1 Introduction
Distinguishing between what can be considered an artwork from what is not, and reach a precise definition of art itself, can be challenging in a dynamic world in which new forms of art are constantly introduced [1, 47]. For this work, we refer to artwork(s) as a “visual object or experience consciously created through an expression of skill or imagination” [
7]. Since an artwork, for its nature, cannot usually be completely understood only from its objective characteristics, it is subjective to observers’ interpretations. In this work, we present a new ontology, ICON, which models art interpretations of the artworks’ subject matter and its possible meanings. In the context of art comprehension, an interpretation is intended as “any kind of assignment of meaning or significance to artworks” [
47, p. 113]. Through interpretations, art historians can claim different kinds of explanations about the artwork concerning different aspects such as the artwork’s content, the tendency of the art of a century, or the reception of artworks by the public [
31, p. 114]. Among them, the interpretations considering the comprehension of the subject matter of an artwork fall under the domain of iconographic interpretation, which is based only on internal aspects of the artwork (e.g., a child with bows, arrows, and wings depicted in an artwork is recognized as Cupid and as a symbol of love) [
31, p. 124]. The interpretation conducted on this basis can be further enriched by other types of interpretations that take as evidence an external source—that is, the cultural context [
31, p. 131]. In this sense, artworks are read as symptoms of the contemporary culture [
37]. For example, the fact that, during the Middle Ages, the classical deities were represented deprived of their classical form can be read as the incapability of the Medieval artists and society of retaining a classical model with its appearance, since it was too far from their taste and from the new Gothic representational conventions [
38].
Therefore, the domain of knowledge of artworks content interpretation is complex and characterized by the subjectivity of the author of each claim. For these characteristics, we believe that Semantic Web technologies, with a focus on ontologies, are a suitable tool to conceptualize this semantic expressivity by means of the high level of granularity and flexibility offered. Nevertheless, the study in our earlier work [
4] highlights how current knowledge graphs, in the Semantic Web, do not express iconographic and iconological statements with the correct granularity, suggesting the introduction of domain-specific ontologies that conceptualize these aspects.
Since the domains of iconography and iconology described previously concern the description and comprehension of artworks’ content, we model art interpretations according to them. Therefore, the ontological modeling proposed here aims to answer to the following research questions:
RQ1:
To what extent is it possible to model the domain of knowledge of iconology and iconography, to provide art historians and cultural institutions a way for expressing complex art subjects and meanings, the interlinking among them, and claims about their interpretations?
RQ2:
Can the newly developed model outperform current work in terms of granularity?
Among the approaches adopted by the scholars in the context of the artworks’ content interpretation, the iconographical-iconological method
1 formalized by Erwin Panofsky had, in the last century, the greatest relevance and influence on contemporary art historians [
43]. Panofsky followed the approach first adopted by Warburg [61] in his studies, which was aimed at understanding the artistic subjects and motifs as witnesses of socio-cultural phenomena. Whereas today his studies are fundamental for the approach itself, the prevailing perspective is the one formulated by Panofsky in a three-layered framework of the artwork’s understanding [
34]. On that basis, some scholars proposed variations over levels subdivision [
3,
8], sometimes including other aspects, such as the artist’s psychology [
62] or the iconic language of the image [
32]. Although the framework is recognized as a valid method or approach, it is commonly accepted to acknowledge that this kind of interpretation is subjective and intuitive when practically applied [
34].
For its complete formalization and historical relevance, we will refer to Panofsky’s approach as a representative method of the discipline,
2 yet considering the enrichment given by other scholars mentioned in Section
2. In detail, Panofsky subdivides the act of interpretation in three levels, the first two of which fall under the traditional domain of iconography—that is, the identification of iconographies (level 2) attributes and variants with which they can be represented (level 1), whereas the last level concerns the socio-cultural interpretation of artworks content, closer to Warburg’s iconological approach [
59].
We consider Panofsky’s theoretical approach as suitable for modeling because it is, from the authors’ point of view, the most complete attempt in the literature to formalize the discipline in detail. Indeed, it not only defines the mechanisms of interpretation and its components but also gives precise indications on (i) the types of subjects, and (ii) how the subjects, their components, and meanings are related.
In this work, we consider the concept of interpretation according to the presented theory—that is, an observer interpreting what is represented by one or more artworks, their possible iconography, and meaning. Therefore, other types of interpretations, such as the results of the observation of the physical object (e.g., measurement) and its metadata definitions (e.g., dating, author, and title attribution) are not considered in the scope of this work. The concept of subjectivity is limited to the described situation. As the content and meaning interpretation always depend on the viewer perception and background knowledge, multiple, incompatible interpretations may derive from different observations of the same artwork. For the sake of clarity, we will use the term interpretation to refer to the overall claim made by an observer in which the artwork’s subject matter and meaning are understood (e.g., “this painting depicting Venus expresses a wish of good marriage”), whereas the term recognition will refer to the identification of subjects taking place at the different levels (e.g., the recognition that a child is Cupid). Therefore, an interpretation is composed of a set of recognitions contributing to the overall understanding of the artwork.
The article is structured as follows. Section
2 introduces the theoretical background of art interpretation, on which the ontology is based. In Section
3, we analyze the state of the art of ontologies and general domain schemas that deal with interpretations, relevant concepts like symbolism, or description of cultural heritage objects. Then, Section
4 explains the requirements that were used to design the ontology, along with potential users and lexical usage. Section
5 describes our design process of the ontology, describing in detail all the design iteration that have been undertaken to model different parts of our work, together with the axiomatization details, the alignments, and reuse of existing ontologies. In Section
6, the evaluation process of the ontology is explained, with both automatic evaluations provided by relevant tools and quality-based evaluation over the granularity potential of the model. Section
7 shortly deals with the release of the ontology and the publication of its documentation. Finally, Section
8 concludes the article with a final discussion about the impact of the ontology, its current limitations, and future work.
2 Theoretical Contribution
In this section, we illustrate Panofsky’s theory of art interpretation [
36,
38] introducing the theoretical aspects that were fundamental for the ontology design phase. During the act of interpretation of an artwork, the formal aspects, such as forms, colors, and compositions, are perceived. When these formal aspects are interpreted as precise objects, the sphere of meanings is considered. According to Panofsky, there are different types of meaning that can be interpreted in an artwork, subdivided in three layers. The depth to which the artwork can be understood depends on the background knowledge of the observer: the more he has knowledge about the artist, stylistic conventions, and cultural context of him/her/them, the more the interpretation at each level is correct, including more profound insights on cultural meanings.
The first layer, namely the pre-iconographical description, requires the knowledge of the representational conventions to allow a correct recognition of factual (e.g., objects, people, actions) and emotional meanings, namely primary or natural subjects. In detail, this description is achieved by the recognition of pure forms (i.e., combinations of forms and colors) as carriers of primary subjects. Pure forms such recognized are called
artistic motifs, and their combinations are
compositions. An enumeration of the recognition of artistic motifs constitutes a pre-iconographical interpretation of the artwork [
36, p. 28].
If the observer is familiar with the literary sources known by the artist, then the subjects already identified at level 1, namely the artistic motifs or compositions, can be recognized at the second level by the combination of them with concepts and themes, obtaining for example characters (e.g., Venus), personifications (e.g., Virtue), or events (e.g., the Battle of Cascina). The artistic motifs such recognized are called
images or
invenzioni, namely the term used by ancient theorists to identify stories and allegories. Allegories are defined in opposition to stories as “combinations of personifications and/or symbols,” although there are many intermediate possibilities between them [
36, p. 29, note 1].
Finally, by knowing and understanding the cultural and societal aspects of the artist’s time, it is possible to read the artwork and the subjects identified at the previous levels as symptoms of the contemporary society, of the artist’s beliefs and personality or as the expression of meanings voluntarily inserted.
The scholar highlights that the first two levels are a description of facts and are under the domain of iconography, whereas the last level is in the domain of iconology, which is a synthetic intuition rather than a description. Table
1 resumes the synoptic table from the work of Panofsky [
36, pp. 40–41], integrating it with further explanation of concepts implemented in the ontology modeling and by adding a practical example.
Although some following scholars made some variations of the model,
3 the subdivision of the interpretation in levels is generally accepted. In detail, we highlight that some scholars put the attention on relevant aspects that we considered during the modeling. Van Straten [
59] highlights the difference between intentional and unintentional meanings by dividing the third level in two layers. In this way, he recognizes that some more profound meanings are voluntarily expressed by the artist (e.g., the concept of “good wishes” that the artist wants to express in an artwork made for a wedding occasion) and more unconscious, cultural meanings. Another relevant addition is made by Imdahl [
32]. He underlines that the iconic sense of the image should not be ignored, since it is the primary means through which visual arts communicate. For example, the disposition of figures in the space can provide insights on their relationships, their actions, or in expressional meanings.
Furthermore, the preliminary studies conducted in previous work [
3], which considered approximately 50 articles of the major scholars of iconography and iconology, collected in several other works [
29,
32,
37,
38,
59,
61,
62], highlight important features that may be involved in an iconographical-iconological interpretation that should not be ignored. Indeed, from the bottom-up analysis emerged that the following aspects may be relevant for the supporting of the third-level meaning, namely (i) the direct citation of visual patterns from other artworks, (ii) the dependency of certain iconographies from specific sources, (iii) the role of style, (iv) the fact that a cultural meaning generally involves more than one artwork, and (v) the fact that scholars often extend claims by other scholars.
4 Requirements
Based on the iconographical and iconological literature analysis described in Section
2, we formulated the requirements using the SEEMP framework [
55]. The terminology was mainly selected from Panofsky’s theory [
36]. The output document is described in Tables
2 and
3.
The purpose of the ICON ontology is to formally represent the domain of knowledge of iconology and iconography with a high granularity level, to allow specific quantitative analysis that can be interesting for domain experts. It is intended to be used by (i) cultural institutions willing to publish their data about artwork content in linked data, (ii) art historians interested in answering iconographical and iconological research questions in a quantitative way, and (iii) developers who plan to use computer vision to associate recognized elements to portions of artworks. Therefore, the ontology aims at being implemented in different contexts, meeting the needs of different types of users. We use the OWL2 format to make the ontology available and reusable.
Therefore, the main non-functional requirement
14 is the reuse and alignment to the standards shared across the community to allow of reusability. Furthermore, the
Competency Questions (CQs) formulated for the functional requirements aim at expressing the various aspects of the iconographical-iconological approach described in Section
2. We summarize the main themes that can be extracted from the requirements listed in Table
3 as follows:
(1)
The identification of subjects at each level of interpretation needs to be included.
(2)
The variations of iconographical subjects (e.g., Cupid represented with a bandage and griffon talons rather than only with traditional attributes, namely wings and arrows [
38]) must be described.
(3)
The symbolic and cultural meanings attributed to each subject must be included.
In addition, relevant characteristics of the approach are considered, namely:
(4)
The attribution needs to be subjective.
(5)
The sources used by the scholar to state its claim need to be present.
(6)
The clear distinction between the subjects described at a general level (i.e., the background knowledge necessary for iconographical descriptions cited in Table
1, found in standard vocabularies, describing, e.g., Cupid as a “child with wings and arrows”) and their specific manifestation in a single artwork (e.g., Cupid with griffon talons) needs to be done to allow us to describe variations.
(7)
The ontology must allow the integration of one claim within the agreeing claims quoted by the art historian as a source of shared and accepted knowledge.
(8)
The ontology must allow to gather sets of agreeing recognitions made in a coherent situation (e.g., a scholar making an interpretation in a specific paper expanding on other scholars’ interpretations, therefore including their claims in his own), which may gather the interdependent recognition made at different levels (e.g., a scholar recognizes the level 2 subject “Cupid,” since he recognized at level 1 the subjects “child,” “arrows,” and “wings”).
(9)
The description of the iconic language of the visual artwork needs to be included (e.g., the relative position of objects and the structure in which they are organized).
(10)
At least a description of style should be included.
As Panofsky’s theory is considered a representative formalization of the iconological approach, we take the majority of the ontology’s terms from his theory. Therefore, we decided to populate the pre-glossary of terms (i.e., the relevant terms extracted by the CQs and their answers) contained in Table
3, point 7, by extracting the terms that are answering to CQs directly from the definition of his theory. The number following each word indicates its frequency in the selected article,
15 in which Panofsky’s theory is fully illustrated.
5 Ontology Design
The ICON ontology
16 was designed following the SAMOD [
39] and eXtreme Design [
42] methodologies. SAMOD is an agile methodology that focuses on the application of small iterative steps to model parts of an ontology. Each step is individually documented and combines motivating scenarios that derive from general domain descriptions with data-centric examples of descriptions formalized with the ontology. We reuse the SAMOD methodology for the main part of the design, as we adopt the iteration-like structure and its outputs. In fact, the design process was divided into four SAMOD iterations, each dedicated to a particular aspect of the ontology. Each iteration contains a motivating scenario, a glossary with the definition of specific terms, a self-contained ontology prototype that contains only classes and properties relative to the corresponding iteration (with no references to external ontologies), the alignments to external ontologies, the aligned prototype, a series of CQs formulated both in natural language and SPARQL (referring to the aligned prototype), and a Jupyter notebook that contains unity tests. All the CQs were tested on real interpretations by Panofsky [
37] expressed using the ontology schema. For a more detailed description of the test dataset, see Section
6.1. eXtreme Design is another agile methodology that divides the development of an ontology through iterations but focuses on the reuse of ODPs. In fact, the methodology tries to solve the “local problems” included in the so-called “local space,” or the modeling issues related to the specific ontology that is being developed, with the reuse of modeling patterns that come from the “solution space,” such as the ODP. We specifically adopted this methodology when dealing with the reuse of ODPs that were specialized in the context of our domain. The following paragraphs describe (i) each SAMOD iteration, (ii) the specialization of ODP to facilitate some modeling issues, and (iii) the refactoring of some classes and property through alignment to relevant ontologies.
5.1 First Design Iteration: Recognitions
As explained in Section
2, works of art can be analyzed through different layers of interpretations that depend on recognitions. A recognition, in the context of this ontology, is an interpretation act made by an agent (or interpreter, which can be a biological or electronic being) that links works of art to something related to their content. From a conceptual perspective, it is a mental entity reflecting the agent’s subjective point of view. From a technical viewpoint, it is an n-ary predicate that cannot be modeled using OWL due to expressivity limitations; therefore, it was turned into an n-ary relationship class.
17 Coherent recognitions on the same artwork are collected and documented by interpretation descriptions (requirement 8, Section
4).
18 In this iteration, we conceptualize the elements that revolve around recognitions. From the n-ary relationship class
icon:Recognition, several properties were designed (or reused from existing ontologies) to link it to its interpreter(s) (or agents), the artwork that is being interpreted, supporting sources for the recognitions. In particular, the
aboutWorkOfArt property links the recognition to the artwork (
Artwork class). Then, the
dul:includesAgent property (from DOLCE [
23]) links the recognition to the agent who performed it (requirement 4, Section
4). The class
InterpretationDescription is linked to (one or many)
Recognition class(es) that comply with it through several properties according to the type of the recognition, namely
isCompliantWithPreiconographicalRecognition for pre-iconographical recognitions and formal motif recognitions,
19 isCompliantWithIconographicalRecognition for iconographical recognitions, and
isCompliantWithIconologicalRecognition for iconological recognitions. The CiTO [
51] properties
cito:citesForInformation and
cito:citeAsEvidence can be linked to a
icon:Recognition class to provide sources or other information that support a recognition (requirement 5, Section
4). Finally, a recognition can also be used to support further recognitions made on the same artwork or another one. For example, Panofsky recognizes that the figure of Chastity sculpted by Giovanni Pisano on the Pulpit of Pisa cathedral is represented with the same appearance of the nude classical iconography of Venus Pudica (formal motif recognition).
20 This interpretation provides support to the third-level recognition of the characteristics of the Proto-Renaissance movement in the cultural context of the Medieval Tuscany [
37, p. 157]. To express this using our ontology, the property
cito:givesSupportTo can link the supporting recognition to another one (requirement 7, Section
4). These elements are also the object of interest of the general CQs (see Table
3, Q0.1–Q0.5).
Depending on the level of interpretation presented in Table
1, four
Recognition subclasses have been defined:
•
PreiconographicalRecognition (level 1)
•
FormalMotifRecognition (level 1)
•
IconographicalRecognition (level 2)
•
IconologicalRecognition (level 3)
Recognitions at each level of interpretation may be based on the results of the recognition at one of the previous levels. Therefore, they can be linked together but ultimately are modeled as independent of one another. This choice is made since (i) the describer may not have available the descriptions of the lower level(s), (ii) the corresponding subjects in the other levels may not be relevant for the recognition, and (iii) it may be possible that a level 3 recognition (i.e., an
IconologicalRecognition) is linked to level 1 subjects rather than level 2 ones (e.g., iconological interpretations of a landscape painting, which may not have level 2 subjects [
36]).
These classes and their specific usage will be further described in the following sections. Figure
1 shows a rendering of the classes and properties of this iteration.
5.2 Second Design Iteration: Pre-Iconographical Recognitions (Level 1)
In this iteration, we model the recognitions that happen on a pre-iconographical level. In this level, an interpreter recognizes artistic motifs present in the artwork, and associates to them (i) natural objects (a tree, a man, a sword) without identifying specific individuals from those classes (tree of life, Saint Joseph, Excalibur) that are recognized in level 2 (Section
5.3), (ii) in the form of expressional meanings
21 (emotions of the depicted elements), (iii) qualities about these elements (size, color, positions), and (iv) performed actions (see Table
1 in Section
2). Assuming that the agent doing the interpretation act might also be a computer, as in the case of the results of object detection through computer vision, we give the possibility to express coordinates of the portion of the image of the artworks where these elements are detected. Furthermore, these coordinates can be expressed using IIIF URIs [
53] that point to a specific portion of the work of art. A series of artistic motifs can be grouped together in a composition that can have a compositional structure
22 (e.g., pyramidal). Additionally, an interpreter might recognize similarities between artistic motifs present in a work of art with other artistic motifs of another work of art, recognizing a prototypical artistic motif or composition that is reused in another artwork. For example, the level 1 description of Pisano’s figure of Chastity cited earlier is linked through a formal motif recognition to the level 1 description of Venus Pudica, from which its appearance is derived (i.e., a nude woman covering herself with her arms). Artistic motifs and compositions are linked to the class
PreiconographicalRecognition respectively through the properties
recognizedArtisticMotif and
recognizedComposition. Only one artistic motif or composition can be linked to a recognition. Compositions are linked to the artistic motifs that take part in them through the
hasPart property. If the artistic motif refers to a natural object or action with a factual meaning, it is linked to the classes
NaturalElement or
Action through the property
hasFactualMeaning. Otherwise, if what is recognized in the artistic motif is an expressional meaning, the property that links it to expressional meanings is
hasExpressionalMeaning. If actions, expressional meanings, or natural elements have some specific quality that needs to be highlighted, from the artistic motif the qualities are expressed with the DOLCE
hasQuality property. When the pre-iconographical recognition is performed by a computer with an object detection algorithm, or when an IIIF URI is provided, it is possible to associate not only the detected objects but also the coordinates of the image in which they are found. Coordinates of the detected object can be expressed through the data property
hasRegionDescription that has the
ArtisticMotif or
Composition classes as the domain. As mentioned earlier, the use of IIIF URIs for the format of this data property is also welcomed. The
FormalMotifRecognition class links the prototypical motif to the copied motif respectively using the
hasPrototypicalMotif and
hasCopiedMotif properties. Finally, all the coherent formal motif recognitions and pre-iconographical recognitions that take part in an interpretation about a work of art can be linked to an
InterpretationDescription class, through the property
preiconographicallyCompliesWith. Figure
2 shows a graphical rendering of the classes and properties used in this interpretation level.
5.3 Third Design Iteration: Iconographical Recognitions (Level 2)
In this third iteration, we focus on the Panofsky’s second level of art interpretation: the iconographical interpretation. In this level, the interpreter recognizes images and invenzioni
23 in an artwork. An image represents the subject depicted as a manifestation in the specific artwork taken into account. It is then linked to second-level subjects, which are characters, places, events, named objects,
24 symbols, and personifications, identifying iconographies from an abstract and general point of view. This distinction between the general subject level (i.e., characters, symbols) and the artwork-specific one (image) is functional to identify the variants of a subject in relation to the specific context (i.e., Thor as represented in a specific painting may differ from its common one). An invenzione, instead, is the subject matter represented by the combination of general subjects linked to the single images recognized.
25 For example, in an artwork, you might recognize three images: the first refers to the general subject of Mary, the second refers to the general subject of Angel Gabriel, and the third refers to the general subject of the Holy Dove. The combination between the general subject of Mary, Angel Gabriel, and the Holy Dove is the Annunciation, which, in our ontology terms, would be considered the invenzione. The same invenzione could be present in multiple artworks, but each artwork maintains its uniqueness by having different images. The classes
Story and
Allegory are subclasses of the class
Invenzione. Stories are more likely to contain characters, named objects, places, and events, whereas allegories are more likely to contain symbols and personifications. We give the possibility to express symbols as just symbolic meanings recognized, or, for a more thorough description, as Simulations (see Section
5.5). The classes
Image and
Invenzione are linked to the class
IconographicalRecognition through the respective properties
recognizedImage and
recognizedInvenzione (one image or invenzione per recognition). The artistic motif belonging to a pre-iconographical level that refers to the recognition of an image can be linked to it with the property
refersToArtisticMotif (i.e., the recognition of the image that represents Mary Magdalene can be linked to the artistic motif that has the factual meaning of woman). This link is important to ensure that the connection between pre-iconographical elements and the respective iconographical subjects is preserved. If the artistic motif is the principal element that enabled a recognition of an image, then it can be linked to that image through the property
hasRecAttribute (i.e., the recognition identifying Cupid has recognizing attributes the artistic motifs linked respectively to “wings” and “arrows”). Images are linked to the general subject portrayed through specific properties according to the subject class. The property
hasCharacter links an image to the class
Character, likewise
hasEvent refers to the class
Event,
hasPlace refers to the class
Place,
hasNamedObject refers to
NamedObject,
hasSymbol refers to
Symbol, and finally
hasPersonification refers to
Personification. The cited ICON classes represent second-level subjects represented in the fictional representational space, therefore including both real and fictional, non-existent subjects (e.g., Medusa, the Greek mythological character appearing in various media), in compliance with the modeling of subjects in narratology [
12,
14]. An invenzione is linked to the elements that compose it through the property
composedOf. Finally, multiple iconographic recognitions that take part in an interpretation of an artwork are linked to the interpretation using the
iconographicallyCompliesWith property. Figure
3 shows the classes and properties relative to this level of recognition.
5.4 Fourth Design Iteration: Iconological Recognitions (Level 3)
Iconological interpretations (third level) focus on the recognitions of intrinsic meanings.
26 An intrinsic meaning links the whole artwork or some parts of it to a cultural phenomenon and a concept that defines it. The
IconologicalRecogniton class is linked to the
IntrinsicMeaning class
27 through the property
recognizedIntrinsicMeaning. From there, the n-ary class
IntrinsicMeaning can be linked to a specific composition, image, or artistic motif that can be the focus of the intrinsic meaning through the properties
hasComposition,
hasImage, and
hasArtisticMotif. Then, it is linked to the expressed concept through the property
recognizedConcept. For the range of this property, we reuse the DOLCE class
SocialObject because there was no need to create an ad hoc class for this element.
28 Additionally, since an intrinsic meaning can also reflect some cultural phenomena, it is linked to the class
CulturalPhenomenon through the property
recognizedCulturalPhenomenon. Currently,
CulturalPhenomenon has four subclasses, which specify the type of cultural phenomenon, namely
Attitude,
Belief,
CulturalValue, and
Tendency. These terms are taken from Panofsky’s vocabulary in the description of the third level of artistic interpretation.
29 Finally, all the iconological recognitions that take part in an interpretation made on an artwork are linked to it with the property
iconologicallyCompliesWith. A graphical rendering of this fourth iteration, representing the third level of the interpretation, can be found in Figure
4.
5.5 Refactoring: Reuse and Alignment to Relevant Ontologies and Ontology Design Patterns
To promote ontology interoperability and reusability, we connect to several external ontologies through means of alignments and reuse. We present our alignments and reuse by following guidelines proposed by the state of the art [
9,
35]. Our ontology selection for reuse and alignment was guided by different principles: (i) standardization for CIDOC-CRM [
5] and FRBRoo [
46] because they are considered standard frameworks in the domain, (ii) cognitive and formal analysis for the choice of DOLCE foundational ontology [
6,
23] in its OWL version (DOLCE Zero), Simulation Ontology [
50], VIR [
8], HiCO [
18], and CiTO [
51], as all of them offer design solutions to the CQs defined from the requirements in Section
4.
Due to the complexity of the field, the number of ontologies to be reused, and the heterogeneous domains from which they come, we adopted a hybrid reuse approach [
9], which, depending on the specific cases explained in the following, considers either reusing directly the classes and properties of the aforementioned ontologies (either by importing the whole ontology or parts of it), or (indirect reuse) using them as fully extensional ontology patterns, or just as intensional patterns.
Extensional reuse happens when classes or properties of an ontology O1 are logically aligned to an external ontology O2, which we want to reuse with its full-fledged semantics, because it is compatible, desirable, or necessary. For example, if we extensionally align a O1 class Organisation to a O2 class dul:SocialObject, we intend to inherit the semantics of DOLCE’s social objects (e.g., that they are not physical).
On the contrary, we use parts of an external ontology O3 as purely intensional constructs when we want a limited interoperability, which does not include accepting in O1 all the semantics provided in O3, because it may be partly incompatible. For example, we may intensionally align a O1 class Image to a O3 class crm:E36_Visual_Item because we might not want to inherit the axiom stating that crm:E36_Visual_Item is a subclass of crm:E89_Propositional_Object.
To implement this distinction, indirect reuse is designed using different mapping properties, according to the semantics they provide, and its impact into the resulting reasoning. We have used RDFS (rdfs:subPropertyOf, rdfs:subClassOf) and OWL (owl:equivalentTo) logical properties when we want the alignments to provide first-order extension to ICON schema and data, and we have used SKOS skos:broadMatch, skos:related, and skos:closeMatch for purely intensional mapping, which can be used at query time to integrate data represented with ontologies that may harm the logical integrity of ICON knowledge.
Among the reused ontologies, we have used an intensional (or “terminological”) mapping for CIDOC, VIR, and FRBRoo, because we have noticed potential problems when reasoning is jointly made with both the axioms from ICON, and from those ontologies. For example, a full extensional alignment of the class icon:Image as rdfs:subclassOf crm:E36_Visual_Item would make an automated reasoner infer that icon:Image rdfs:subclassOf crm:E89_PropositionalObject, which is not defendable, since propositional entities typically exclude visual, musical, or other information modalities. In other words, CIDOC contains here a debatable assumption, which should be ignored when reusing data that use CIDOC as their schema. Now, if we use a purely intensional mapping: icon:Image skos:broadMatch crm:E36_Visual_Item, we make a commitment that can be discussed, and the triple can be used to make SPARQL-based data integration, but we will not get the inference that images are propositions.
In this section, we give a thematic overview of the classes and relations reused for satisfying a specific task, and we refer to the documentation (see Section
7) for further details on the single alignments. Table
4 shows the direct reuse of external classes and properties in ICON, and Table
5 shows the indirect alignments.
5.5.1 Recognitions as Situations.
According to the guidelines of eXtreme Design [
42], we defined our local problem (in our local space) as the expression of recognitions through n-ary relationship classes due to the inability of expressing n-ary predicates in OWL. As explained in the previous paragraphs, our conceptualization of the
icon:Recognition class required a good deal of contextual information (e.g., the agent performing it; what is recognized in the form of the first, second, or third level of interpretation entities; the artwork). We have chosen the Situation ODP
30 as a solution because it was designed to solve modeling issues regarding multiple contextual information connected to the same class in the form of n-ary relationships. The Situation ODP is reused via the import of DOLCE Ultralite.
31 The n-ary relationship ODP is specialized by our
icon:Recognition class, by making it a subclass of
dul:Situation.
5.5.2 Interpretations as Descriptions.
The types of recognitions that we have presented are formalized as situations. In the Descriptions and Situations pattern
32 that is also formalized in DOLCE Ultralite and DOLCE Zero, situations are loosely associated with
descriptions (i.e., intensional entities that are used criteria for a situation to occur). The pattern is used in most domains: in medicine, a pathological situation depends on the diseases or syndromes that are used to interpret it, and which can have different probabilities to correspond to the actual situation; in law, different norms may apply to a same legal case; and in an everyday situation, an observer may interpret it differently according to her perspective, culture, or intention. In the iconographical and iconological domain, as also applied in the ArCo ontology network [
10,
11], all recognitions and high-level interpretations are based on perception criteria, which make a rationale emerge, and eventually motivate a particular interpretation with respect to others. A description is therefore a conceptual entity, constituted by parameters, roles, tasks, and so forth [
24], which is satisfied by a situation when it involves entities that are classified by one of the parameters, roles, tasks, and so forth that constitute a description. For example, the interpretation of a painting (Named
A) such as “in this painting, there is a lion which symbolizes courage” is compliant with (i) a pre-iconographical recognition (recognizing an artistic motif as a carrier of the factual meaning of a lion) and (ii) an iconographical recognition (recognizing the image of the lion as the simulation of lion-courage). These recognitions would involve the recognizer, a source, and the time period, as well as (potentially) additional iconographical aspects. Hence, we formalize this complex relation in terms of compliance:
InterpretationDescriptionPaintingA isCompliantWithPreiconographicalRecognition LionRecognitionInA and
isCompliantWithIconographicalRecognition LionCourageRecognitionInA. The property
isCompliantWithPreiconographicalRecognition is made a sub-property of
dul:isSatisfiedBy that links a
dul:Description (our
InterpretationDescription is subsumed under description) to one or more
dul:Situation (our
Recognition is subsumed under situation).
5.5.3 Describing Artwork Content.
Since CIDOC CRM offers a way of describing the content of visual elements (
crm:E36_Visual_Item,
crm:P138_represents,
crm:E1_Entity), we modeled the more specific elements recognized in each level of interpretation following this modeling principle as a guideline, and aligning our classes to CIDOC’s ones through SKOS relations. As illustrated by Figure
5, all the classes representing the general subject as represented in the contest of the artwork (i.e., Artistic Motif, Composition, Image, IntrinsicMeaning) are a
skos:broadMatch of
crm:E36. Furthermore, the recognized subjects at every level are a
skos:broadMatch of
crm:E1_Entity. In this way, the patterns linking the visual elements recognized in each level and the general subject can be seen as a specification of
crm:P138. The identification of the artwork at the abstract level (Artwork,
skos:broadMatch of
crm:E36_Visual_Item) is intended to make the ontology compliant with the CIDOC-CRM modeling of cultural objects, whereas the alignment of Artwork with
dul:InformationObject is motivated by the DOLCE conceptualization of Information Object that fits with our Artwork definition.
5.5.4 Interpretation Details.
The class
Recognition has been aligned with classes from HiCO, CIDOC-CRM, and DOLCE, as shown in Figure
1. The class
hico:InterpretationAct is intended to represent the context in which a recognition
33 is made—that is, furnishing more information about the recognition to validate the claim. The recognition such represented can be further specified by
hico:interpretationType and
hico:Interpretation- Criterion. For its purpose and formal structure,
icon:Recognition was made a subclass of it. Since also the purpose expressed by
crm:E13_Attribute_Assignment is of documenting the context in which an assertion about a cultural object was made, it is a
skos:broadMatch of Recognition, since Recognition is more specific than the more generic concept expressed by
crm:E_13. Furthermore,
crm:E13 is practically used as an n-ary relationship class linking two individuals through ancillary properties,
crm:P140,
crm:P141, identifying respectively the element to which the assignment is made and the assigned one. Therefore, when this logical structure is respected, the respective properties in the subclass of
icon:Recognition are aligned to
crm:P140 and
crm:P141 through
skos:broadMatch, and respectively,
RecognizedArtisticMotif or
RecognizedComposition at level 1, and
RefersToArtisticMotif and
RecognizedImage or
RecognizedInvenzione at level 2.
By the alignment with
hico:InterpretationAct, and
dul: Situation, the ontology not only enhances interoperability but also inherits a variety of means for expressing further detail about each recognition act at each level. For example, the possibility to express an agent using
dul:Agent which includes both humans and computers, the time of the recognition using the
includesTime property of dolce, the interpretation criterion
34 InterpretationDescription class, and type (HiCO) allows the user to fully document the recognition acts, giving a comprehensive representation of the subjectivity of the recognition itself.
The Motif Recognition is developed as a specialization of the VIR property
K4i_has_visual_prototype, documenting the use of a visual prototype for an image, enriching the latter by giving the possibility to add further details about the interpretation and to highlight the direct correspondence between the portions of the copying and copied artworks. For example, the derivation of the visual arrangement of the relief
Allegory of salvation from the Roman relief depicting
Hercules and the Caledonian Boar described by Panofsky and Saxl [
38, p. 228; Figure 4–5, p. 231] can be further described by recognizing that the deer in the former is derived from the boar in the latter, and so on. Our property
icon:hasPrototypicalMotif was aligned with
skos:broadMatch to
K4_is_visual_prototype_of.
5.5.5 Subjects.
As it is the closer definition of artistic subject intended as an object represented by an artwork, we align all the subjects of the ontology to the ArCo’s class
arco:Subject. Specifically, we indirectly reuse
arco:Subject by subsuming
icon:Place,
icon:NamedObject,
icon:Character,
icon:Event,
icon:Symbol,
icon:Personification,
icon:Action,
icon:NaturalElement,
icon:ExpressionalQuality,
icon:CulturalPhenomenon,
icon:Invenzione,
dul:SocialObject to it. In doing so, we also propose a new way of attributing a subject to a work of art compared to ArCo. In fact, whereas ArCo directly links a subject to the physical representation of the work of art, we link it to an interpretation made on the visual representation of what is in a physical work of art. By reusing the class
arco:Subject and not its properties, which consider the physical artwork as the domain, we also avoid possible logical inconsistencies between ArCo’s description of physical artifacts and ICON description of visual items. In contrast, the representation of the subjects as manifested in the artwork (i.e., Artistic Motifs, Compositions, Images, and Intrinsic Meanings) are a subclass of
icon:VisualSubject, which is disjoint with
arco:Subject to underline their different nature and role. Table
6 displays the division between subjects and visual subjects according to the different iconographic and iconological levels.
5.5.6 Symbols.
In an artistic interpretation, an interpreter might recognize a symbol of a specific cultural context in an artwork. For the modeling of symbols, we reuse the entire Simulation Ontology [
50]. This ontology, designed to conceptualize cultural symbols, uses the n-ary
sim:Simulation class to link together a symbol, expressed by the class
sim:Simulacrum, its symbolic meaning, expressed by the
sim:RealityCounterpart class, the cultural context in which the symbol denotes the symbolic meaning (
sim:Context), and the source of the claim (
sim:Source). We aligned our class
icon:Symbol to the
sim:Simulation class to allow the expression of symbolic meanings using the Simulation Ontology structure.
5.5.7 Expression of Style.
The expression of style is an important feature related to iconographical and iconological studies (requirement 10, Section
4). Knowing the history of styles is, according to Panofsky [
37], a fundamental requirement for the correct interpretation of level 1 objects. Furthermore, as is evident, among the others, from Warburg’s studies on
Pathosformeln and
Nachleben der Antike, forms of style are a subject of interest in iconology. Therefore, we reuse CIDOC-CRM to model it according to the solution adopted by linked.art project,
35 using the structure
where the last object is the Getty AAT vocabulary term defining style. Although the property’s domain is
crm:E1_Entity, it is suggested to use it with
crm:E36_Visual_Item, in compliance with linked.art directions. Even if we do not express
icon:VisualSubject and
icon:Artwork as direct subclasses of
crm:E36, it is possible to reuse this pattern since ICON’s classes are not disjoint with those of CIDOCs. Therefore, we reuse this existing solution to model requirement 10 of Section
4. In this way, both the artwork itself and every portion of the image identified at each level can have its own style specification declared.
5.5.8 Citations, Sources, Evidence.
As shown in Figure
1, the CiTO ontology is directly reused to represent the source (
cito:citesForInformation) from which the Recognition is extracted, the evidence (
cito:citesAs- Evidence) on which it is based, and the supporting (
cito:givesSupportTo) between acts of recognition. This representation is fundamental to encourage a documented description of the recognition, its reference, and support.
5.6 Logical Constraints in Artistic Interpretations
Artistic interpretations are, by definition, subjective. Nevertheless, the characteristics and relationships that surround recognized elements can be subject to logical constraints. Each recognition is made on exactly one artwork, involves exactly one agent, and different recognitions require adequate elements. A pre-iconographical recognition targets as recognized elements, either artistic motifs or compositions. A formal motif recognition instead deals with a prototypical motif and a copied motif, and both of them can be either an artistic motif or a composition. Then, an iconographical recognition recognizes exactly one image or one invenzione. Finally, an iconological recognition refers to an intrinsic meaning. For what it concerns the elements that are recognized, the recognition of an artistic motif, in Panofsky terms (and thus in our ontology) implies that the interpreter associates either a natural or expressional meaning to a portion of the artwork. At the same time, the recognition of an image implies the presence of either a character, event, named object, symbol, personification, or specific place in the artwork. Furthermore, images, artistic motifs, and intrinsic meanings cannot be instantiated without having a recognition that addresses them. To ensure this, we added several restrictions. For example, images must be linked to exactly 1 recognition through the property isIconographicallyRecognizedBy. The difference between stories and allegories is that the former generally includes characters, places, events, and named objects, and the latter is more focused on symbols and personifications. Nevertheless, a story might contain symbols and an allegory may contain characters, so the logical restrictions in these cases are not very strict. An intrinsic meaning can refer to either a cultural phenomenon or a conceptual object.
An exemplification of some restrictions on main classes through OWL axioms follows, formalized in Manchester Syntax
36:
•
InterpretationDescription:
•
PreiconographicalRecognition:
•
IconographicalRecognition:
This list contains only the directly created axioms. The restrictions inherited by the alignment to external ontologies are available in the documentation.
378 Conclusion, Limitations, and Future Work
In this article, we presented ICON, an ontology dedicated to the conceptualization of artistic interpretations designed by formalizing the content of several interpretation theories. In line with the principles of reuse and interoperability of the Semantic Web, the ontology reuses (directly and indirectly) several existing ontologies. It is released alongside a documentation that guides potential users in formalizing art interpretations using our model. ICON was evaluated on its extraction potential, syntax, metadata, and FAIRness. Moreover, its granularity was highlighted through a comparison to current ontologies on their respective serialization of the same interpretation. The results show how our work elevates the potential of expression of artistic interpretations in the context of Semantic Web by providing a granularity level that was not reached by other ontologies on this topic. Finally, its effectiveness of describing the iconographical and iconological complex domain is confirmed by the results of the proposed CQs, formalized in SPARQL queries, ran on a test dataset containing artistic interpretations.
Although ICON provides users the option to describe a plethora of concepts related to art interpretation, the ontology presented here is still a work in progress. Future work will be directed to a more thorough conceptualization of cultural phenomena, personifications, and allegories in the same way that cultural symbols were defined by the Simulation Ontology [
50]. Furthermore, modeling artistic interpretations required the use of several agglomerated n-ary relationship classes, drawing a long path from the artwork to its meaning. On the one hand, this modeling offers a very high granularity level, as interpretations can be dissected into recognitions representing different levels (pre-iconographic, iconographic, iconological), allowing the potential extraction of very specific information as shown by the testing of the CQs. On the other hand, potential users might only be interested into separating what is depicted in an artwork into the aforementioned levels, without having to describe the whole process of interpretation. Future work will be devoted to the creation of property chains that can be used to declare that an artwork represents elements of a pre-iconographic, iconographic, or iconological level. Both solutions (series of n-ary classes) and the property chains could be adopted, in a knowledge graph, at the same time, as in ArCo [
10], which uses property chains as shortcuts, and n-ary relationship classes to describe the same information with more granularity. As for the alignments, although, as Table
5 shows, there are more than 50 alignments to external ontologies, we decided to keep a conservative approach for this version of ICON. Future versions will foster the interoperability of ICON even more by (i) finding other ontologies that cover similar aspects to increase alignments (e.g., [
14]), and (ii) by specializing the alignments with the currently aligned ontologies by finding more specific classes and properties. The
icon:Invenzione class, for instance, is aligned to a very generic
crm:E1_CRM_Entity. Future versions of CIDOC might introduce new classes that are closely related to ours, leading to a more specific alignment.
On another note, the current debate of interpretation provenance and the difference between “asserting and expressing” [
17] in a Semantic Web context using recently introduced technologies of RDF-star
46 and conjectures [
17] could use a dataset of contrasting interpretations described with our ontology as a case study.
In conclusion, by combining our model with the external ontologies to which it is aligned, it is now possible to thoroughly describe artworks both in their standard metadata (i.e., creator, date of creation, dimensions, place of creation) and on their content side based on interpretations. The result is that, finally, these two evenly important types of information are now treated equally and can both exploit the potentiality offered by the Semantic Web. This contribution opens up the possibility to link artworks in their content level, allowing content-based research questions in the field of art history to cross into the linked open data realm.