Skip to content

Ontology design

In this section we will elaborate on the rationale supporting the selection of ontologies we chose to reuse for our model, with a focus on the economy of description, query optimization, and the inference potential of mereological and content type relations. Finally, we will present the final output of the knowledge organization process, a formal model meant to guide the preparation of the dataset.

Reuse of bibliographic ontologies

After establishing the need for WEMI levels to represent the different kinds of relationships between objects in our domain, the question arose as to which LRM implementation to base our model on. As of January 2025, the main frameworks that have integrated the LRM into Linked Open Data are Bibliographic Framework Initiative (BIBFRAME), Resource Description and Access (RDA) and Library Reference Model object-oriented (LRMoo).

  • BIBFRAME was developed in 2012 by the U.S. Library of Congress as a linked data alternative to MARC for the creation of bibliographic data. Expressed in RDF, it is based on three core categories of abstraction (work, instance, item), where work collapses LRM's work and expression entities. While we can see the reasons for this simplification, since bibliographic data describe works in terms of publication (manifestation) and edition (expression), we required the distinction between work as an artistic endeavor and expression as the result of an edition. For example, the contribution of editor Valerio Riva, as distinct from that of author Edoardo Sanguineti, to the first edition of Il Giuoco dell'Oca would be flattened not considering the two levels as separate. Similarly, it could not be clearly represented how Magdalo Mussio's article in Marcatrè is a “sketched” version of his animated film Go for your money. For this reason, we could not integrate BIBFRAME into our model.

  • RDA was released in 2010 as the new standard for descriptive cataloging in the U.S., a successor to the second edition of the Anglo-American Cataloguing Rules, and is jointly managed by a committee of Anglophone library associations, despite being the reference in an increasing number of other countries. In 2014, the vocabularies that represent RDA entities and controlled terminologies were published as RDF element sets in RDA Registry. The RDA data model adopts LRM as its theoretical underpinning and faithfully reproduces its structure with domain and range constraints. Its RDF representation provides rich detail relationship designators through a hierarchy of subproperties from general to specific and ensures interoperability with other Semantic Web initiatives by mapping its properties as subproperties of both LRMer properties and Dublin Core Terms. In addition to LRM properties, RDA includes several elements to accommodate additional information needed or currently part of the cataloging practice. For the sake of flexibility, its properties are further declared as both datatype and object properties, and can therefore be used with literals and other entities alike.

  • LRMoo succeds the 2010 FRBRoo model, its first official release was approved in 2024 and has been appointed by IFLA as a preferred implementation of the LRM. It translates the LRM entities and relationships into the formal ontology of ICOM's CIDOC Conceptual Reference Model (CRM), enabling alignment with cultural heritage data standards and facilitating integration between museum and bibliographic datasets. The CIDOC CRM is the most widely used ontology in cultural heritage modeling, and is particularly well suited for representing complex cultural object production events with nuanced designations of temporal relationships, part-whole inferences and roles of the agents involved, the latter thanks to the extension of properties with types (reification). Being an object-oriented, event-centered framework, to describe a cultural object a strict application must forcibly pass through production events and entities representing appellations, time-spans, and the like. In LRMoo, these are multiplied because the cultural object itself is already conceptually separated into multiple levels, each requiring its own related event.

We identified RDA and LRMoo as the two main ontologies that we could reuse in our model. To test the potential and critical aspects of each, we modeled the metadata of the most complex item in our domain—a comics magazine—as precisely as possible, applying the two ontologies separately.

Modelling of Serials

Modelling a comics magazine is challenging because it is a periodical that in specialized sources is described analytically, as each story within it has different subjects and authors, often masked behind pseudonyms.

In the Library Reference Model, serials—publications issued in parts, regardless of regularity or frequency, of which periodicals as a subset—are complex constructs that combine whole/part relationships and aggregation relationships. A serial work has a whole/part relationship with the individual issues published over time, while each issue expression is an aggregate of articles.

The whole/part relationship is hierarchical and structural: it describes a situation where a whole entity is composed of discrete parts that contribute to its existence, are integral to its identity, and cannot typically exist independently in the same context, such as chapters in a book. In contrast, the aggregation relationship is associative, grouping together expressions that remain independent, can exist separately, and each realize their own work, as short stories within an anthology. In the case of serials, the serial work provides the overarching issuing plan, which informs the aggregation of each individual issue.

According to IFLA LRM, a work is defined as having a certain commonality of content among its expressions. For serials, this commonality “resides in both the publisher’s and the editor’s intention to convey the feeling to end-users that all individual issues do belong to an identifiable whole, and in the collection of editorial concepts (a title, an overall topic, a recognizable layout, a regular frequency, etc.) that will help to convey that feeling.”[@lrm2018] This inherent characterization of serial works by their publication features—such as title, language, and publisher—means that any variation in these features results in the identification of a new serial work.

Reflecting this intrinsic link between the serial work and its defining publication features, LRM enforces the “WEM lock”: a cardinality restriction specifying that a serial work is realized by one and only one expression, which is embodied by one and only one manifestation. For this reason, an article is not considered a part of a periodical but is the expression of its own work. This expression is aggregated within the expression of an issue, whose work is part of the serial work.

In this model test, we limited ourselves to the description of the cover of issue 26 of Young Romance, the piece that Richard Hamilton chose to include in the iconic collage that started pop art. For the diagrams we used RDF Diagram, a Draw.io plugin (written in Rust!) that allows for real-time syntax validation and conversion in Turtle. We thank the Institute for Applied Informatics (InfAI) of Leipzig for sharing such a useful tool.

LRMoo in action

Modelling periodicals in LRMoo is currently limited by the fact that PRESSoo, the extension dedicated specifically to this, has not yet been updated to LRM and is only available in an earlier version based on FRBR and incompatible with the most recent version (7.1.3) of CIDOC CRM, so we chose not to use it. PRESSoo would offer the possibility of describing the planning aspects of a serial work (thanks to the extension of the “forsees use of” properties), although in our case, in which the periodical is not only described as a whole but the perusal of single articles is carried out, this information can be included in the issue description. However, inclusion of the issue in the magazine is possible only through a very generic property, lrmoo:R10_is_member_of, via the serial creation event1.

At first glance, the description is structured around creation events corresponding to the various WEMI levels, with actors participating through typed properties that define their roles. For the sake of this example, all concepts were drawn from Getty's Art & Architecture Thesaurus, as it is a widely used resource in the cultural heritage domain. For brevity, some entities connected to objects are nested, and their details can be expanded by clicking the plus sign. This includes a class for dimensions, which enables a detailed description of size, separating numerical values from their units of measurement.

See RDF triples
Describing Periodicals with LRMoo
@prefix : <https://f3mf.github.io/gdo/test-cidoc#>.
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>.
@prefix xsd: <http://www.w3.org/2001/XMLSchema#>.
@prefix dc: <http://purl.org/dc/elements/1.1/>.
@prefix crm: <http://www.cidoc-crm.org/cidoc-crm/>.
@prefix lrmoo: <http://iflastandards.info/ns/lrm/lrmoo/>.
@prefix aat: <http://vocab.getty.edu/aat/>.
@prefix gn: <https://sws.geonames.org/>.

<https://f3mf.github.io/gdo/test-cidoc#>
    dc:creator "Francesca Massarenti";
    dc:date "2025-01-09";
    dc:description "A test of the descriptive capabilities of the CIDOC-CRM ecosystem, taking as a case study the perusal of a periodical, the most complex publication to model within LRM."@en;
    dc:title "Describing Periodicals with LRMoo"@en.

:10-c a crm:E97_Monetary_Amount;
    crm:P90_has_value 0.1;
    crm:P91_has_unit aat:300411994.

:1913-10-11 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "1913-10-11T00:00:00"^^xsd:dateTime.

:1915-12-21 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "1915-12-21T00:00:00"^^xsd:dateTime.

:1917-08-28 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "1917-08-28T00:00:00"^^xsd:dateTime.

:1984-11-28 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "1984-11-28T00:00:00"^^xsd:dateTime.

:1994-02-06 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "1994-02-06T00:00:00"^^xsd:dateTime.

:2011-12-14 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "2011-12-14T00:00:00"^^xsd:dateTime.

:52-pages a crm:E54_Dimension;
    crm:P90_has_value 52;
    crm:P91_has_unit aat:300194222.

:ben-oda a crm:E21_Person;
    crm:P100i_died_in :ben-oda-death;
    crm:P67i_is_referred_to_by :ben-oda-n1,
        :ben-oda-n2,
        :ben-oda-n3,
        :ben-oda-n4;
    crm:P98i_was_born :ben-oda-birth;
    rdfs:label "Ben Oda".

:ben-oda-birth a crm:E67_Birth;
    crm:P4_has_time-span :1915-12-21;
    crm:P7_took_place_at gn:5392114.

:ben-oda-death a crm:E69_Death;
    crm:P4_has_time-span :1984-11-28;
    crm:P7_took_place_at gn:5097672.

:ben-oda-n1 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Ben Oda";
    crm:P2_has_type aat:300266386.

:ben-oda-n2 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Ben Kazuhito Oda";
    crm:P2_has_type aat:300404681.

:ben-oda-n3 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Ben";
    crm:P2_has_type aat:300404651.

:ben-oda-n4 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Oda";
    crm:P2_has_type aat:300404652.

:cover-creation a lrmoo:F27_Work_Creation.

:editing-1 a crm:PC14_carried_out_by;
    crm:P01_has_domain :issue-editing;
    crm:P02_has_range :jack-kirby;
    crm:P14.1_in_the_role_of aat:300025526.

:editing-2 a crm:PC14_carried_out_by;
    crm:P01_has_domain :issue-editing;
    crm:P02_has_range :joe-simon;
    crm:P14.1_in_the_role_of aat:300025526.

:editorial-policy a crm:E29_Design_or_Procedure;
    crm:P68_foresees_use_of aat:300014137,
        aat:300312356.

:feature-publications a lrmoo:F11_Corporate_Body;
    rdfs:label "Feature Publications"@en.

:inking a crm:PC14_carried_out_by;
    crm:P01_has_domain :cover-creation;
    crm:P02_has_range :joe-simon;
    crm:P14.1_in_the_role_of aat:300443955.

:issue-editing a lrmoo:F28_Expression_Creation.

:issue-publication a lrmoo:F30_Manifestation_Creation;
    crm:P14_carried_out_by :feature-publications;
    crm:P4_has_time-span :october-1950;
    crm:P7_took_place_at gn:5110629.

:jack-kirby a crm:E21_Person;
    crm:P100i_died_in :jack-kirby-death;
    crm:P67i_is_referred_to_by :jack-kirby-n1,
        :jack-kirby-n2,
        :jack-kirby-n3,
        :jack-kirby-n4,
        :jon-henri;
    crm:P98i_was_born :jack-kirby-birth;
    rdfs:label "Jack Kirby".

:jack-kirby-birth a crm:E67_Birth;
    crm:P4_has_time-span :1917-08-28;
    crm:P7_took_place_at gn:5128581.

:jack-kirby-death a crm:E69_Death;
    crm:P4_has_time-span :1994-02-06;
    crm:P7_took_place_at gn:5402405.

:jack-kirby-n1 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Jack Kirby";
    crm:P2_has_type aat:300266386.

:jack-kirby-n2 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Jacob Kurtzberg";
    crm:P2_has_type aat:300404681.

:jack-kirby-n3 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Jack";
    crm:P2_has_type aat:300404651.

:jack-kirby-n4 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Kirby";
    crm:P2_has_type aat:300404652.

:joe-simon a crm:E21_Person;
    crm:P100i_died_in :joe-simon-death;
    crm:P67i_is_referred_to_by :joe-simon-n1,
        :joe-simon-n2,
        :joe-simon-n3,
        :joe-simon-n4,
        :jon-henri;
    crm:P98i_was_born :joe-simon-birth;
    rdfs:label "Joe Simon".

:joe-simon-birth a crm:E67_Birth;
    crm:P4_has_time-span :1913-10-11;
    crm:P7_took_place_at gn:5134086.

:joe-simon-death a crm:E69_Death;
    crm:P4_has_time-span :2011-12-14;
    crm:P7_took_place_at gn:5128581.

:joe-simon-n1 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Joe Simon";
    crm:P2_has_type aat:300266386.

:joe-simon-n2 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Hymie Simon";
    crm:P2_has_type aat:300404681.

:joe-simon-n3 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Joe";
    crm:P2_has_type aat:300404651.

:joe-simon-n4 a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Simon";
    crm:P2_has_type aat:300404652.

:jon-henri a lrmoo:F12_Nomen;
    lrmoo:R33_has_string "Jon Henri";
    crm:P2_has_type aat:300404678.

:lettering a crm:PC14_carried_out_by;
    crm:P01_has_domain :cover-creation;
    crm:P02_has_range :ben-oda;
    crm:P14.1_in_the_role_of aat:300025115.

:no-26 a crm:E41_Appellation;
    crm:P190_has_symbolic_content "NO. 26".

:october-1950 a crm:E52_Time-Span;
    crm:P82_at_some_time_within "1950-10-01T00:00:00"^^xsd:dateTime.

:penciling a crm:PC14_carried_out_by;
    crm:P01_has_domain :cover-creation;
    crm:P02_has_range :jack-kirby;
    crm:P14.1_in_the_role_of aat:300443954.

:september-1947-june-1963 a crm:E52_Time-Span;
    crm:P82a_begin_of_the_begin "1947-09-01T00:00:00"^^xsd:dateTime;
    crm:P82b_end_of_the_end "1963-06-30T23:59:59"^^xsd:dateTime.

:serial-creation a lrmoo:F27_Work_Creation;
    crm:P14_carried_out_by :feature-publications;
    crm:P4_has_time-span :september-1947-june-1963.

:standard-golden-age-us a crm:E54_Dimension;
    crm:P90_has_value "7.75 x 10.5";
    crm:P91_has_unit aat:300379100.

:vol-4-no-2 a crm:E41_Appellation;
    crm:P190_has_symbolic_content "VOL. 4, NO. 2.".

:young-romance-26-cover-e a lrmoo:F2_Expression;
    rdfs:label "Cover of Young Romance 26"@en.

:young-romance-26-cover-w a lrmoo:F1_Work;
    lrmoo:R16i_was_created_by :cover-creation;
    lrmoo:R3_is_realised_in :young-romance-26-cover-e;
    crm:P2_has_type aat:300255278;
    rdfs:label "We've got to keep our love a secret, Marge..."@en.

:young-romance-26-e a lrmoo:F2_Expression;
    lrmoo:R17i_was_created_by :issue-editing;
    lrmoo:R4i_is_embodied_in :young-romance-26-m;
    lrmoo:R75_incorporates :young-romance-26-cover-e;
    crm:P2_has_type aat:300263751,
        aat:300264388;
    crm:P72_has_language aat:300387822;
    rdfs:label "Young Romance 26, Edited"@en.

:young-romance-26-m a lrmoo:F3_Manifestation;
    lrmoo:R24i_was_created_through :issue-publication;
    lrmoo:R70_has_dimension :10-c,
        :52-pages,
        :standard-golden-age-us;
    crm:P1_is_identified_by :no-26,
        :vol-4-no-2;
    crm:P2_has_type aat:300311879;
    rdfs:label "Young Romance 26, October 1950"@en.

:young-romance-26-w a lrmoo:F1_Work;
    lrmoo:R10_is_member_of :serial-creation;
    lrmoo:R3_is_realised_in :young-romance-26-e;
    rdfs:label "Young Romance 26"@en.

:young-romance-dc-sw a lrmoo:F18_Serial_Work;
    rdfs:label "Young Romance (DC, 1963 Series)"@en.

:young-romance-prize-sw a lrmoo:F18_Serial_Work;
    lrmoo:R11_has_issuing_rule :editorial-policy;
    lrmoo:R16i_was_created_by :serial-creation;
    lrmoo:R1i_has_successor :young-romance-dc-sw;
    crm:P102_has_title :young-romance-title;
    crm:P2_has_type aat:300203177;
    rdfs:label "Young Romance (Prize, 1947 Series)"@en.

:young-romance-title a crm:E35_Title;
    crm:P190_has_symbolic_content "Young Romance"@en.

RDA in action

Modelling in RDA is more concise and takes advantage of the coupling of properties with controlled vocabularies to assign the correct type designation to each WEMI level: category/genre for work, content type (i.e., the medium it is intended for) for expression, physical description of material and extent for manifestation. Serial work does not have a dedicated class but is defined by the properties “has extension plan” and “has frequency,” plus the designation of the issue number with “has numbering within sequence.” The issue is traced back to the magazine by the specific property “has issue,” which establishes a partitive relation.

The flexibility to choose between datatype and object properties as needed allows not to create entities for individual name variants when not relevant, but also to do so in cases such as the shared pseudonym “Jon Henri.” However, it should be noted that making this distinction explicit in the ontology would require specifying, case by case, which type of property is being employed. This is achieved by using distinct namespaces—such as rdamo: for object properties and rdamd: for datatype properties—both of which have the manifestation class as their domain. Moreover, a downside of the model's lack of typed properties is that, to express more artist-specific roles such as penciler and inker, it would need to be extended.

See RDF triples
Describing Periodicals with RDA
@prefix : <https://f3mf.github.io/gdo/test-rda#>.

@prefix dc: <http://purl.org/dc/elements/1.1/>.
@prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#>.
@prefix xsd: <http://www.w3.org/2001/XMLSchema#>.

@prefix rdaa: <http://rdaregistry.info/Elements/a/>.
@prefix rdac: <http://rdaregistry.info/Elements/c/>.
@prefix rdae: <http://rdaregistry.info/Elements/e/>.
@prefix rdai: <http://rdaregistry.info/Elements/i/>.
@prefix rdam: <http://rdaregistry.info/Elements/m/>.
@prefix rdan: <http://rdaregistry.info/Elements/n/>.
@prefix rdaw: <http://rdaregistry.info/Elements/w/>.

@prefix rdacc: <http://rdaregistry.info/termList/RDAColourContent/>.
@prefix rdaco: <http://rdaregistry.info/termList/RDAContentType/>.
@prefix rdaep: <http://rdaregistry.info/termList/RDAExtensionPlan/>.
@prefix rdafr: <http://rdaregistry.info/termList/frequency/>.
@prefix rdatb: <http://rdaregistry.info/termList/RDATypeOfBinding/>.
@prefix rdamat: <http://rdaregistry.info/termList/RDAMaterial/>.

@prefix lcsh: <http://id.loc.gov/authorities/subjects>.
@prefix lcgft: <http://id.loc.gov/authorities/genreForms/>.
@prefix gn: <https://sws.geonames.org/>.

<https://f3mf.github.io/gdo/test-rda#>
    dc:creator "Francesca Massarenti";
    dc:date "2025-01-12";
    dc:description "A test of the descriptive capabilities of the RDA ecosystem, taking as a case study the perusal of a periodical, the most complex publication to model within LRM."@en;
    dc:title "Describing Periodicals with RDA"@en.

:ben-oda a rdac:C10004;
    rdaa:P50115 "Ben Kazuhito Oda";
    rdaa:P50117 "Ben Oda";
    rdaa:P50118 gn:5097672;
    rdaa:P50119 gn:5392114;
    rdaa:P50120 "1984-11-28T00:00:00"^^xsd:dateTime;
    rdaa:P50121 "1915-12-21T00:00:00"^^xsd:dateTime;
    rdaa:P50291 "Oda";
    rdaa:P50292 "Ben";
    rdfs:label "Ben Oda".

:feature-publications a rdac:C10005;
    rdaa:P50025 "A Prize Publication"@en,
        "Prize Group"@en;
    rdaa:P50041 "Feature Publications"@en;
    rdfs:label "Feature Publications"@en.

:jack-kirby a rdac:C10004;
    rdaa:P50103 "Jacob Kurtzberg";
    rdaa:P50117 "Jack Kirby";
    rdaa:P50118 gn:5402405;
    rdaa:P50119 gn:5128581;
    rdaa:P50120 "1994-02-06T00:00:00"^^xsd:dateTime;
    rdaa:P50121 "1917-08-28T00:00:00"^^xsd:dateTime;
    rdaa:P50291 "Kirby";
    rdaa:P50292 "Jack";
    rdaa:P50428 :jon-henri;
    rdfs:label "Jack Kirby".

:joe-simon a rdac:C10004;
    rdaa:P50103 "Hymie Simon";
    rdaa:P50117 "Joe Simon";
    rdaa:P50118 gn:5128581;
    rdaa:P50119 gn:5134086;
    rdaa:P50120 "2011-12-14T00:00:00"^^xsd:dateTime;
    rdaa:P50121 "1913-10-11T00:00:00"^^xsd:dateTime;
    rdaa:P50291 "Simon";
    rdaa:P50292 "Joe";
    rdaa:P50428 :jon-henri;
    rdfs:label "Joe Simon".

:jon-henri a rdac:C10012;
    rdan:P80068 "Jon Henri".

:young-romance-26-cover-e a rdac:C10006;
    rdae:P20006 "en-US"^^xsd:language;
    rdae:P20387 :ben-oda;
    rdfs:label "Cover of Young Romance 26"@en.

:young-romance-26-cover-w a rdac:C10001;
    rdaw:P10078 :young-romance-26-cover-e;
    rdaw:P10088 "We've got to keep our love a secret, Marge..."@en;
    rdaw:P10256 lcsh:sh92001504;
    rdaw:P10451 :jack-kirby,
        :joe-simon;
    rdfs:label "We've got to keep our love a secret, Marge..."@en.

:young-romance-26-e a rdac:C10006;
    rdae:P20001 rdaco:1014,
        rdaco:1020;
    rdae:P20059 :young-romance-26-m;
    rdae:P20319 :young-romance-26-cover-e;
    rdae:P20338 :jack-kirby,
        :joe-simon;
    rdfs:label "Young Romance 26, Edited"@en.

:young-romance-26-m a rdac:C10007;
    rdam:P30011 "1950-10-01T00:00:00"^^xsd:dateTime;
    rdam:P30014 "no. 26",
        "vol. 4, no. 2";
    rdam:P30088 gn:5110629;
    rdam:P30169 "standard Golden Age U. S."@en;
    rdam:P30182 "52 pages"@en;
    rdam:P30304 rdamat:1025;
    rdam:P30309 rdatb:1007;
    rdam:P30420 :feature-publications;
    rdam:P30456 rdacc:1003;
    rdfs:label "Young Romance 26, October 1950"@en.

:young-romance-26-w a rdac:C10001;
    rdaw:P10078 :young-romance-26-e;
    rdfs:label "Young Romance 26"@en.

:young-romance-dc-sw a rdac:C10001;
    rdfs:label "Young Romance (DC, 1963 Series)"@en.

:young-romance-prize-sw a rdac:C10001;
    rdaw:P10004 lcgft:gf2014026515;
    rdaw:P10088 "Young Romance"@en;
    rdaw:P10141 :young-romance-26-w;
    rdaw:P10191 :young-romance-dc-sw;
    rdaw:P10218 gn:6252001;
    rdaw:P10365 rdaep:1003;
    rdaw:P10368 rdafr:1007,
        rdafr:1008;
    rdfs:label "Young Romance (Prize, 1947 Series)"@en.

Comparison

Criterion CIDOC CRM + LRMoo RDA
Focus Formal ontology expressivity Bibliographic cataloguing precision
Perspective Temporal Synchronic
Modelling patterns Historical relation pattern, typed properties Property proliferation (generally seen as bad ontology design)
Ease of querying Item metadata must be reconstructed through the related events Data are directly connected to the item they relate to
Inferencing potential Fine transitivity of partitive properties, including the chain of core links between WEMI levels Limited to entailment from subclass relationship chain
Size Streamlined, 8 000~ axioms Large, 80 000~ axioms
Maturity De facto standard for data integration in cultural heritage Still settling, but likely the foundation of semantic libraries
LRM integration LRM classes and properties are inserted into the CIDOC CRM taxonomy, with some friction2 Seamless
Reuse issues Mandatory declaration of entities not relevant to the domain Somewhat arbitrary breakdown between datatype and object properties

In comparing the two ontologies, a need emerged to balance the accuracy of item descriptions and the convenience of information retrieval with the expressiveness required to situate them within a complex scenario to reason on. Using the LRM classes as a central pivot, we therefore integrated RDA for metadata representation and LRMoo to model interactions between entities at the WEMI levels. This approach enables us to streamline the creation of entities and maintain a concise model without compromising the formal representation of conceptual interactions across the layers of the works, which is the most critical aspect of our domain.

We believe this practice is justified not only by the specific requirements of our case but also by the emerging necessity to integrate library data into cultural heritage Linked Open Data in a practical manner, while preserving the descriptive richness of cataloguing records. The criterion we have established ensures a clear separation of concerns and we think it could be applied to contexts beyond our own.

Reuse of referencing ontologies

CiTO, HiCO, PROV-O were reused for the interpretation acts extracted from the three scholars' publications where most of the intermedial relations have been pointed out.

The model formalizes three key aspects of the observed domain: 1) the description of the novel; 2) the description of items connected to the novel’s content; 3) the relation’s types (i.e. between text and items, between chapters, and object-object).
Each item, including Sanguineti's text, is described in its possible implemented entity in WEMI levels, following FRBR model, to which the relative alignment with the RDA classes corresponds.

Formal model for the intermedial relations, including interpretation acts

All relationships were modeled integrating LRMoo, CIDOC CRM, and RDA to achieve more granularity. Intermedial relations were divided into three subcategories: media transposition, media combination, and media reference. Each category concerns two different media considered under the perspective of WEMI levels and related by a specific connection.

An example of integrated modelisation: WEMI levels, relations and interpretation act.

The example below formalizes the description of Just what is it that makes today's homes so different, so appealing?, a collage by Richard Hamilton, described within the chapter LXXXVIII and its relationships with the comics Young Romance.


  1. Oury, C., Boulet, V., Dunsire, G., Howlett, L., & Reynolds, R. (2017). Definition of PRESSoo: A conceptual model for Bibliographic Information Pertaining to Serials and Other Continuing Resources (Version 1.3). International Federation of Library Associations and Institutions (IFLA). https://repository.ifla.org/handle/20.500.14598/1173 

  2. Beyond the minutiae, it seems redundant to have separate classes for F12 Nomen and E41 Appellation.