A Foundation Ontology for Global City Indicators

Tracking #: 905-2116

Mark Fox

Responsible editor: 
Guest Editors Smart Cities 2014

Submission type: 
Full Paper
This paper defines a generic, reusable ontology that forms the foundation for representing the definitions of ISO37120 Global City Indicators and their instantiation on the Semantic Web by cities. It makes two contributions. The first contribution is to the field of city indicators. The ontology makes it possible to: • create precise definitions of indicators thereby reducing the ambiguity of their interpretation and making them computationally accessible; • represent the “data behind the data” or supporting data, enabling drilling down; • determine the consistency of metrics. E.g., is the supporting data of the same scale, refer to the same place, measured in the same way, covering the same time period, etc.; and • achieve interoperability, namely the ability to access, understand, merge and use measurement data available from datasets spread across the Semantic Web by providing a standard representation using OWL 2. The second contribution is the successful integration of seven basic ontologies and their extension with trans-foundation axioms. The ontology spans analytical models (e.g., ratios), statistical models (e.g., population measurements), geo-spatial models (e.g., city boundaries), temporal models (e.g., time periods) and description logic models (e.g., definitions of students, teachers, etc.). It also integrates meta-knowledge such as provenance, validity and trust.
Full PDF Version: 

Major Revision

Solicited Reviews:
Click to Expand/Collapse
Review #1
By Raúl García-Castro submitted on 04/Dec/2014
Major Revision
Review Comment:

The paper describes an ontology for city indicators; the representation of such indicators is currently a very relevant issue.

The paper describes in detail the main modules of the ontology through a concrete example (using the Student/teacher ratio indicator). Furthermore, it defines a set of consistency rules that can be used to validate data represented according to the ontology. Finally, it evaluates the ontology according to different aspects.

Things that I like about the paper is the holistic view of an indicator as more than just a number and that it presents interesting issues to take into account (city boundary changes, validity, etc.).

The paper reads well, but it has more of technical report than of research paper. It is too long for just including a description of an ontology, because it is too verbose with detailed descriptions of every component in the ontology.

One problem that the ontology has is that it considers that an indicator is a number and that the only measurement scales to take into account are interval and ratio. This is not correct, the other two existing measurement scales (nominal and ordinal) can also be applied to cities and must be taken into account when defining an ontology like this. For example (just out of my mind):
- Nominal: Define the city you visited with one world ("clean", "dangerous").
- Ordinal: Rate the noise in your city from 1 to 5.

One effect of taking this into account (and that should be mentioned in the paper) is that units of measurement are related to physical quantities (i.e., only interval and ratio scales).

A proper time interval is not an interval where the start time is less than the end time; it is an interval where the extremes are different (for example, negative interval durations are possible).

In general, the paper presents the outcome of the ontology development process (i.e., the final ontology obtained) but it does not discuss any decisions taken or modelling alternatives.

Related to this, there should be other ontologies/models that have the goal of representing indicators in cities. However, the paper does not discuss them and compares with them (e.g., what is new/better?, do all of them follow the same modelling principles?).

The consistency rules in section 6 are totally specific to the example (Student/teacher ratio). However, some of them are or could be easily generalized. Instead of these specific rules, the paper should provide generic rules that can be applied in general to indicator data generated according to the ontology. Clearly, these rules will also apply to the example, but the interesting contribution here is that they are general for any (or a broad set of) indicator.

The evaluation of the ontology is quite poor. Apart from the use of a reasoner for evaluating its consistency, the evaluations presented are of qualitative nature (subjective). Besides, one important aspect regarding the ontology is whether it has already been used in some practical use case; right now this is not clear.

Some tips to make the paper shorter and to seem more as a research paper:

.- To avoid tech-report elements (e.g., boxes with questions in section 4).
.- Some figures that are presented in an incremental order could be removed, leaving only the final figure.
.- Captions must be included in figures.

Some writing issues:
.- Page 2. "or against itself" -> "or against a city itself"
.- Page 6. "an indicator in measurement ontology" -> "an indicator in a measurement ontology"
.- Page 7. Table. "On tology" -> "Ontology"
.- Page 19. "among its supporting data" -> "among their supporting data"

Review #2
Anonymous submitted on 22/Dec/2014
Major Revision
Review Comment:

I am not an expert in ontology or the semantic web. But I am a GIscientist with interests in urban measurement. I cannot vouch for the technical qualities of the urban indicator ontology developed in this paper. I can say that standard, computable definitions for urban indicators is a valuable and timely development.

From a user perspective, it is unclear to me how well this ontology can represent higher-order concepts. When I read this paper, I enjoyed the opening example of student-teacher ratio: it nicely illustrates how a seemingly simple concept is fraught with measurement ambiguity. However, I was disappointed as this example continued as the central focus of the paper. While student-teacher ratio is intricate, it is not as complex as concepts such as "congestion," "crime" or "poverty" that have multiple layers of measurement as well as ambiguous, even contested, semantics. Perhaps this is trivial, but a stronger motivating example would be more convincing (at least to a user).

Another unclear aspect is how to support the assembly of individual indicators into composite indicators of multidimensional concepts such as "quality of life." This requires representing concepts such as attribute weights, higher-level dimensions and trade-offs (compensatory, non-compensatory). There is a substantial literature on environmental and quality of life indicator construction using the tools of multicriteria analysis.

Minor comment - the author is opening himself to perhaps endless streams of critical deconstruction with the comment about taking urban indicators out of the realm of humans and into the realm of computers. I also think this is unfair: ultimately, this is all about human decision support - not everything will be automated. I hope.

Review #3
By Mathieu d’Aquin submitted on 18/Jan/2015
Major Revision
Review Comment:

This paper describes the building of an ontology of city indicators, which are based on existing, although not formalised standard definitions. The ontology is certainly interesting and I can see some use for it. The main claimed contribution are the ones generally associated with building an ontology in any domain. My main issue is therefore the lack of validation of these claims. Indeed, while again the ontology seems mostly valid from what is said in the paper, that it facilitates interoperability and traceability in the domain can really only be verified once it has been adopted and tested on real cases. Unfortunately, the paper only provides a toy example. As usual for this type of papers (which should really be an "ontology paper", rather than a full research paper) the plans to make the ontology adopted and used should be made explicit, so that the potential impact could be assessed.

This a paper describing an ontology for a particular domain, like there are many others. I am not aware of other ontologies of city indicators even if the work is based on an existing set of standardised indicators. Some indication of the methodology and the requirements used to build the ontology would have been useful, as compared to similar ontologies (i.e. making it more specific than a broad list of expected benefits).

Again here, this is hard to assess. There are a lot of cities in the world, but whether or not they need this ontology is unclear. The efforts it requires to adopt it are not really detailed, and in the absence of real adoption, nothing can really be said about the benefits to real users.

Quality of writing:
The paper is easy to follow and the design choices made clearly explained. It is not in the layout expected by the journal, but that is easily fixable.

- I would also make a reference to RDF Data Cube, which share a lot of common aspects with the GCI ontology
- A link to the ontology and to examples of use would be good
- As already mentioned, you should consider making this an ontology paper