Grounding the development of an ontology for narrative and fiction

The paper addresses the methodological and theoretical foundations for building computational ontologies for modeling narrative and fiction, focusing on literary characters. It critically surveys existing ontologies in the domain, assesses their alignment with foundational ontologies (DOLCE, BFO, CIDOC-CRM), and proposes a two-step ontology mapping strategy (harmonization and alignment) to foster interoperability and semantic clarity. It concludes with a proposed ontology pattern for fictional characters and illustrates potential mappings with foundational ontologies, underlining the challenges posed by the non-referential nature of fictional entities.

The paper is well-written and technically precise, although many sections are quite dense, with long paragraphs and few line breaks, which complicates reading. The paper has a well-defined objective, aiming to propose a methodology for developing interoperable ontologies for narrative and fiction, with a specific focus on literary characters. The methodology is well explained, and the inclusion of a state-of-the-art section with a motivated selection of relevant domain ontologies provides a strong foundation for the analysis. The proposed framework for harmonizing and aligning domain ontologies with foundational ontologies is significant, illustrating applicable ontological patterns and mapping diagrams useful for researchers who wish to implement semantic models for narrative and characters.

However, while the paper proposes a clear and systematic methodology for harmonization and alignment, the absence of a practical case study or experimental validation using real narrative data means that the effectiveness and robustness of the proposed approach remain untested. Including an implementation with a reasoner to check consistency, or applying the ontology pattern to a real corpus to demonstrate semantic interoperability across models, would significantly strengthen the contribution by moving it from a purely theoretical proposal to a practically validated framework.

In light of the above, I suggest that the authors:

1) Make the text less dense in several sections to enhance clarity and accessibility for the reader while maintaining the scholarly depth of the analysis. For instance, in the Introduction, the paragraph beginning with “In the age of Open Science...” spans multiple ideas (Open Science and FAIR guidelines, the state of the art in hard sciences vs. humanities, the need for FAIR data in the humanities, the role of DH and the Semantic Web) without clear breaks, making it challenging for readers to follow the logical progression toward the paper’s objectives. Another example of section that would benefit from added line breaks is the State of the Art section. The current presentation of ontology models for narrative and fiction appears as a dense, uninterrupted list, which can overwhelm readers and obscure the key distinctions between different approaches. I would suggest inserting clear breaks to distinguish the different ideas within these paragraphs and facilitate smoother reading.

2) Apply the proposed ontology pattern to one small corpus (e.g., characters in a novel or a movie) to validate it on real data, or, if possible, use a reasoner to check its consistency. This would provide practical evidence and the validation of the framework’s effectiveness.

The paper addresses an important and difficult topic: the construction of an ontological infrastructure for narrative and fiction that takes into account existing approaches. I consider it an interesting contribution not only as a survey of the existing literature, but also for its effort to harmonize and align different models.

First of all, the title promises an ontology for “narrative and fiction” in general, but the paper focuses almost exclusively on literary characters. The title should more accurately reflect this specific focus on characters—for instance, “Grounding the Development of Ontologies for Fictional Characters” or something similar.

I admit I am not deeply familiar with the technical aspects (except partially with the DOLCE ontology), and therefore it is not entirely clear to me, for example, how the authors handle quality spaces in UML. My question may be naive, but it would be worth clarifying more precisely how these aspects are formalized.

In the end, what matters in practice is how an ontology is actually populated.

Regarding the issue of populating the ontology, the paper would benefit from an example that illustrates, at least for the main aspects, how their approach works in practice.

The effort toward mapping and alignment is significant, but the literature cited on the topic is presented somewhat linearly and appears slightly outdated. In fact, the conceptual problem is quite complex: if I change an axiom that characterizes a concept in an ontology, can I still say I have the same concept? Or is it a different concept? It is easy to see how this problem transfers to the issue of mapping between ontologies, as the authors themselves seem to acknowledge.

In short, beyond trivial cases, ontology mapping appears far from solved, and there is a rich body of literature on the subject. I would suggest discussing these limitations and the underlying conceptual debates more critically.

For a very recent approach and further references, I suggest:

Masolo, C., Compagno, F., & Borgo, S. (2025). On the Formal Alignment of Foundational Ontologies: Building Mappings Between Basic Formal Ontology and Descriptive Ontology for Linguistic and Cognitive Engineering. Applied Ontology, 20(2), 153–180. https://doi.org/10.1177/15705838251334527

In this paper, the authors show that even seemingly minor modeling differences can have a significant impact when formal alignments are attempted, including with respect to the issue of qualities. Masolo et al. argue, for example, that integrating DOLCE and BFO in this regard would require extending both ontologies.

The final proposal (three classes with two relations) does not address the fundamental issue: attributing qualities to fictional characters is a subjective interpretive act. Who decides, for example, whether Hamlet is mad? How are conflicting interpretations managed? How can the ontology be populated in the presence of such ambiguities? This limitation (of the current literature) should be acknowledged and discussed. An attempt is done in:

Sanfilippo, E. M., Masolo, C., Ferrario, R., & Bottazzi, E. (2024). Interpreting Texts and Their Characters. In Proceedings of the 14th International Conference on Formal Ontology in Information Systems (FOIS 2024) https://www.utwente.nl/en/eemcs/fois2024/resources/papers/sanfilippo-et-...

Of course, no one expects this issue to be solved here, as it lies beyond the scope of this work.

In conclusion, I find the contribution to be original. To the best of my knowledge, this is the first time that a comprehensive comparison between domain ontologies and foundational models has been undertaken specifically in the field of literary fiction. The result is significant in its own right: the comparison provides an indispensable basis for anyone wishing to develop or reuse data models for computational literary studies.

Moreover, the quality of the writing is high. The text is clear, precise, well-structured, and employs appropriate academic language.

My overall recommendation is accept with major revisions, although these revisions are not “major” in the sense of requiring a complete overhaul of the work’s core structure. I would assign a score of 70/100. Please note that my evaluation focuses primarily on the general framework and conceptual aspects, not on the technical details.

This manuscript was submitted as 'full paper' and should be reviewed along the usual dimensions for research contributions which include (1) originality, (2) significance of the results, and (3) quality of writing.

This paper is clear in its goal, formulation, and explication. By applying foundational ontologies such as DOLCE and BFO, and aligning them via CIDOC-CRM, the article outlines the work undertaken to propose a shared conceptualisation for fictional narrative entities, with a particular focus on literary characters. The approach undertaken is appropriate and follows well-known and established ontology engineering methodology.

The paper introduces an experimental ontology pattern for fictional characters, aligning it with foundational ontologies (DOLCE, BFO) via CIDOC-CRM. The article’s application of DOLCE and BFO to the formulation of ontological descriptions of fictional characters is clearly articulated and convincing. This is novel because fictional entities are notoriously difficult to formalise due to their non-referential and interpretative nature.
The paper's methodological approach applies a two-step ontology mapping methodology (harmonisation and alignment) specifically tailored to narrative and fiction, rather than creating a new ontology from scratch, to critically judge the classes and relations proposed. The application of foundational ontology classes and reasoning is well articulated and logically consistent with those frameworks.
The overall organisation of the paper’s content is clear and the research references the relevant literature in both ontology engineering and literary studies.

I have no hesitation to recommend it for acceptance for publication.

Other Minor issues
1. Table 1, P2 should 2 be a subscript?
2. Caption Table 2 ...across ontologies -->across domain specific ontologies