Representing Narratives in Digital Libraries: The Narrative Ontology

Tracking #: 2357-3570

Carlo Meghini
Valentina Bartalesi
Daniele Metilli

Responsible editor: 
Special Issue Cultural Heritage 2019

Submission type: 
Full Paper
Digital Libraries (DLs), especially in the Cultural Heritage domain, are rich in narratives. Every digital object in a DL tells some kind of story, regardless of the medium, the genre, or the type of the object. However, DLs do not offer services about narratives, for example it is not possible to discover a narrative, to create one, or to compare two narratives. Certainly, DLs offer discovery functionalities over their contents, but these services merely address the objects that carry the narratives (e.g. books, images, audiovisual objects), without regard for the narratives themselves. The present work aims at introducing narratives as first-class citizens in DLs, by providing a formal expression of what a narrative is. In particular, this paper presents a conceptualization of the domain of narratives, and its specification through the Narrative Ontology (NOnt for short), expressed in first-order logic. NOnt has been implemented as an extension of three standard vocabularies, i.e. the CIDOC CRM, FRBRoo, and OWL Time, and using the SWRL rule language to express the axioms. An initial validation of NOnt has been performed in the context of the Mingei European project, in which the ontology has been applied to the representation of knowledge about Craft Heritage.
Full PDF Version: 

Major Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
Anonymous submitted on 18/Mar/2020
Review Comment:

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 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.

The paper presents the idea of representing narratives as first class citizens in Cultural Heritage (CH) semantic web applications, and a detailed formalization of an ontology called NOnt based on logic with alignments with CIDOC CRM, FRBoo, and OWL Time Ontology.

(1) Originalit
The paper has originality in its view of addressing the problem of representing intangible narrative CH. A fairly comprehensive related works section is provided. Also a more general philosophical and historical introduction to the topic including, e.g., a reference to Aristotle is provided.

(2) Significance of the results

The focus and the main contribution of the paper lies in presenting a very detailed logical formalization/axiomatization of NOnt (section 7) and its implementation formally using OWL and SWRL is discussed, too. This a complex system including e.g. Allen’s time algebra as part of it. Two paragraphs are provided regarding the mapping to CIDOC CRM, FRBRoo, and OWL Time.
The presentation looks professional from a mathematical point of view, but it remains unclear why such heavy logical machinery is needed for representing narratives. The idea that we need to represent narratives is good but why should one need this heavy-weight complex system should be explained to the reader. What problems are actually solved here? Not even a single example of a narrative is presented in the paper and the use case of the Mingei European Project presented in section 8 is far too superficial to be convincing from a validation point of view, not to mention from a use case point of view.

Furthermore, the paper does not include an evaluation of the system. This acknowledge by the authors in the conclusions by saying: the project is “at an initial stage”.

In short, the paper presents an interesting problem and approach but fails in its current stage in showing the significance of the results.

(3) Quality of writing. The paper is well-written.

In conclusion, in spite of the virtues of the paper, I cannot recommend publication of the paper at its current stage due to criterion (2). More work than a major revision would be needed before publishing the work in the SWJ.

Review #2
Anonymous submitted on 10/May/2020
Review Comment:

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.

The paper proposes an ontology for representing narratives in digital libraries. After discussing the background and related work on the representation of narratives, it presents a First-Order Logic formalization of narratives, an implementation in OWL and SWRL and a mapping of the proposed ontology to standard ontologies such as CIDOC-CRM.

The initial sections (2-5) of the paper motivate well the need for an ontology for narratives and identify the data modelling requirements by reviewing the relative literature from Narratology and Knowledge Representation.

Section 6 presents the NOnt ontology, which implements the ideas discussed in the previous sections. The proposed formalisation captures well many of the discussed concepts, focusing more on the representation of time. Although the proposed formalisation seems correct, the paper does not present any examples, which would help clarify the meaning of each element and how the ontology works in practice. Furthermore, some important concepts discussed in the previous sections, such as those of the characters involved in an event or the place of an event, are missing from the ontology and the paper does not make clear how they can be represented.

The implementation of the ontology is Section 7 seems technically correct but again no examples or any kind of validation are given making it difficult to assess whether it is actually correct. Section 7.3 presents the mappings of NOnt to external ontologies using only two tables and without explaining why these mappings are meaningful and how they were derived. It is, therefore, again, difficult to judge whether the mappings are correct.

The validation of the ontology presented in Section 8 does not make things clearer. It only consists of a mapping of NOnt to an ontology developed for the needs of a research project. This does not, in my opinion, constitute a validation of the ontology, but only a demonstration of its use. Furthermore, without explaining the meaning of the elements of the target ontology, it is impossible to assess whether this mapping is correct and meaningful.

For all the above reasons, and mainly because of the lack of any evaluation of (or plan for evaluating) the proposed ontology, I can't recommend the acceptance of the paper. The results that the paper presents are well motivated, potentially significant and in the right direction, but need to be refined and properly evaluated before being published in a journal.

Review #3
By Beatrice Markhoff submitted on 22/May/2020
Minor Revision
Review Comment:

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.

(3) quality of writing
The paper is well written, well structured, main concepts and ideas are clearly presented.

(1) originality
The article presents a conceptualization of the domain of narratives and its specification through an ontology, expressed in first-order logic and then implemented using semantic web standards. Narratives are of utmost importance for endowing Digital Libraries with advanced visualization and exploration tools. The related works section demonstrates that authors perfectly master the topics addressed here. They have already proposed preliminary versions of an ontology of narratives and demonstrated interesting applications in the field of Digital Libraries, in particular in the context of Europeana. But in this article they present how to express the semantics of a narrative in the form of a Web Knowledge Base, following a rigorous formal methodology, which to my knowledge does not already exist in related works.

(2) significance of the results
In this paper, the focus is put on extensively covering the domain of narratives. To this end, the authors recall the distinction between narrations (told stories) and fabulae (actual stories) and their link by reference functions. It is explained, on the one hand, how a narration can be formalized by decomposing it into meaningful events, each event being formally represented, the connection between fragments of the narration and the formalized events being performed via reference functions. On the other hand, starting from the events of a fabula, each of them can be documented to build a narrative, using the reference functions the inverse way. In this context, the authors propose to treat the fabula of a narrative as a knowledge base composed of a set of statements giving the best available approximation of the fabula according to the narrator of the narrative. In order to build such KBs, they develop a language to express formal fabula statements.

Of course, the presence of different versions of the same story manifests different point of views, that are all important and must be kept and documented, so each narrative should be a separated KB for being able to reason on one narrative while avoiding inconsistencies with other ones. The paper is about the formalization needed for building a KB representing only one narrative, which is in itself challenging.

In particular, the core of the paper is about representing time and basic temporal relations in narratives, formally defining them and rigorously stating their behaviors. It is very common, in a narration, that only a relative relation between intervals is known, i.e. it is said that an event occurs before, after, or during another event. In the fabulae KB, the authors propose to represent the relationships between the time intervals by a network called Qualitative Temporal Knowledge, whose nodes represent the time intervals in the narrative, while arcs are labeled by sets of Allen’s relations. It is absolutely essential to be able to reason about the temporal aspects of a narrative and it is also well known that the representation of time is a tricky subject, which in my opinion the authors treat here with all the necessary rigor.

Beyond the framework of narratives for digital libraries, such a work may be reused. Rather than coding ad-hoc programs to deal with time representation, with the risk of errors, this proposal defines in a declarative way the qualitative temporal knowledge networks, and, thanks to semantic web standards, their definitions could be reused. I notice that interoperability is also well taken into account by authors, by aligning their proposal with OWL Time for general applications, and CIDOC CRM for Cultural Heritage ones. This is why I consider that this article has its rightful place in a special issue of the semantic web journal devoted to cultural heritage applications.

I would only have two demands: the first one is to provide in the paper a link for accessing the current state of the implementation of NOnt with OWL and SWRL, and the second one is to add at least one concrete example to illustrate how the implementation of this very interesting proposal is used in practice in the Mingei European project.

Review #4
Anonymous submitted on 18/Jun/2020
Major Revision
Review Comment:

This paper addresses an important problem: offering services about narratives in digital library. A thourough related work section shows the limitations of state-of-the-art solutions to handle this problem. The proposition consists of an original ontology called NOnt which can be used to define narratives. This ontology is formally defined to ensure the consistency of the narratives. Thus the paper has a clear and original contribution completely in the topic of the special issue.

Despite the paper being well-written, the definiton of NOnt lacks examples. It would be useful to have a concrete example used throughout the paper. Another limitation of the proposition is a lack of a thourough validation of the proposition. The validation section provided in the paper does not show the impact in practice of this ontology. This section must be completely edited. In my opinion, this section should contain a concrete instantiation of the ontology with examples of queries that could be written (e.g, comparing two narratives on the same subject, etc.). I would also like to see in this section the kind of applications that could be done with this ontology (e.g, tool to build narratives or to show the history of a narrative, etc.).