SeMFIS: A Flexible Engineering Platform for Semantic Annotations of Conceptual Models

The author present in the paper a tool called SeMFIS for semantic annotation of conceptual models. They review existing proposals in the area, presents the metamodel that underpins their tool, its architecture and present two use cases that were implemented using SeMFIS. The tool presented by the author exhibit features that are attractive, in particular the decoupling from the specificities of modeling languages. Nevertheless, I have major comments (concerns) on the paper that I present in what follows.

C1. The major contribution that is put forward by the author is that “SeMFIS does not require a specific type of modeling language or the modification of an existing modeling language. Due to the decoupling of the semantic annotations in separate annotations models, SeMFIS can be easily added to the existing modeling methods without affecting their structure nor behavior.”. It is well recognized that with more flexibility (which is in this case is the result of decoupling from any specific modeling language) comes more complexity. In particular, the user of SeMFIS will have to deal with understanding of the semantics of the modeling language in question, understanding SeMFIS metamodel, and maping the two. This task seems to me to be complex and tedious. Yet, I did not find in the paper any discussion on this topic, e.g., how SeMFIS users can be assisted.

C2. Another point that is related to C1 is that usually a modeling language comes with its own notation, syntax but also semantics. This semantics may require adding new knowledge in the ontology used for reasoning about annotation. Here again, I did not find a convincing answer in the paper.

C3. The author presents two use cases taken from business process modeling and semantic web services. These areas have already been covered by a large number of proposals that illustrated how semantic annotations can be used to guide the design of business process, identify constraint violation. It would have been IMHO better to focus on a modeling language that has not received much attention from the semantic web community so far, e.g., UML class diagram. This in my opinion would have given more weight to the contribution made in the paper. It is also fair to say that the presentation of the two use cases is not deep and does not show how the tool was effectively used and the lessons learned in the process.

C4. The validation of the paper does not demonstrate the claim made in the introduction as to that SemFIS supports different (multiple) modeling languages.

C5. In summary, I think that the paper presents a nice tool that can be used by the community (provided that more evaluation can be done). However, I don’t think that the paper makes a solid contribution, and does not back the claims made as to the flexibility provided by the tools and its ability to support the annotation of diverse modeling languages. Because of this, I am afraid I don’t think that the article as is warrant a journal publication.

The paper contains the system description of the SeMFIS tools. SeMFIS is a platform for the creation of conceptual modelling tools extended with semantic annotations. SeMFIS is based on the (commercial) ADOxx meta modelling platform, though a freely available version exists. The key feature is that it can be used to attach annotation to an arbitrary modelling language with no need of adapting the language. A limitation of the tool is that it is only Windows based.

The description of the tool, its architecture and the techniques for attaching (semantic) annotation is reasonably clear. What is not really clear is the effort required for users to create interfaces for (new) modelling languages and proper evaluation of the tool (the paper mentions it has been used by a group of students, and it received positive feedback, I feel the paper would benefit from a more description of the evaluation and presentation of the feedback).

From what I have understood from the presentation in the paper a key mechanism is that of scripting in the native scripting language, I think the paper should expand on the scripting language and provide more information and examples.

If I have understood correctly the tool can call external services. However, this part is not properly described in the paper. In my view this is an essential feature. Very often semantic annotations are used to perform some form of analysis on the modelling language they are attached to. So while SeMFIS could be used to add annotation, then if no analysis service is attached to it, then one has to export and used different tools for the analysis reducing the utility SeMFIS.

The English language is appropriate even if there are a few slips here and there:

Abstract: "Conceptual models have in the past been used..." I would use "Conceptual model have been used in the past ..."

Page 1, col 2: Sentence starting with "When creating conceptual ..." please check the sentence for meaning and grammar.

General: the verb "revert" has been used in several occasions, but it seems to me that it does not convey the right meaning (e.g., second line of section 3.2, page 8, mid first column).

Page 7, col 2: "as has been shown above with the ex- ample for integrating the web service menu" please check it for meaning and grammar.

This paper presents a new version of the SeMFIS tool. The authors focus in this paper on the modeling, the query and the exchange functionalities. They clearly motivate their work and its importance through the literature review. The authors described their contribution through a detailed description of the tool:

Quality: This paper is of good quality. The authors start by giving an overview of the literature by pointing out the lacks and on this base an architecture was designed and then they clearly explain the different features of the tools and proposed a use cas. Regarding the architecture it is not very clear why the modelling subsystem is considered in the repository layer since it is more a managing module; And maybe the authors can present within the figure 5 what are the parts/modules that have been developed through their work and what are the other parts/modules that have been rue-sed or adapted.

Importance: The tools presented has its importance in the domain of engineering of conceptual model. This importance has been well described by the authors in the introduction part. However we can hardly estimate the time consumed to put in place the ontologies and to integrate the necessary semantics and thus it is hardly possible to estimate is this important need of this tool is easy and efficient to use and to model. I think it is important, in order to complete this paper, to integrate an evaluation section. “The feedback we gained from students was very positive” is not measurable. how the feedback were given; what does it mean “very positive”? etc.

Impact of tool: The authors present different tools/works and distinguish themselves with new features regarding the semantics; they argued that PRO-SEAT permits to create semantic annotation but no open accessible version of that tool can be found. This is an advantage for their tools and the impact it may have since it can be reused; However their approach is based on a tools based on Windows which may alter the impact of the tool in a sens that not every organisation would be able to use their tool. In my point of view the authors can add a section for performance and evaluation where they can state in a more qualitative way (then the section 6) the effect of having this platform only available on Windows.

Clarity & Readability : The paper is very well written. Every section is clearly presented and motivated. As a small suggestion I would propose that the authors presents the concrete context of the use cases and the overall architecture before going in details in the functionalities of the the platform.

Illustration: The figures are well presented and self-contained. It would be preferable to provide a better view for the figures 3. in priority and the figure 2 if possible.