Review Comment:
This paper presents a set of ontologies applied to the elderly people in Africa. The authors claim that the ontologies were built based on surveys with the goal to play a key role in a learning and recommendation systems. Although I am very pleased with the application of using ontologies and semantic technologies to the domain, I am not convinced that the current paper is ready for publication in the Semantic Web Journal, mainly due to the lack of originality and the significance of the results. The paper is easy to read even though there are some figures that are not necessary (e.g., Figures 9 and 10 just to mention those two figures). I would suggest the authors (re)submit a mature version of the manuscript in a different track, such as “Descriptions of ontologies”.
Detailed comments
1- It is hard to evaluate the ontologies since Section 5 is mainly a screenshot of a SPARQL query. I suggest to authors to use SoA ontology evaluations, like using Competency Questions (CQs) and other sounded metrics to evaluate the scope of the ontologies.
2- In the Conclusion, the authors mentioned “OntoSen has been proposed [..] to make learning systems smarter”. This is exactly what we want to see described in the paper. How has the use of ontology actually improved the learning process? A comparison of both approaches will be useful to assess this work.
3- Section 6: “Enrich the ontology by extending its fields of action”. Does that mean the actions are not in the current scope of OntoSen? If yes, how do you evaluate the scope of OntoSen?
4- Figure 9: Retired is SubClassOf Professional. What about modelling this as an Event happening at some period of time? How do you capture the “life-cycle” of a Person from (adult, professional, Senior and retired) in a human life? I don’t see any mention of time. For example to say “Bob was Lecturer at the University of Maroua from 2000 to 2020 and he has been retired since January 2021”.
5- I miss a link to the set of ontologies for assessment. Please, provide a link of the resources so that we can easily assess them. It is difficult to understand your modeling choices. The class “Have_income” is an example. What are possible instances of this class? Sometimes, it is also hard the choice between the classes instead of enumerated values for instance. This modeling choice should also be clearly described in the manuscript.
6- You take the reference from the surveys to make your ontology. Would that not be a restriction to your application domain? If tomorrow you have a different survey, how would it impact the changes of your modeling choices?
7- One of the objectives of creating ontology is also to support FAIR principles and reuse as much as possible existing ontologies. That criteria is not available in your ontology. For example, why don’t you reuse foaf:Person or any other ontology available at https://lov.linkeddata.es/dataset/lov/terms?q=Person&type=class This also applies to any other classes in the ontology.
8- Section 4.6: I wonder if you have checked which reasoner can deal with the rules. Maybe write them in SWRL, RIF, N3 or other semantic languages to be able to make reasoning on top of the instances.
9- Figure 2: The architecture is not clear to me. Are the links representing the “semantic relations” between the modules? How do you use the different modules? I suggest also to check the works on NeOn Methodology and Ontology Design Pattern (ODP) to get inspired on how to “interconnect” different modules of ontologies.
10- In section 4.3.1 “Most African seniors spend their day at home …”. Please, add a reference.
11- Table 2: I miss in the SoA (section 3) other works/studies using ontology in context-aware systems with users to complete table 2.
12- Figure 1: Where is the reasoner? On the server side, what contains “DB”? I guess this figure if completed would be the architecture of the purpose of this ontology. Hence, it could constitute the main application of the designed ontology.
13- Section 2.1: In your assumption of an elderly person, why don’t you model an elderly_person according to the country of origin? What happened to someone in a different country outside of Africa to use OntoSen? Moreover, I do not see the axioms associated with the assumption “an elderly person is a person from 50 years old located in those 10 countries in Table 1.
Definitely, this is a nice use case for leveraging semantic technologies, but the current manuscript contains many weak points. Also, the use of the title “framework” is somehow misleading because the paper mainly describes a set of ontologies. IMHO, the authors do not meet their goal as stated in the introduction to propose a system using ontologies to anticipate the intentions of the users, hereafter called seniors.
|
