This paper introduces ALIGHIERoo (Advanced Literary and Iconographic Graph for Hermeneutics of Interconnected Editions and Resources - object-oriented), an ontological framework currently under development within the Naples Dante Project. Drawing upon CIDOC CRM and LRMoo, ALIGHIERoo focuses on the representation of intricate textual and iconographic elements within Dante's Commedia, though it aims for a high level of scalability. By prioritizing semantic granularity and inferential reasoning, ALIGHIERoo is envisioned as a versatile and expandable framework for digital philology and heritage studies. The project constitutes a structured proposal for future refinement and collaborative validation.
promise as an adaptable model, ofering both domain-specific insights and generalizable strategies for digital humanities research.
The ontology not only incorporates the foundational distinctions of LRMoo—such as the layered segmentation into Work, Expression, Manifestation, and Item—but also extends them through domainspecific inferences.
The fundamental inspirations for encoding the elements of ALIGHIERoo originate from ongoing and
completed initiatives within the Naples Dante Project (NDP), particularly the Illuminated Dante
Project (IDP) [
Regarding OWL-specific components, ALIGHIERoo utilizes previously established ontological
frameworks. Among these, MeMO (Medieval Manuscript Ontology) [
ALIGHIERoo is conceived within this framework as an innovative tool for Dante studies. By merging the semantic capabilities of OWL with the structural coherence of CIDOC CRM, ALIGHIERoo aims to provide a comprehensive and interconnected representation of the Commedia’s textual and iconographic traditions. This approach enhances scholarly research on Dante and establishes a scalable model for the examination of other literary corpora, supporting the broader advancement of digital humanities research.
The current version (1.3) includes 68 classes, 10 object properties, and 123 individuals, totaling 151 classes and 370 object properties when integrated with CIDOC’s framework. Its design supports the structured classification of manuscripts, early printed editions, scholarly works, and the extensive iconographic apparatus surrounding them.
At its core, CIDOC CRM organizes knowledge around six primary branches deriving from the overarching class E1 CRM Entity: Dimension, Persistent Item, Place, Spacetime Volume, Temporal Entity, and Time-Span. These categories enable the modeling of physical, temporal, spatial, and conceptual relationships across cultural heritage domains. The framework is event-centric, structuring data around events connecting people and objects (persistent items), places, and times. This design facilitates the depiction of dynamic processes, including production, transmission, and interpretation, which are particularly significant in the analysis of intricate textual traditions.
ALIGHIERoo adapts and extends this structure to the specific needs of Dante studies. It introduces enhancements for domain-specific scenarios such as illuminated traditions, miscellaneous manuscripts, and fictive topographies. The result is a model that maintains interoperability with CIDOC-compliant ontologies while providing the specificity required for fine-grained reasoning.
In E54 Dimension, a new dimension, header_length, represents the number of bytes used for the header of a packet or digital document. This may enable automated agents to detect scans falling below quality standards or to measure data loss over digital preservation processes. Tracking degradation is crucial for preserving the integrity of digitized materials, especially in long-term archival projects.
ALIGHIERoo adopts the same dimension-recording methodology as EpiOnt [
These concepts enhance precision in representing fragmented textual materials, supporting a more lfexible and rigorous approach to manuscript and epigraphic studies. The methodology aligns with CADMUS, strengthening its applicability within scholarly frameworks.
We turn to the Persistent Items class. Within the Person subclass, ALIGHIERoo expands the range of recognized agents to trace the complex transmission and reception of textual works. These key actors, defined through projects such as the IDP and IMAGO, document the intellectual, material, and cultural transformations of manuscripts and printed works. The newly introduced agent classes are listed in Table 1.
Role Annotator Author Commentator Concepteur Copyist Curator Dedicatee Holder Illustrator Patron Publisher Restorer Scholar Stationer Translator Typographer
Description Adds explanatory or interpretative notes to a text, typically in the margins The original intellectual creator of a work Provides systematic interpretation or analysis of a text Conceptual architect of a project or structure Manually transcribes texts, potentially introducing variations Oversees preservation and scholarly interpretation of manuscripts or prints Recipient of the dedication of a work Holds or held a manuscript or printed item Creates visual elements that accompany a text Commissions or supports the production of a work Responsible for reproduction and dissemination of a work Repairs or conserves textual artifacts Conducts critical and historical analysis of texts Supplies or sells writing materials and books Renders a text into another language Designs and arranges printed type
Table 1: Agent roles introduced in ALIGHIERoo
These roles enable a more precise representation of the networks of intellectual, artistic, and material agents involved in the production, transmission, and reception of textual works. By distinguishing these figures, ALIGHIERoo refines the model to support scholarly inquiry in philology, book history, and manuscript studies. 3.3. Thing The Person subclass refers to real individuals; fictional characters are not accommodated within this CIDOC category. ALIGHIERoo reintroduces a Character class to represent personas depicted in textual works, ensuring a structured representation of characters. Within the broader CIDOC CRM ecosystem, an equivalent class has been assigned to CRMsoc, the extension dedicated to modeling social constructs and interactions. While CRMsoc is still in development, integrating such a class in ALIGHIERoo guarantees future interoperability without relying on external modules.
Additionally, two types of CIDOC’s E89 Propositional Objects are introduced: Iconographic_Action and Scene_Macro_Area. The former serves to describe highly specific, non-generic actions within a visual or textual context, while the latter, subdivided into Typical_Scene, enables the grouping and classification of visual elements such as miniatures or drawings: these items are defined with the CIDOC class E36 Visual Item. The individuals of Scene_Macro_Area have been arbitrarily determined by the NDP team, serving as broad conceptual categories into which multiple Typical_Scenes can be grouped. While the former provides a stable set of reference points, the number of Typical_Scenes remains theoretically infinite, reflecting the variability of artistic representation. Crucially, the combination of unique scene elements facilitates both inferential classification and content-based retrieval. Each miniature or picture can be seen as a Visual Item formed by a background (Fictional_Place), protagonist characters (Character), and distinctive actions (Iconographic_Action). Analyzing these components allows the reasoner to identify the subject of an illustration. For example, an image featuring Dante and the usurers likely refers to Inferno, Canto XVII, not merely due to Dante’s constant presence, but to the distinct presence of the usurers.
However, iconographic references are rarely univocal. It is not uncommon to find illustrations that merge elements from diferent cantos, such as a depiction of the Selva Oscura, the dark forest before Hell’s Gates, featuring Minos, the judge of Hell, found much later in the poem. Such variations arise from stylistic choices, illustrative constraints, established conventions, or artist misinterpretations. Leveraging SPARQL queries, the ontology determines whether illustrations adhere to canonical correspondences or incorporate references to other passages, enhancing interpretative scholarship and digital humanities methodologies. Such mix-ups justify focusing on unique elements rather than broader, omnipresent ones.
A core feature of LRMoo and its predecessors, FRBR and FRBRoo, is the structured division of propositional and information objects into four fundamental levels. Such hierarchical distinction allows for a precise representation of textual and non-textual heritage, ranging from abstract concepts to their physical instantiations: • F1 Work - conceptual essence of a creation, independent of any realization; encapsulates intellectual or artistic content. • F2 Expression - specific realization of an F1, capturing the multiple particular1 textual forms that define an intellectual variant of the content. • F3 Manifestation - a concretized form of one or more F2; it encompasses the medium with no reference to a tangible object. This level is crucial as it distinguishes between diferent editorial states. • F5 Item - discrete material instantiation of an F3, reflecting unique physical characteristics and provenance.
Building on this structure, ALIGHIERoo introduces the following subclasses:
For Expression 1As CIDOC specifies, every variation due to errors in copying or more purposeful editorial choices constitutes a new instance of F2. • Main Text - core intellectual production of the author, subdivided into: – Author Draft , preliminary writing stages, possibly including corrections, marginalia, or incomplete sections; – Author Final Draft , the ultimate version of a text completed or approved by the author. Such distinction arose from discussions within the NDP team comparing Dante and Petrarca: the Commedia lacks a surviving autograph, requiring reconstruction from multiple witnesses, while Petrarca documented multiple drafts. This diferentiation provides a nuanced approach to textual stratification, reflecting both authorial intent and textual transmission. • Paratext - includes supplementary textual materials that accompany, contextualize, or elucidate the main text, made by the author of the main text or by other agents. Rather than separating into distinct classes, the class groups textual elements such as glosses, annotations, ownership notes, rubrics, introductions, and appendices. These diverse components are viewed as interpretative or contextual layers, “ancillary texts” enhancing the comprehension, dissemination, or application of the main work. Future distinctions are possible, but the current version embraces a unified approach for clarity and adaptability.
For Manifestation • Critical Edition - editions produced with scholarly intervention, combining a base text with paratexts (apparatus, commentary, annotations) synthesized from multiple witnesses; • Manuscript Edition, hand-copied textual witnesses, varying in accuracy, scribal fidelity, or decoration; • Hand Printed Book, early printed versions (before circa 1830), including incunabula and other typographic products retaining manuscript traits.
An additional subclass is presented as a peculiar instance of manuscripts: Miscellaneous_Manuscript. These manuscripts are repositories of multiple F1 Works and their corresponding F2 Expressions. This category is crucial for studying medieval and early modern textual transmission, ofering insights into reading practices, intellectual networks, and the circulation of works.
In Design or Procedure, two key additions have been introduced: Book_Decor_Style and Book_Decor_Technique. These subclasses capture the stylistic and methodological aspects of decoration in manuscripts and printed editions. Avoiding rigid distinctions between manuscript and print traditions, the framework is able to describe even modern decorative elements that reference or revive historical styles and techniques.
The individuals populating the ontology reflect distinctions aligned with the classification criteria established in the Illuminated Dante Project, which are integral to the broader project and will be maintained or refined elsewhere. This ensures consistency across ontological components and supports detailed annotation of artistic and technical features across historical periods. 3.3.3. Type The Type class in CIDOC CRM serves as the most general categorization mechanism across domains, as it strategically enables classification without rigid constraints, allowing flexible attribution of types to objects, events, and concepts alike.
Within ALIGHIERoo, the class Book_Decoration_Type encompasses various forms of book ornamentation, as presented in Table 2. Each subclass targets specific individuals representing decorative elements, ensuring a structured and expandable framework. Although established primarily for Danterelated case studies, Book_Decoration_Type remains open to refinements and additions.
Beyond book decoration, additional type-based classes have been introduced in Table 3.
These expansions ensure that ALIGHIERoo can accurately classify and diferentiate textual and decorative elements, supporting a more precise analysis of historical manuscripts and printed materials. Subcategory Cover Decor Type Diagram Illustration ImageDecor (W.I.P.) InitialLetter Ornamentation
Description Structural categorization of folios in manuscripts and early printed copies.
Poetic meter of a text, crucial for analyzing verse structures.
Distinct textual components like colophons, dedications, and other defined segments. 3.4. Place In ALIGHIERoo, the Place class follows CIDOC CRM’s distinction, representing only real-world locations. To accommodate fictional settings -essential for literary studies- a separate Fictional_Place class was introduced. For real places, ALIGHIERoo expands the ontology with three subclasses: City, Region, and Nation. A Nation can encompass Regions, and a Region can contain Cities, all linked through the standard relation P10_falls_within. This approach facilitates SPARQL queries retrieving historical and administrative data relevant to manuscript provenance and textual transmission, ensuring compatibility while supporting spatial analyses of physical and narrative geographies.
Given the extensive relationships in CIDOC CRM, only a few new properties were introduced in ALIGHIERoo to enhance the modeling of narrative and visual content. These relationships align with the ontology’s class structure and integrate literary, artistic, and material aspects of textual heritage. • has_action (Typical_Scene → Iconographic_Action): connects a scene to a specific, non-generic action depicted within it. • has_background (Typical_Scene → Fictional_Place): specifies the setting or location within which the scene takes place, distinguishing real and imaginary environments. • has_character (Typical_Scene → F38_Character): links a scene to the characters it portrays. The inverse property, participates, associates a character with the scene in which they appear. • has_dedicatee (Human-Made Thing → Actor): captures the act of dedication, such as in manuscripts, printed works, or artistic creations. The inverse, is_dedicated_to, expresses the recipient’s role in this relationship. • has_fictional_counterpart (Actor → F38_Character): due to CIDOC CRM’s division between nonand real entities, this property models cases where a historical figure also exists as a fictional character, such as Dante the Author versus Dante the Character. More broadly, it applies to real ifgures appearing in literary works with altered or symbolic identities, as in The Divine Comedy.
The ALIGHIERoo ontology employs an inferential engine to analyze and classify scenes and textual
elements in Dante’s works. This approach aligns with methodologies like VideOWL [
Our inferential system relies on specific classes to formalize this process. Typical_Scene serves as a container for visual and textual representations, linking them to their essential components. Each Typical_Scene may be characterized by one or more: • Iconographic_Action, capturing distinct actions within the scene (e.g., “Virgil gestures toward the city of Dis” or “Dante faints”); • F38_Character, identifying key figures, distinguishing real from their fictional counterparts via has_fictional_counterpart; • Fictional_Place, classifying the setting into Earthly-, Infernal-, Purgatorial- and Heavenly_Place; • Scene_Macro_Area, grouping Typical_Scene instances into broad inferential categories for pattern recognition and high-level reasoning.
For instance, an illustration of Dante among the usurers in Inferno XVII is identified not just by the protagonist’s presence (constant throughout the work) but by unique F38_Character instances (e.g., Reginaldo Scrovegni) and the distinct Infernal_Place (the sandpit).
The system also accommodates artistic and interpretative variations, as manuscript illuminations often combine elements from diferent cantos due to stylistic choices, constraints, or illustrator misunderstandings. For example, a depiction of the Selva Oscura (the dark forest of Inferno I ) might include Minos, canonically appearing at the threshold of Hell later in Canto V. The Scene_Macro_Area classification mechanism detects such atypical combinations, enabling queries on adherence to the textual sequence or the introduction of other elements.
A crucial component of this system is the use of E41 Appellation to match text segments (F2 expressions with their corresponding E33 Linguistic Object) with their corresponding Canto. By associating each Canto with an Appellation instance, such as “Inferno Canto V ”, the ontology ensures that all textual entities are properly grouped according to CIDOC conventions. This parallels VideOWL, where stable appellations unify distinct but related elements in video game environments under a common reference.
Through these structured relationships, ALIGHIERoo ofers a robust mechanism for analyzing Dante’s works and more, demonstrating the power of ontological modeling in the humanities.
The high granularity of ALIGHIERoo enables multiple inference rules supporting classification, review processes, and reasoning across works and materials. By leveraging structured relationships and constraints, the ontology enables automated deductions, streamlining research and analysis.
A straightforward example involves the material properties of manuscript illuminations: if a miniature is colored gold, the corresponding manuscript2 necessarily contains actual gold, as historically the color could not be replicated synthetically.
A more articulated inferential mechanism addresses the identification of miscellaneous manuscripts, allowing one to grasp a key distinction: a single-work manuscript contains a unified textual expression, whereas a miscellaneous manuscript compiles multiple independent works. When a Manuscript_Edition3 embodies F2 Expressions distinct kinds of F1 Work, the new specialized property perpetuatesWork semantically links the manuscript to its associated multiple works. This is inferred through the property chain R4_embodies ∘ R3i_realises, able to bridge the F3/F2/F1 layers.
Applied at scale, these mechanisms enable more precise analysis of historical artifacts, assisting researchers in identifying and categorizing complex manuscripts and tracing text-image patterns. By embedding these reasoning capabilities into the ontology, ALIGHIERoo provides an advanced framework for studying medieval texts and their transmission. 2Linked via the relationship has_decoration_type to the class Book_Decoration_Type. 3As mentioned, a subclass of F3 Manifestation.
Although it is still a work in progress, ALIGHIERoo shows considerable promise in organizing and enhancing the analysis of textual works, particularly manuscripts and early print traditions. By integrating CIDOC CRM and LRMoo, it ensures compatibility with established standards, while new classes and properties broaden its applicability to complex textual and visual phenomena. These additions are a key strength of the ontology’s approach to such elements, heavily shaped by the contributions of the NDP. Consequently, ALIGHIERoo extends beyond being a simple augmentation of LRMoo, evolving into a valuable resource for the semantic analysis of manuscript illumination, decorative styles, and text-image connections.
As development continues, further refinements will strengthen the ontology’s capacity to facilitate in-depth research in textual heritage. The foundational work provides a robust basis for future developments, ofering a model adaptable to evolving scholarly demands in exploring historical texts and material culture.
The author(s) have not employed any Generative AI tools.