The ontology's central design claim is that knowledge extracted from interpretive historical texts is not merely discovered but constructed — and that the knowledge graph must therefore model the construction process itself, not only its output. Every extracted entity and every mention annotation is traceable, via typed PROV-O activities, to the paragraph it came from, the software that produced it, and the prompt template or index resource that guided extraction.
The ontology is developed within the eXtreme Design (XD) methodology and has been validated iteratively against a set of 39 competency questions derived from domain expert interviews with art historians, digital humanities researchers, and data engineers.
Interactive visualization of the Viewsari ontology class hierarchy and relationships. Drag nodes to rearrange, scroll to zoom, click a node for details.
| Expression class | Manifestation class | Superclass |
|---|---|---|
viewsari:volume | viewsari:volume_web_representation | fabio:WebSite |
viewsari:biography | viewsari:biography_web_representation | fabio:WebPage |
viewsari:page | viewsari:page_web_representation | fabio:WebPage |
doco:Paragraph — carries viewsari:hasText, viewsari:hasLengthInCharacters, viewsari:hasStartPage, and viewsari:hasEndPagedoco:TextChunk — a surface-form span within a paragraph, linked via oa:hasSelectordoco:TextPositionSelector — character-level oa:start / oa:end offsets| Class | Description |
|---|---|
viewsari:person | Historical individual from the Index of Names |
viewsari:artwork | Creative work mentioned in the text |
viewsari:location | Geographical place |
viewsari:organization | Guild, order, academy, or other institution |
viewsari:cooccurrence | Pairwise joint appearance of two entities in one paragraph |
| Class | Pipeline stage |
|---|---|
viewsari:named_entity_recognition | NER and coreference resolution over the Index of Names and paragraphs |
viewsari:cooccurrence_analysis | Pairwise co-occurrence extraction |
viewsari:entity_linking | LLM-guided candidate ranking and Wikidata linking |
| Property | Domain | Range | Description |
|---|---|---|---|
viewsari:hasStartPage | doco:Paragraph | viewsari:page_web_representation | First printed page of a paragraph |
viewsari:hasEndPage | doco:Paragraph | viewsari:page_web_representation | Last printed page of a paragraph |
viewsari:isBasedOn | viewsari:translation | viewsari:edition | Derivation of a translation from its source edition |
viewsari:inParagraph | viewsari:extracted_content | doco:Paragraph | Anchors a co-occurrence to its source paragraph |
viewsari:involves | viewsari:cooccurrence | viewsari:extracted_content | Links a co-occurrence to a participating entity (min. cardinality 2) |
| Property | Domain | Range | Description |
|---|---|---|---|
viewsari:hasText | doco:Paragraph | xsd:string | Full normalized paragraph text |
viewsari:hasLengthInCharacters | doco:Paragraph | xsd:nonNegativeInteger | Character count of the merged paragraph |
| Prefix | Namespace | Used for |
|---|---|---|
fabio | http://purl.org/spar/fabio/ | FRBR-aligned bibliographic types |
doco | http://purl.org/spar/doco/ | Document component types |
prov | http://www.w3.org/ns/prov# | Extraction provenance |
oa | http://www.w3.org/ns/oa# | Web Annotation (mentions, selectors) |
frbr | http://purl.org/vocab/frbr/core# | Part-of and embodiment relations |
dct | http://purl.org/dc/terms/ | Relation subproperties |
The ontology defines four named individuals that serve as the stable bibliographic root nodes of the knowledge graph.
| Individual | Type | Description |
|---|---|---|
viewsari:le_vite | fabio:Work | The abstract intellectual work by Vasari (wd:Q1645493) |
viewsari:le_vite_1568 | viewsari:edition | The 1568 Giunti edition |
viewsari:the_lives_1568 | viewsari:translation | The 1912 Du Vere English translation |
viewsari:the_lives_gutenberg_version | fabio:ManifestationCollection | The Project Gutenberg web publication of the Du Vere translation |
The full provenance chain from a raw paragraph to a canonical entity can be traversed
in SPARQL in at most two prov:wasGeneratedBy hops:
Every triple in the knowledge graph is traceable to a documented computational activity and ultimately to a specific paragraph in a specific volume of the Du Vere edition.
The ontology was designed and validated against a catalog of competency questions (CQs) derived from user stories across two requirements engineering phases, elicited from four personas.
Phase I 29 ontology-focused CQs from 32 user stories Phase II 41 extended CQs covering extraction provenance, bibliographic depth, and epistemic modeling
| Persona | Role |
|---|---|
| Elena Rossi | Art historian researcher |
| Nazeera Marfi | Professor of digital art history |
| Aaron Warner | Computer scientist / knowledge engineer |
| John Saffron | Art history student |
viewsari:cooccurrence_cim-gio_1viewsari:inParagraph → viewsari:the_lives_1568_volume-1_paragraph-13dct:isPartOf → viewsari:the_lives_1568_volume-1_cimabue-bio_page_7frbr:isPartOf → viewsari:the_lives_1568_volume-1_cimabue-biofrbr:isPartOf → viewsari:the_lives_1568_volume-1frbr:isPartOf → viewsari:the_lives_1568 (Du Vere translation)viewsari:isBasedOn → viewsari:le_vite_1568 (Giunti edition)
wd:Q1479873 (Church of S. Cecilia) appears in
Giotto's, Cimabue's, and Cavallini's biographies — linked via shared Wikidata QID.
viewsari:the_lives_gutenberg_version (fabio:ManifestationCollection)viewsari:the_lives_1568 (viewsari:translation) — Du Vere, 1912 Englishviewsari:le_vite_1568 (viewsari:edition) — Giunti, 1568 Italianviewsari:le_vite (fabio:Work) — wd:Q1645493
viewsari:the_lives_1568), which in turn viewsari:isBasedOn
the 1568 Giunti Italian edition. The FRBR distinction lets queries target either level.
| Biography | Mentions | Wikidata-linked | OOKB |
|---|---|---|---|
| Michelangelo | 395 | 59 (15%) | 57 (14%) |
| Ghirlandaio | 235 | 53 (23%) | 64 (27%) |
| Giotto | 203 | 44 (22%) | 52 (26%) |
| Botticelli | 110 | 41 (37%) | 16 (15%) |
| Pollaiuolo | 127 | 12 (9%) | 52 (41%) |
owl:sameAs, enabling federated queries.
viewsari.ise.fiz-karlsruhe.de/kb/1.0#f19bb56f…).
viewsari:obliquer_run_1 (prov:Activity · NER)doco:Paragraph_96viewsari:m_0008 (implicit_artwork_mention)doco:TextChunk [start:1487, end:1530]viewsari:entity_0006viewsari:entity_linking_run_1 (prov:Activity · EL)viewsari:entity_0006Some CQs address multiple clusters. Phase II includes additional sub-questions generated via OntoChat from extended user stories.
viewsari:cooccurrence instances rather than annotated edges, following the Participation Ontology Design Pattern. This allows provenance metadata, paragraph anchoring, and statistical scores (e.g., PMI) to be attached directly to the co-occurrence node.viewsari:mention is a subclass of both oa:Annotation and prov:Entity. This reflects the dual role of every mention as a scholarly annotation grounded in the source text and as the output of a documented computational process.owl:sameAs links to Wikidata QIDs, not through coreferential annotation.