This is an archive of an inactive wiki and cannot be modified.

SKOS ISSUE-39 ConceptualMappingLinks: "Consolidation of existing SKOS mapping vocabulary" Proposal

This is a proposal for resolution of [ SKOS ISSUE-39]. See also IssuesProcess.

Resolution of this issue should enable SKOS to meet requirement [ R-ConceptualMappingLinks] (see the [ SKOS Use Cases and Requirements] document).

0. Introduction

NOTE (from Antoine): this proposal has been written following a [ discussion on the SKOS list] after the realease of the [ first proposal] written for SKOS mapping vocabulary

This proposal differs from this wiki's [ first proposal] and [ Alistair's proposal], which both tried to re-use as much as possible the basic SKOS semantic relationship vocabulary.

This proposal extends SKOS with the possibility to define mapping links between SKOS concepts coming from different concept schemes, which is currently allowed by the - unstable - SKOS mapping vocabulary [#SWBP-SKOS-MAPPING [SWBP-SKOS-MAPPING]].

This proposal avoids mixing with the existing SKOS semantic relations (e.g., skos:broader). Even if semantics of mapping links and "standard" (paradigmatic) semantic relations is similar, different properties are used for the mapping links. For instance, instead of advising the use of skos:broader between concepts from different concept schemes for mapping purposes, this proposal advocates the use of the skos:broadMatch property.

It also postpones the task of representing "complex" entities that could appear in mapping statements until a resolution is made for [ ISSUE-40 ConceptCoordination]. For the moment it is assumed that mapping links hold between simple concepts.

In the following, the namespace prefix skosm: is used to denote the SKOS mapping namespace

1. Vocabulary

The proposal transfers the following vocabulary from existing SKOS mapping vocabulary into the standard SKOS vocabulary:





The proposal introduces the following property in the SKOS vocabulary


The following properties are deprecated: skosm:majorMatch, skosm:minorMatch.

The properties skosm:AND, skosm:OR, skosm:NOT from existing SKOS mapping vocabulary may be 'moved' to SKOS core by a resolution on [ ISSUE-40 ConceptCoordination]. But as far as the SKOS mapping vocabulary is concerned in this proposal, they are temporarily deprecated.

2. Axiomatic Triples

RDFS statements:

  skos:mappingRelation rdfs:domain skos:Concept.
  skos:mappingRelation rdfs:range skos:Concept.
  skos:exactMatch rdfs:subPropertyOf skos:mappingRelation.
  skos:broadMatch rdfs:subPropertyOf skos:mappingRelation.
  skos:narrowMatch rdfs:subPropertyOf skos:mappingRelation.
  skos:relatedMatch rdfs:subPropertyOf skos:mappingRelation.

NOTE: this implies that, according to the RDFS semantics, the mapping relations are of type rdf:Property.

OWL statements:

  skos:exactMatch rdf:type owl:SymmetricProperty.
  skos:relatedMatch rdf:type owl:SymmetricProperty.
  skos:broadMatch owl:inverseOf skos:narrowMatch.

NOTE: according to the OWL semantics, the mapping relations are of type owl:ObjectProperty. owl:SymmetricProperty is a subclass of owl:ObjectProperty, and owl:inverseOf has owl:ObjectProperty as domain and range.

PROPOSAL 1: No statement is made here about subsumption links between subproperties of skos:mappingRelation (e.g. skos:exactMatch rdfs:subPropertyOf skos:broadMatch). This is left to a new ISSUE on Semantics of mapping relations.

PROPOSAL 1 (continued): No other statement statement is made here about OWL-semantics of mapping links, such as e.g. the transitivity of skos:exactMatch, skos:broadMatch, skos:narrowMatch and the symmetry of skos:relatedMatch. These problems should be dealt with by a new ISSUE on Semantics of mapping relations, which should be resolved in accordance to [ ISSUE-44 BroaderNarrowerSemantics].

3. Semantic Conditions

For the moment there is no further semantic condition. Depending on the choices made regarding semantic mapping links and coordination constructs, we could have here consistency checks of the form: "a concept cannot be declared a narrower (resp. equivalent, related) concept of another which is a negation of it". But this should first be discussed in the scope of [ ISSUE-40 ConceptCoordination]

Also, there are consistency checks that could be inspired from the resolutions made for [ ISSUE-44 BroaderNarrowerSemantics]: shall we allow cycles for broadMatch and narrowMatch? Should relatedMatch be disjoint with the transitive closure of broadMatch and the one of exactMatch?

PROPOSAL 1 (continued): the semantic conditions are left to the resolution of a new ISSUE on Semantics of mapping relations.

4. Consistent Examples

The example below shows conceptual mapping links between simple concepts.

 ex1:platypus rdf:type skos:Concept;
   skos:prefLabel "platypus"@en;
   skos:inScheme ex1:allAnimalsScheme.
 ex1:animals rdf:type skos:Concept;
   skos:prefLabel "animals"@en;
   skos:inScheme ex1:allAnimalsScheme.
 ex1:allAnimalsScheme rdf:type skos:ConceptScheme;
   dc:title "Extensive list of animals"@en.
 ex2:eggLayingAnimals rdf:type skos:Concept;
   skos:prefLabel "animals that lay eggs"@en;
   skos:inScheme ex2:eggSellerScheme.
 ex2:animals rdf:type skos:Concept;
   skos:prefLabel "animals"@en;
   skos:inScheme ex2:eggSellerScheme.
 ex2:eggs rdf:type skos:Concept;
   skos:prefLabel "eggs"@en;
   skos:inScheme ex2:eggSellerScheme.
 ex2:eggSellerScheme rdf:type skos:ConceptScheme;
   dc:title "Egg-seller view on animals"@en;

 ex1:platypus skos:broadMatch ex2:eggLayingAnimals.
 ex1:platypus skos:relatedMatch ex2:eggs.
 ex1:animals skos:exactMatch ex2:animals.

NOTE: when [ ISSUE-40] is solved, an example with compound concepts (e.g. from should be added here.

5. Inconsistent Examples

PROPOSAL 1 (continued): the introduction of inconsistent examples is left to [ ISSUE-40] and a new ISSUE on Semantics of mapping relations.

6. Entailment Rules

Many of the entailment rules concern the combination of mapping links with coordination construct s[ ISSUE-40 ConceptCoordination], as for instance: "if A is equivalent to concept B, and B is the intersection of C and D, then A is narrower than C and narrower than D".

PROPOSAL 1 (continued): entailment rules for mapping links are left to [ ISSUE-40] and a new ISSUE on Semantics of mapping relations.

7. Syntactic Constraints

PROPOSAL 1 (continued): syntactic constraints for mapping links are left to [ ISSUE-40] and a new ISSUE on Semantics of mapping relations.

8. Discussion

8.1 Conceptual mapping issue and other issues

The point of this proposal is minimum change to the current SKOS vocabulary. Indeed it delegates to other issues some of the problems that occur when considering conceptual mappings.

Mapping concept combinations, and general concept coordination issue

The existing SKOS mapping vocabulary proposes constructs for building combination of concepts (skosm:AND}}, {{{skosm:OR, skosm:NOT). Following [ discussions] on the SKOS mailing list, this proposal does not include them in the standard SKOS vocabulary. Rather, it postpones this issue until a solution has been found to [ ISSUE-40 ConceptCoordination] and [ ISSUE-45 NaryLinksBetweenDescriptorsAndNonDescriptors], which are about very similar problems.

Semantic interpretation of mapping links

A first possible interpretation of conceptual links considers the set of documents annotated/indexed by the concepts involved in the mapping links, as done in [#SWBP-SKOS-MAPPING [SWBP-SKOS-MAPPING]].

This proposal does not make the same assumption, which explains the minimal set of formal axioms it includes. Rather, it assumes that mapping links, as a parallel vocabulary to the SKOS semantic relations (see [ discussion]), should somehow "inherit" the semantics of these relations. With the fundamental difference that mapping does not come with the same confidence and authority status than established semantic relations. For instance, a mapping statement may not be endorsed by the creator(s) of the concepts that are mapped.

Mapping links between non-conceptual entities

Covered by [ ISSUE-49 LexicalMappingLinks]

Provenance of a mapping

Covered by [ ISSUE-47 MappingProvenanceInformation]

8.2 Other discussion points

The problems here are specific to the conceptual mapping case, and are tackled by this proposal.

skos:exactMatch or owl:sameAs?

owl:sameAs, as a property which denotes logical equivalence between classes, may be thought of for mapping purposes. However, the formal semantics of this property are different from what is needed for mapping concept schemes. When two resources are mapped by owl:sameAs, they are considered as one and a same merged resource. This imply for example that the preferred label of a first mapped concept would become the preferred label of the second one, even the latter already had a (different) preferred label. This is likely to make the resulting knowledge base completely useless for SKOS applications.

The majorMatch/minorMatch case, and motivation for skos:relatedMatch

The existing SKOS mapping vocabulary [#SWBP-SKOS-MAPPING [SWBP-SKOS-MAPPING]] introduces an arbitrary threshold of 50% overlap for distinguishing between majorMatch and minorMatch mappings. It is unclear whether this threshold has some legitimacy. What if someone wants to introduce a finer-grained characterization, having links for matches corresponding the 0-30%, 30-60% and 60-100% overlap ranges? This proposal advices SKOS user to create such specific weighted links as a specialization of the skos:relatedMatch property.

This proposal indeed includes skos:relatedMatch as a mapping link that corresponds to the generic association as found in many controlled vocabulary standards. From the [ SKOS use cases], there is no explicit motivation for such a link. Yet, there is a need for an umbrella property that would cover the situations mentioned above. Additionally, there was some [ consensus] on having a SKOS mapping vocabulary that would offer features parallel to the standard (paradigmatic) semantic relations (skos:braoder, skos:narrower and skos:related).

A previous version of this proposal has included a skos:overlappingMatch property as a super-property of the deprecated majorMatch and minorMatch properties. It was dropped for the two reasons below, summed up from different threads on the SKOS mailing list (e.g. [ here] and [ here]):

Mappings within one concept scheme

This proposal considers that mapping links typically hold between concepts from different concept schemes. It however does not forbid the creation of mapping statements within one single concept scheme.

This could happen for instance if a given application release an unstructured list of tags as a SKOS concept scheme. Other applications may compute hierarchical links between these tags, in order to provide the list with useful structure. In this case of one scheme being mapped to itself, these new links cannot be considered as (paradigmatic) semantic relations endorsed by the creator of the orginal scheme. The creators of the mapping links may also be reluctant to publish the new structure as a brand new concept scheme.

Representing rationale for a mapping

Some use cases (e.g. [ HILT]) bring the motivation for introducing different types of mapping links that mirror the way conceptual similarity is obtained, e.g. by lexical match based on plural/singular comparison.

This is however orthogonal to the general specification of conceptual links: the lexical matching criterion mentioned in this use case can apply both to exactMatch, broadMatch or relatedMatch links. In fact this flavor of links, focused more on the mapping process than on the semantics of the links, is probably out of SKOS scope.

Furthermore, it is unsure wether we can provide a list of such kinds of mapping rationale, even at a broad level. Survey of the ontology matching research gives some hints at general families of techniques, but it is unclear whether we can produce an exhaustive list that would cover all needs.

It is therefore proposed here to leave to specific SKOS users the task of specializing the semantic mapping properties, or even to choose another solution (e.g. reification of maping statements) in order to represent such mapping rationales.

Conceptual Mapping links between conceptual entities that are not of type skos:Concept

Some situations may imply to release the domain and range assumptions for the semantic relations used for mapping:

If these are allowed, then we face again the problem that motivated [ ISSUE-33 GroupingsInConceptHierarchies]. And the different options proposed for this issue shall be considered again:

NOTE FROM THE EDITOR: I would favor the first option.

PROPOSAL 2: a NEW ISSUE is raised on mapping links between conceptual entities that are not of type skos:Concept

SkosDesign/ConceptualMapping/ProposalTwo (last edited 2007-12-16 16:48:09 by AntoineIsaac)