Author and biographic data

The author name is a required metadata in the original database: it needs to be properly represented both when it is expressed as a plain name and both when the authoriality assignment is complex . Thus, the author's representation was a pivotal point in the conceptual model design.

Each author is represented as an entity belonging to the class efrbroo:F10_Person. The relation between the expression creation URI and the author URI is expressed by the property ecrm:P14_carried_out_by. Usually, the record of each author string in the source database embeds biographic information, namely the place and date of birth and the place and date of death (e.g. Passerotti Bartolomeo (Bologna, 1529 - 1592).

Following the CIDOC-CRM event-centric approach to modelling, the author URI is connected to two URIs representing the event of the birth and the event of the death of the author. Then, each event is annotated with a machine-readable place and time of occurrence. In particular, the author URI is related to the birth URI having class crm:E67_Birth via the property crm:P98i_was_born. The birth URI is linked via the property crm:P4_has_time_span to the “Bdate” URI, of class crm:E52_Time-Span. The “Bdate”, in order to be machine readable as a proper period of time, is better defined with a starting point and an ending point: through the property crm:P82a_begin_of_the_begin as a Literal date object (in the xsd:date format) and via the property crm:P82b_end_of_the_end as a second Literal date object. The “Bplace” URI represent the place of the event, as expressed by the property crm:P7_took_place_at. This Bplace entity has class crm:E53_Place and is defined more explicitly via the label (RDFS:label property) expressed as a Literal string object. The same detailed model applies to the death event, represented by a death URI of class crm:E69_Death linked with the author URI through the property crm:P100i_died_in.

Dating

The expression creation is an event that occurs (property ecrm:P86_falls_within) in a period of time, represented with the “dating” URI, having class ecrm:E52_Time-Span. The model here follows as before: the two points in time are defined by a first machine-readable date object (crm:P82a_begin_of_the_begin property) and a second similar date object (crm:P82b_end_of_the_end property); and the URI is also labelled (RDFS:label property) with a Literal string object.

In the original database, dating information is not mandatory and is expressed in different ways: as an exact date, as a period delimited by two dates or as a more general century indication. While it is important to convert all the dates in a machine-readable format - thus with one or two precise points in time in the xsd:date format - these different expressions are crucial to explicitly mark different kinds of fuzziness in chronological data (e.g. “1575” is a simple date, “1740-1815” is a period, “prima metà XVII sec.” is a century ). This is the reason why a “dating type” entity has been added; three URIs can be assigned consequently: dhlod:data_semplice for exact dates, dhlod:data_periodo for a delimited period of years, and dhlod:data_secolo for centuries.

Technique

Despite its importance, information about technique is unavailable for all items in the archival collection, but only for the categories “Dipinti” and “Disegni”. In the model the technique is an entity of class crm:E55_Type and is directly linked to the expression creation entity to the property ecrm:P32_used_general_technique.

The alignment is represented by the property owl:sameAs, which connects the technique URI with the URI of the term from the AAT taxonomy.

Publisher and place of publication

The publication and the relative metadata are described in the model following the CIDOC-CRM approach. As shown in figure 4.6, the expression is physically embodied (property efrbroo:R41_has_representative_manifestation_product_type) by a manifestation product type (automatically generated URI of class efrbroo:F3_manifestation_product_type)82. The latter is an object that carries (property efrbroo:CLR6_should_carry) the content as graphically edited and organised by the publisher: this concept is expressed by the publication expression URI, an entity of class efrbroo:F24_publication_expression. The publisher is thus the person responsible (ecrm:P14_carried_out_by property) for the publication expression. The publisher URI is generally classified as actor (crm:E39_Actor) and specified by the property ecrm:P14.1_in_the_role_of, which assigns the role of publisher (role URI) among other possible types of roles (entity of class ecrm:E55_Type). It is also made available the string expressing the publisher, through the RDFS:label property.

Moreover, when available, the data about the place of publication are expressed as the place in which the publisher resides (publisher URI, ecrm:P74_has_current_or_former_residence, place of publication URI; and place of publication, is a object of class ecrm:E53_Place, having a string as a label (RDFS:label property).

Category

The archival items have been classified by the Collections Taxonomy. In the conceptual model, a type assignment (property ecrm:P2_has_type) draws a direct relationship between the intellectual work (expression URI) and the category term from the Collections Taxonomy.

Bibliography and Keywords

The item expression is about (property dct:subject) some main themes, that are he expressed as a Literal object (the keywords string). Additionally, some bibliographic references (string) describe (ecrm:P67_refers_to) the content of an item (efrbroo:F2_Expression).

Identification: title, item id and inventory number

The model comprehends three ways for item identification:

1. the title, which is here represented as a Literal string object and is linked to the manifestation URI via the property dct:title.
2. the inventory number (unique code assigned to the item in order to catalogue it and retrieve it in the physical archive, represented as instance of class ecrm:E42_Identifier)
3. the item identifier (alphanumeric code, unique for each item, but automatically generated by the database management software and referring to database item, represented as an instance of the class efrbroo:F13_Identifier) are both related to the manifestation URI through the property ecrm:p1_is_identified_by as shown in figure 4.9.

Physical description: dimension, pagination, ligature, support, iconography

The physical description describes a plethora of archival objects of different types that carry a variety of physical features.

The dimension is common to bibliographical and artistic objects and is represented in the data model as in CIDOC CRM: a manifestation URI has dimension (ecrm:P43_has_dimension), a URI belonging to crm:E54_Dimension. Each dimension is connected with the measurement unit URI (ecrm:E58_Measurement_Unit), and the actual dimension value (ecrm:E60_Number).

The further descriptions of the manifestation regard its vessel and the manufacturing of the object itself: the manifestation URI is related via the property ecrm:P45_consist_of to the physical man made thing URI (class ecrm:E24_Physical_Man-Made_Thing), that “comprises all the persistent physical items that are purposely created by human activity” (ICOM/CIDOC Documentation Standards Group, 2020). Artwork iconography is specifically represented with the triple “manifestation URI ecrm:P62_depicts iconography (string)”.

For the manuscript description, some particular metadata are required: pagination, ligature and support.

The pagination URI (of class ecrm:E18_Physical_Thing) reports how many pages or folios the physical man-made thing is composed of (ecrm:P46_is_composed_of property) and it is also annotated by a Literal string object.

The ligature URI accounts for the material and the technique used for the book ligature (ecrm:P19_bears_feature property): it is an entity of class ecrm:E29_Physical_Feature with its rdfs:label.

Finally, the support URI of class ecrm:E57_Material describes the material of the writing support with the property ecrm:P45_consist_of.

Ownership: artwork owner or artwork origin

The concept of ownership is reported in the model with two fields: “artwork origin” and “artwork owner”. So, in the model the concept of the former owner is expressed by the relation ecrm:P51_has_former_or_current_owner between the manifestation URI and a former owner URI belonging to class ecrm:E39_Actor. The former owner URI can be identified (ecrm:P131_is_identified_by property) by both the artwork origin and the artwork owner data values, expressed as URIs of class ecrm:E82_Actor_Appellation.

Other information: notes, linked resources and related artwork

References of other institutions’ digital catalogues describing the same artwork are represented with the property rdfs:seeAlso. Relations between objects inside the collection are generically represented through the property skos:related. Finally, the commentary notes by the cataloguer are represented as Literals via the property ecrm:P3_has_note.

Workshop or school

Art historians use the term “Workshop of…” or “School of...” or “Circle of...” when the responsible for the artwork creation is not a single artist but a group of artists, or it is clear that there is a recognisable style thus the author is a member of the school. These groups can be indicated through their common master or via the common city or area of action.

As in LinkedArt, the object of the property ecrm:P14_carried_out_by is a “bottega URI” an entity of class crm:E74_Group carrying the name of the group. The group responsible for the expression creation is then furtherly described: the group have been created through (property crm:P95i_formed_by) a formation action (URI of class crm:E66_Formation); the formation action is a gathering around either a master (property crm:P15_was_influenced_by), which is expressed by URI of class efrbroo:F10_Person, or a place (URI of class crm:E53_Place) in which a style is diffused (property crm:P7_took_place_at).

Anonymous

In the source dataset, unknown authors are annotated as “anonymous”, while in the DHLOD conceptual model RDF object of type “Blank Node” is implied as a high-functioning substitute for the term “Anonymous”. A blank node is defined as “simply indicating the existence of a thing, without using an IRI to identify any particular thing”(W3C 2014). The model represents the blank node as a unique entity of class efrbroo:F10_Person and labelled with the string “Blank Node Anonimo”.

When available, other data are added to better define the anonymous author in question. For example, if the author is drawing inspiration from a known author the blank node is linked to a known author URI (class efrbroo:F10_Person) via the property crm:P15_was_influenced_by.

Or for example, an anonymous author can be defined by the place where he was active - in this case, the blank node is related to the place of activity URI (crm:E53_Place) via the gndo:PlaceOfActivity property - or by the period of time when he was active - represented by a blank node, the property gndo:periodOfActivity and the century URI (class crm:E52_Time-Span).

Hypothetical author assignment (uncertain attribution)

The last case of uncertainty in the metadata describing the responsibility for an expression creation is the so-called hypothetical author assignment: the responsibility is not clearly stated, but the cataloguer or the art historian propose a guess about the hypothetical author. The fact that the data is carrying hypothetical information and thus not a certain one is expressed in the original database by adding in the author string a question mark or the abbreviation “attr.”, meaning “attributed” to the attributed author.

The model instead expresses the fact that the data is a hypothesis in an explicit fashion for both the user and the machine (here the database management system). Different conceptual models propose different solutions for the explicit expression of attributions. In the ideation of the conceptual model presented here, we chose to adopt the practice suggested by the CIDOC-CRM model.

The uncertain information is clustered and identified with a unique URL, formed by “attrAss” plus the item identifier, of class crm:E13_Attribute_Assignment. The hypothesis is then expressed with three statements: (1) the attribution subject (the expression creation URI in our case) declaration via the property crm:P140_assigned_attribute_to; (2) the attributed relation statement (in our case ecrm:P14_carried_out_by) via the property crm:P177_assigned_property_of_type; and (3) the attribution object identification through the property crm:P141_assigned. The object will be in our case the attributed author name, represented by a simple author URI, of class efrbroo:F10_Person. It will be recognised as an attributed author thanks to the conceptual structure, where it is not directly connected to the expression creation via the property ecrm:P14_carried_out_by, but through the relation with the attribute assignment URI.

In the original database, the uncertainty is noted, as previously explained, by adding a textual signal to the author string. The different textual signals are not useful for identifying an attribution in the model but they still bear information about the hypothesis formulation. The latter are preserved in the model using the attribution type URI, which is attached to the attribution assignment URI via the property ecrm:P2_has_type, and it can have two values: (1) “uncertain attribution”, if the textual signal was a question mark, meaning that the hypothesis is still a guess, or “attribution”, if the textual signal is the substring “attr.”, then the hypothesis is likely to be widespread and verified among the critics.

Uncertain Dates

Uncertain dates are modelled following the same structure used for the hypothetical author assignment, described in the previous paragraph.The uncertain information is represented by a unique URI of class crm:E13_Attribute_Assignment, composed with the prefix “DateAttrAss” and the item identifier. The attribution is illustrated through the three same statements: (1) the subject will be again the expression creation URI; (2) the relation will be this time the property ecrm:P86_falls_within; and (3) the object will be a date URI, treated as the other dates in the standard version of the model (having a label and a machine-readable expression of time delimiters).

The textual signals in this case are expressed as date qualifier URIs: the question mark signals an uncertain attribution (date qualifier value unknown); the “circa” substring (date qualifier value circa) enlarges the borders of the time-span; the “post” substring (date qualifier value post) and the “ante” substring (date qualifier value ante) define respectively a post quem term and an ante quem term, and finally all the other substrings are qualified with the value approximation.

Uncertain Titles

The titles deemed to be uncertain are represented in the model using the same structure described before, with some slight differences. As shown in figure 4.17, the URI of the attribution assignment is formed by the prefix “TitAttrAss”. The objects of the three statements about the attribution assignment are respectively the manifestation URI, the dct:title property and the Literal string object containing the attributed title.

Finally, the textual signal is here called Title attribution type and can assume two values: “uncertain_title” if it is a question mark and “attributed_title” for square parenthesis wrapping the string, which is a practice to identify information added by the editor.