Representing Narratives in Digital Libraries: The Narrative Ontology

Tracking #: 2539-3753

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. On the basis NOnt, we have developed the Narrative Building and Visualising (NBVT) tool, and applied it in four case studies to validate the ontology. NOnt is also being validated in the context of the Mingei European project, in which it is applied to the representation of knowledge about Craft Heritage.
Full PDF Version: 

Minor Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
Anonymous submitted on 11/Aug/2020
Minor Revision
Review Comment:

The revised submission addressed many of my and the other reviewers' comments. Specifically:
- The content added in Section 2 clarifies the rationale of the proposed methodology and explains why certain methodological choices were made, e.g. regarding the identification of a tractable subset of temporal relations and the evaluation of the ontology.
- In Section 6, two more unary and two binary predicate symbols and three axioms were added to represent the participation of an entity in an event and the place that the event occurred.
- Section 8 was substantially extended with a new section that describes a tool built on top of the proposed ontology, demonstrating the usefulness. In the same section, more details about the Mingei project were added demonstrating better how the ontology was used in the project.

However, the paper is in my opinion still not ready for publication. The most important limitation is the lack of any examples showing how a narrative and other related concepts are modelled using elements of the proposed ontology. What I would expect to see is at least: an example (textual description) of a narrative, and a graph-based representation of the same narrative using elements of the proposed ontology. It would also be useful to present the code of (some of) the SPARQL queries that implement the functionalities of NBVT described at the end of Section 8.1. More generally, this section (8.1) could focus more on how the proposed ontology supported the implementation of NBVT (using examples!) rather than the description of its architecture and features, which are anyway presented in detail in a previous publication (6).

The presentation of NBVT also raises some additional questions. You claim that the tool provides support in identifying the role of a person in an event and defining the type of an event. How are these two features (role of a person in an event, type of an event) modelled in the ontology? I couldn't identify any relevant concepts in the section presenting the ontology.

Regarding the pragmatic evaluation of the ontology, I understand that it may not be possible to evaluate its accuracy or correctness when the relevant concepts (e.g. the notion of narrative) are vague. I also agree that a way to evaluate the ontology is to test it against the requirements of real applications. In this respect, it would be useful if you explicitly described the data modelling requirements of the two applications (the NBVT tool and the Mingei project) and explained the extent to which the ontology met those requirements. You state in Section 2, that you have received positive feedback in both settings. Can you give more details about it?

Some minor comments:

- If I understand well, Partic(c) is meant to represent a person, an object, a concept or any entity that may participate in an event. The description of the predicate ("participant of the event") is rather inaccurate and misleading as it implies a binary predicate with the event as one of its two arguments.

- Figures 1 and 2: I think it would be better if the two figures are merged into one.

- The paper is in general well-written but needs to be carefully proof-read again for typos and minor language errors.

Overall, although the ideas presented in this paper are original and potentially significant, I believe that the paper still requires some small but significant amendments before it is published.

Review #2
Anonymous submitted on 24/Aug/2020
Minor Revision
Review Comment:

Authors have addressed some of my comments/suggestions in the revised submission. The section 2 has been rewritten to clarify the main focus of the paper and to explain how the ontology was evaluated. Moreover, the section 8 was completely rewritten to (1) describe the Narrative Building and Visualising Tool (NBVT) tool ; a software built on top of the proposed ontology that allows creating narratives and visualising them and (2) explain how the NOnt ontology was used in the Mingei European project.

However, the paper still lacks examples. As I suggested in my first review, the paper should contain a complete example of a narrative used throughout the paper. This narrative should be first explained informally. Then, a concrete instantiation of the NOnt ontology should be provided to show how this ontology can be used to represent this narrative. Lastly, examples of SPARQL queries could be provided to show the interest of this instantiation. For example, I would like to see interesting SPARQL queries that could be written thanks to the NOnt ontology such as one comparing two narratives on the same subject.

This example could also be used to illustrate the NBVT tool. Concerning the description of this tool, I would like to see in more details the link between the functionalities of this tool and the NOnt ontology. Does this tool only rely on the NOnt ontology or if it relies on other models to implement some functionalities. I would also like to know if the functionalities have been only implemented with SPARQL queries based on NOnt or if more complex algorithms were necessary.

Concerning the Mingei European project, this section does not counvince me of the interest of the NOnt ontology. Firstly this section explains that this ontology has been extended to cover the requirements of the Mingei European project. It is unclear if this is a limitation of NOnt or just requirements of the project that are outside the scope of narratives. I suggest to give more details on the sentence "CrO adds to NOnt features that are useful in the context of the Mingei project,such as the notion of process and that of presentation of a narrative". The reader should be counvinced that the added features are not relevant to the NOnt ontology. Moreover, the main outcome of the project is that the CrO ontlogy based on NOnt is well suited to support the Mingei activity. If I agree that this is a positive point, I would also like to have some insights of the interests of being able to represent formally narratives and particularly how these instantions of the CrO ontologies were used in practice in the context of this project.

Review #3
By Beatrice Markhoff submitted on 27/Aug/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 authors met my expectations. The article has been improved by a clarification in Section 2 on motivations and methodology, and by the addition of elements in Section 8 demonstrating that the proposals in this article are actually used in concrete applications. One fundamental quality of these applications is that valid reasoning can be performed on one narrative (one knowledge base) represented with NOnt, either for checking the consistency of qualitative temporal knowledge, or for answering queries about such qualitative temporal knowledge. The core of the paper shows how such reasoning is possible with existing semantic web reasoners, by devising a tractable subset of basic temporal relations which is applicable to narratives.
I suggest to accept this article.