IFC Schemas in ISO/TC 211 Compliant UML for Improved Interoperability between BIM and GIS

Round 1

Reviewer 1 Report

The paper explores in detail  links and information exchange between two different standards - IFC (for BIM) and various standards from the ISO TC/211 committee and the Open Geospatial Consortium, presenting a systematic approach to information mapping with a strong focus on deriving generic approaches for mapping from IFC/Express to UML/GML.   This is a particularly important topic for anyone working in GeoBIM as to date many ad-hoc/one-off conversions have been attempted but this paper presents a more systematic approach in that it details how to handle the various packages, data types, enumerations and so forth.  

 

The approach has benefits both for the fact that mapping between the standards is a very complex task due to the number of classes on each side (the results of which could of course not be presented in a  single paper) and also because as the standards are evolving any mapping will need to evolve in parallel. Importantly, the problem is addressed at conceptual level which means that the results may be widely applicable to different implementations.   It is also good that the authors provided a full implementation: https://github.com/jetgeo/IFC2GML and the authors are to be commended on the hard work that must have been undertaken to get to this point.    The results themselves are also well presented with examples relating to specific classes.

 

However, the paper in its current form is not suitable for publication in a journal - in particular as it doesn't take into account a broader readership but also because the sturcutre does not follow a scientific standard.  The following changes are required.

 

 

1. The work needs to be better and more clearly motivated in the introduction - as well as the opportunities arising from the generation of data in both IFC and TC/211 standards, there needs to be a clear, real world, explanation as to why we would want to exchange information between the two standards - who will benefit, give some examples.   (This should extend the paragraph starting "Integration of BIM and GIS is considered a promising topic")

 

 

2. As with standard scientific papers, the literature review should directly follow the introduction. 

 

3. The literature review should also include some history of each of the standards - in particular the context in which they evolved and who uses them/who are the planned users.  This would help to clarify why the standards evolved in parallel as well as some of the challenges that will be encountered when trying to undertake data exchange.

 

3. Basic concepts should be explained for a more general readership - e.g. what is UML, what is EXPRESS - including some simple examples

 

4. "Contrary, the Unified Modelling Language" why 'contrary' ? 

 

5. Rather than just mention a 'range of standards' from TC 211 please clearly list and describe the relevant ones (forward reference to literature review for this).

 

6. Merge section 2 Contribution and Research Questions into section 1 and also give an explanation as to where these questions derived from - what gaps in knowledge exist

 

7. "handling of solid geometry in BIM and GIS are considered out of scope."  - this is fair enough as it is a topic in its own right, but please cite a few papers for the interested reader to look at if they wish.

 

8. "through a conversion from IFC EXPRESS schemas to UML models according to ISO/TC 211 standards and an inverse conversion to implementation schemas for EXPRESS and GML." is not sufficient in terms of method - was this a top down approach, a bottom up approach, a systematic read through, an implementation?  Remember the idea is that you need to provide sufficient detail for someone to be able to reproduce what you have done independently.  YTou start to do this in Figure 1 but all figures should also be accompanied by a written explanation and detail.

 

9. State of the art - literature review - should be before method to follow standard structure for a scientific paper.

 

10. A more fundamental comparison should relate core concepts from IFC to equivalent concepts from ISO/TC 211 standards - I think I understand what you mean with this statement, but it makes no sense to the reader if you haven't first described the work of TC 211  and in particular what you perceive as the 'concepts' - which I assume are things that span multiple standards.  This is fundamental as what you are trying to do is compare - at conceptual level - BIM and GIS rather than just specific standards.       Again, this comparison only makes sense if you first give the reader some history of how the two disciplines evolved in parallel.

 

11. Section 4.2 is important - but too detailed for the average reader so lead the reader in gently.  You can't assume that the reader will know the detail of the versions you are describing.

 

12. You mention that UML will be a future modelling language for IFC - in the discussion, please comment on how this impacts your work? 

 

13. Basic notation on all diagrams should be explained

 

14. Geometry - the paragraph after Figure 15 - deserves its own sub heading as it is an important topic in its own right

 

15. Statements such as "Aggregation datatypes have no equivalent concept in UML" are a little confusing - if you are working at conceptual level, then the concept of aggregation exists, and the fact that UML isn't able to model it is a limitation when using UML for modelling.    Please comment on this in relation to the statement where you say that IFC future implementations may be in UML - does it mean that aggregation data types will be lost as an option? (IFC-UML)

 

 

16. In the discussion, the authors should refer to two additional elements that are not considered in the paper

- as is done currently, re-address the research questions

- firstly, how flexible is this approach in terms of real world data - i.e. logical and physical level - that may contain errors and have data quality issues.  What implementation problems are foreseen, how can they be overcome?

- secondly, the paper talks about information exchange between standards - is a natural next step a merged standard?  If so, why, if not why not?

- thirdly, how does this work feed back to the standards-setting initiatives on both sides  - and how should standards setters move forward when evolving these standards to ensure better integration, is the plan to publish this integration with some automated tools so that it can be used widely?

- fourthly, link back to the motivation - how will this work help the users that are mentioned in the first section, what level of expertise is required for the users to use this approach? 

- fifthly - is the conversion bi-directional?  If not, why not?

Author Response

Thank you for your kind words and constructive comments. Please see the attached file for our replies.

Author Response File: Author Response.pdf

Reviewer 2 Report

This paper proposed conversion patterns between IFC EXPRESS schemas and ISO/TC 211 UML models for their interoperability.
The authors addressed relationships between concepts of each model and tried to design mapping rules between IFC EXPRESS and ISO/TC 211 UML. The study is positioned with a strong background in two-domain standards of BIM and GIS. Also, it provides brief information about two models to make readers understand their contribution. Easy to read. 
However, there are three concerns for acceptance.
1) The main issue is how to evaluate the method. The current manuscript does not show any evaluation of the technique. Even though the reviewer can find the contribution of this study, it is difficult to capture the improvement compared with previous approaches. In Figure 1, the authors mentioned the evaluation, but they don't address it.
2) On page 5, the authors state that IFC-UML and ISO/TC 211 are still heterogeneous due to different usage of the modeling language. Please describe some examples.
3) Finally, what is the main benefit of converting models in the conceptual models? Still, it isn't straightforward to understand why this method provides improved interoperability.

Minor comments: I cannot follow the explanation about Figure 17 and 18. Are they drawn by ShapeChange? 

Author Response

Thank you for your kind words and constructive comments. Please see the attached file for our replies.

Author Response File: Author Response.pdf

Reviewer 3 Report

Thank you for the opportunity to review this paper. It is an exciting piece of work, but I have some fundamental concerns about the article.
My first concern is around the motivation of the paper. We all agree that BIM and GIS integration is essential. To this end, there has been plenty of research on trying to map or unify BIM and GIS data formats. The question is, what is the significance of integration at the modelling language level. What problem we would solve by mapping a GIS specific UML and to a BIM particular EXPRESS. Why is this level of integration better than, e.g. IFC and CityGML integration? Why not using a database integration approach?
My second concern is around the way research questions are formulated. These questions do not lead to objective and measurable answers. Take the first question. The answer to this question is obvious, it is possible, but what is more interesting is what is the advantage/implication/limitation of such conversion.
My last concern is around the evaluation of the developed conversion approach. It appears only IfcAligment and IfcAlignmentCurve are used to testing the conversion. Is this test representative for the rest of the elements/objects, based on what proven methodology?

Author Response

Thank you for your kind words and constructive comments. Please see the attached file for our replies.

Author Response File: Author Response.pdf

Reviewer 4 Report

This draft reads more like a PhD thesis, from the paper structure to the provided level of details. It should be thoroughly repacked for a journal publication. 

Besides, I can hardly be convinced by the significance of this work. BIM and GIS integration has been well discussed for over a decade, there is nothing new reported in this paper. The introduction and methodology parts are weak and pale and not clearly state why authors are spending time on this work and why the devised research questions are important. Using UML for depicting IFC is ubiquitous and almost all papers related to BIM/GIS integration adopt UML to clarify, relate and map the differences data model structures.  The main body of this paper locates in Section 5 which contains three sub-sections. 5.1 ("Conversion of IFC models to ISO/TC 211 compliant UML models") and 5.2 ("Relating the IFC model to abstract concepts from ISO/TC 211 standards"), and I reckon that removing section 5.1 will not impact the readability of this paper at all. 

Given these, I think this paper is far from a good shape for journal publication.

Author Response

Thank you for your comments. Please see the attached file for our replies.

Author Response File: Author Response.pdf

Reviewer 5 Report

This paper presents a research projects for the integration of BIM and GIS that turns out to be a very relevant and current topic.

The aim of the article is to improve interoperability between BIM and GIS with the scope to investigate whether conceptual UML models for BIM and GIS can be a way forward for improved interoperability between the two application domain.

The conversion pattern developed for conversion from IFC information models to an UML model with the aim to reduce heterogeneity between BIM and GIS by moving IFC schema into the same UML environment as GIS information model is really interesting, especially the in-depth study of the express scheme and its internal composition in order to understand how to best relate the express concepts with the UML ones.

The paper is well structured and the methodology used is very clear. The state of the art and the bibliography are and very thorough and exhaustive

Author Response

Thank you for your kind words. We have rewritten and improved the abstract, introduction, discussion and conclusions based on comments from other reviewers.
Thank you for your time and consideration!

Reviewer 6 Report

# overview

The authors describe how IFC schemas can be converted to UML that are ISO/TC211-compliant. They describe the mappings they did and show how from those GML schemas (.xsd) can be derived.

The paper is clear and well-written, and there's a scientific contribution.

That being said, I am struggling to write this review. I've been thinking about it for a while, and I think my main issue with the work done is: "but what is it good for?". I understand that many use UMLs and having UMLs for IFC would help undertand the issues, but in practice, will it really help practitioners dealing with IFC and 3D GIS datasets?

I guess that sentence on p.2 "The scope is to investigate whether harmonized conceptual UML models for BIM and GIS can be a way forward for improved interoperability between the two application domains." hasn't been totally answered for me after reading the paper...

Those comments could be used by the authors to better specify and explain why this is important. I deal with IFC and CityGML myself, and I am not convinced. But I could be :)

# Some specific comments

p.24: I would generally disagree that statement: "Still, most studies have investigated integration at the data instance level and not at the conceptual level where the semantics are defined"
--> Many have done it, with a specific data model that is "comparable" to that of IFC (eg CityGML, or LandInfra).

p.4: "However, CityGML is only one of several application schemas based on core abstract concepts from ISO/TC 211 standards."
--> but if CityGML mappings were done, isn't it because the generic ISO concepts are too generic? mapping generic to generic would make sense, but here you have IFC to generic. Why not CityGML?

p.4: some papers are missing in my opinion, eg [Nagel07], [El-Makewy12], and [Donkers16]. These all perform mappings from the schemas of IFC to CityGML, but they are relevant.

p.26: "A shared understanding of the semantics for digital representation of real-world phenomena is more important than the conversion between modelling languages."
--> but only if that helps concretely in practice I would add. And here, I am not convinced. [Donkers16] among others show how difficult it is to convert IFC files to CityGML: how will your work help them?

# References

@misc{Nagel07,
Author = {Nagel, Claus and Kolbe, T. H.},
Howpublished = {Presentation OGC 3DIM WG},
Title = {Conversion of {IFC} to {CityGML}},
Year = {2007}}

@article{El-Mekawy12,
__Markedentry = {[Sjors:4]},
Author = {El-Mekawy, Mohamed and {\"O}stman, Anders and Hijazi, Ihab},
Date-Modified = {2014.08.29},
Journal = {International Journal of Advanced Computer Science and Applications},
Number = {5},
Pages = {159--171},
Title = {An Evaluation of {IFC-CityGML} Unidirectional Conversion},
Volume = {3},
Year = {2012}}

@article{Donkers16,
Author = {Donkers, Sjors and Ledoux, Hugo and Zhao, Junqiao and Stoter, Jantien},
Journal = {Transactions in GIS},
Number = {4},
Pages = {547--569},
Title = {Automatic conversion of {IFC} datasets to geometrically and semantically correct {CityGML} {LOD3} buildings},
Volume = {20},
Year = {2016}}

Author Response

Thank you for your kind words and constructive comments. Please see the attached file for our replies.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Dear Authors

Thank you for the revised version of the paper, in which many of my comments were addressed.  However, there are still some things to sort out before this paper can be published:

1.  You did change the structure of the paper, but not as required. A standard paper has an introduction and then literature review.  As per the previous comments "the literature review should directly follow the introduction. "   Your introduction should motivate the work and set the research questions - the reader shouldn't have to read to far into the paper to see these questions and thus be able to decide if it is worth their while to read the rest.   The literature review then summarises and synthesizes the current state of the art in the topic.  Please stick to this expected structure.  Materials and method should be Section 3.     2. As requested, please put legends (as a small box somewhere on the figure, much as you would do for a map) in the figures to explain the notation for those not familiar  in this case with UML - e.g. what do the boxes mean, what do the arrows mean, what do dashed arrows mean, what do the diamonds mean.  Do this for all figures in the paper.      3.  The work still needs to be better and more clearly motivated in the introduction - as well as the opportunities arising from the generation of data in both IFC and TC/211 standards, there needs to be a clear, real world, explanation as to why we would want to exchange information between the two standards - who will benefit, give some examples.   What applications will use this (you can look at applications using GeoBIM for examples - planning and asset management, site security, cadastres and so forth are some topics that are relevant)

 

Author Response

Thank you for your review and the suggestions for further improvements. We have considered your comments and modified the document accordingly. 

The following list describes our response to each of your comments.

1. On the structure: We have moved and modified the literature review to section 2, as requested. 
2. On legends in UML diagrams: The UML diagrams are package and class diagrams following the standardized UML notation, as defined in the UML Specification [1] and ISO 19505-2:2012 [2]. It is not common practice to include legends in UML Class diagrams. Instead, we have added references to supporting literature on UML in the methodology description (section 3). Besides, we have defined the type of UML diagram in each figure text.
3. On motivation: We have extended the description of the motivation for integration in section 1, as requested. 

References:

1. Object Management Group, Unified Modelling Language Specification Version 2.5.1. 2017, Object Management Group: Needham, MA, USA.
2. ISO/IEC JTC 1/SC 7, ISO/IEC 19505-2:2012 Information technology — Object Management Group Unified Modeling Language (OMG UML) — Part 2: Superstructure. 2012, ISO: Geneva, Switzerland.

Reviewer 3 Report

Thank you for addressing my comments.

Author Response

Thank you for your review. We have modified the introduction and the literature review based on feedback from other reviewers. Besides, we have checked the language carefully.