference on August 19 in São Paulo, Brazil.

Date: August 19, 2026

Venue: International Society for Knowledge Organization – ISKO 2026

Location: São Paulo, Brazil

Broad Topic: Taxonomy

Presenter: Joseph Busch, Principal Analyst, Taxonomy Strategies

Watch this Presentation:

Download Presentation: MP4 | PDF

Download Slide Script: PDF

[Slide 1]

This presentation is part of the panel “International standards for KOS supporting information retrieval and machine learning assessment.”  It was presented in person at the 2026 International ISKO Conference on August 19 in São Paulo, Brazil. This talk is a short report on ISO 25964 progress and outcomes focusing on Part 2: Thesauri and interoperability with other vocabularies.

[Slide 2]

First published by ISO in 2013 25964 Part 2 covers interoperability between different thesauri and with other types of structured vocabulary. It is essentially a primer on how to map concepts.

[Slide 3] Overview of ISO 25964 Part 2

Clauses 1-12 and 14-16 of ISO 25964 Part 2 deal with the principles and practicalities of interoperability, especially mapping.

Clause 13 provides additional guidance for handling pre-coordinated classes that occur in classification schemes and other vocabularies using the classification approach.

Clauses 17-24 deal with different types of vocabularies.

  • Classification schemes, taxonomies, subject heading schemes, and name authority lists – used for classifying or indexing resources
  • Terminologies, synonym rings, and ontologies – used for other purposes such as understanding specialized terms, query expansion, logic-based reasoning, retrieval augmented generation, etc.

[Slide 4]

Thesaurus mapping connects terms or concepts across different controlled vocabularies, enabling data interoperability and unified search. Simple Knowledge Organization System or SKOS uses standardized mapping properties exactMatch, closeMatch, broadMatch, and narrowMatch between concept URIs. Revisions to ISO 25964 align it with SKOS.

[Slide 5] Interoperability

Vocabularies can support interoperability by including relations to other vocabularies (i.e., mapping), by presenting data in standard formats, and by using systems that support common computer protocols.

[Slide 6] Mapping

Mapping is the process of establishing relationships between the concepts of one vocabulary and those of another. Three types of relationships are identified in the Thesaurus standard – equivalence, hierarchical, and associative. Any relationship may be a one-to-one or a one-to-many mapping. Equivalence mappings may be differentiated by degree, or not. An undifferentiated mapping is one that does not distinguish between different types of mapping, nor indicate different degrees of equivalence.

[Slide 7] Equivalence mapping

An equivalence mapping is one where the concept in the target and source vocabulary are considered identical in scope. A so-called Simple equivalence is a one-to-one mapping, for example, “Mobile Phone” is equivalent to “Cell Phone.” There are several one-to-many equivalence mapping types including compound equivalence, intersecting compound equivalence, cumulative compound equivalence, and compound equivalence involving a combination of target vocabularies.

[Slide 8] Exact, inexact, and partial equivalence mapping

Consider how much overlap is enough or too much in the context of search results to warrant an exact, inexact, or partial equivalence mapping.

[Slide 9] Hierarchical mapping

Hierarchical mapping may be established between concepts when one is clearly broader or narrower than the other. The thesaurus standard identifies three types of hierarchical mapping – generic, instances of, and whole-part.

[Slide 10] Associative mapping

The associative mapping type may be established between concepts when they are semantically associated to such an extent that items indexed with one are likely to be relevant in a search for the other. The distinction between an associative mapping and inexact equivalence is subjective, but may be made pragmatically according to the context in which the mappings are to be used taking into account, for example, the interests of the users, the volume of relevant resources to be searched, and the capability of the retrieval and display systems to make fine distinctions.

[Slide 11] Usefulness of types of mapping

Some types of mapping are more useful than others. Exact, simple equivalence is the only type that can be applied reversibly in any context. Cumulative compound equivalence can also be highly effective, provided it is applied in the direction intended and the combination of narrower concepts really adds up to the whole concept being mapped. Intersecting compound equivalence generally reduces the precision of searches. Neither form of compound equivalence can be used directly in the opposite direction. Hierarchical mappings tend to give more acceptable results in the direction from narrower to broader concept. Greater flexibility will be available in a particular application if the available mappings are not restricted to one per concept.

[Slide 12] Automated mapping techniques

There are several automated techniques for thesaurus candidate mappings including:

  • Lexical similarity between terms
  • Consideration of hierarchical and associative structure
  • Co-occurrence in indexed data
  • Reference to external knowledge sources, such as dictionaries or linguistic resources

Techniques can be used sequentially in an automatic mapping tool or combined in a voting system.

[Slide 13] Different types of vocabularies

The thesaurus standard provides a clause on each of 8 different types of vocabularies that might be mapped to a thesaurus. For each type of vocabulary the standard identifies key characteristics of the vocabulary, semantic components and relationships compared to a thesaurus, and recommendations for mapping.

[Slide 14] Examples of mapped thesauri

In reviewing the 2013 version of ISO 25964 Part 2, we recognized that one of the great strengths are the clauses that discuss how to map thesauri to different types of vocabularies and what the benefits are to do so. We also observed the need for more and varied examples of vocabularies that had been mapped to thesauri. There are not many and these are the well-known ones. We anticipate that there will be interest in using language-oriented AI to enable more mapping projects but the people doing the mapping may or may not have knowledge organization experience, and certainly not the depth of experience that has worked on these well-known mapped vocabularies.

[Slide 15] Status of ISO 25964 Part 2

What is the status of ISO 25964 Part 2? At the end of May the working group completed their initial review, comments, and draft editing of the current 2013 version of the standard. We also completed ISO Form 04 New Work Item Proposal. We identified the relation of the proposal to existing international standards and standardization work; how it supports of UN Sustainable Development Goals (SDGs); benefits to industry and commerce, government, consumers, labor, academic and research bodies, standards application businesses, and NGOs; the key nations and regions where the proposal is important for national commercial interests; international organizations or other ISO and/or IEC committees engaged in this work; and relevant international, regional, or national standards and regulations. If the revision request is approved, the working group will reconvene in Fall 2026.

[Slide 16] References

This reference slide provides an abbreviated bibliography for the presentation. I invite discussion and point you back to the standards and scholarly sources.

[Slide 17] Thank you

Please reach out to me with any questions.