Cities use a variety of metrics to evaluate themselves. With the introduction of ISO 37120, that contains 100 indicators for measuring a city’s quality of life and sustainability, it is now possible to
consistently measure and compare cities, assuming they adhere to the standard. With the adoption of Open Data policies, cities are now publishing vast amounts of data which can potentially be used to analyse their performance. But much of this open data lacks standard formats and
vocabularies to enable cross city analysis. Nor do tools exist for automating the analysis of this data.
The goal of this research is to automate the longitudinal analysis (i.e., how and why a city’s indicators change over time) and transversal analysis (i.e., how and why cities differ from each other at the same time), in order to discover the root causes of differences. In
particular, to develop a system that takes as input:
- All of the information and knowledge with respect to an indicator,
- A set of consistency axioms,
- A set of diagnosis axioms, and applies the axioms to determine why indicators change.
1. Develop ontologies for representing the complete definition of each indicator in the ISO 37120 standard.
2. Represent each city’s indicator value (for a particular year), including the supporting data used to derive it, using the aforementioned ontologies.
3. Develop ontologies for representing indicator theme-specific knowledge, such as basic knowledge about education, such as school, teacher, student, grade, etc.
4. Represent a city's theme-specific indicator knowledge. For example, in order to diagnose the education theme of indicators, PolisGnosis needs to understand concepts such as what grades comprise a Primary school in the city, and what schools are defined as Public schools.
5. Develop axioms to determine the consistency of indicators. 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; and
6. Develop a reusable, interoperable logical theory of longitudinal and transversal analysis of Metrics.
The general method we will use to create the ontology is the methodology defined in Grüninger & Fox (1995). The process begins by defining a set of usage scenarios. Based on the scenarios, we identify a set of competency questions that the ontology must answer. These are the requirements for what is to be represented and the deductions to be performed. Next, the terminology (i.e., classes and properties) required to represent the competency questions are defined. Next we will specify the semantics of the terminology by constructing a set of axioms that
define and/or constrain the interpretation of the classes and properties. The axioms are important as they precisely define the Metrics, and can determine whether the data that underlies the Metrics
are consistent (e.g., the time periods during which the student and teacher populations are the same). The ontologies will be defined using First Order Logic, and will be made available on the web using the Web standard OWL and the axioms will be implemented in Prolog.
An ontology-based representation of a Metric can be depicted as a labeled, directed graph. For longitudinal analysis we are analyzing two sub-graphs that represent the same Metric (and its supporting sub-metrics) for the same system but for two different time periods. For
transversal analysis we will be analyzing two sub-graphs for the same Metric for the same time period but for two different systems of the same class. Our approach to automating comparative analysis is to characterize the change between two subgraphs in the form of consistency axioms. For example for longitudinal analysis, a growth in a ratio metric (such as “Student Teacher Ratio”) is consistent with the rate of change in the numerator being greater than the rate of change of the denominator (i.e., the growth in students over time exceeds the growth of teachers). With this knowledge, we would then have to move the analysis down to the numerator to understand why it is increasing so quickly, or the denominator to see why it is not increasing as quickly. For each type of knowledge used to represent a Metric, such as Populations, placenames, statistics, etc., comparative analysis consistency axioms will be defined, in order for analysis to drill down to root causes of change. An important aspect is determining which inconsistencies are significant. Where possible, sensitivity analysis can be used to distinguish significant changes from insignificant
Strong, successful cities are prerequisites for a strong, vibrant nation. In Canada, over 80% of Canadians currently live in urban areas. The sustainability of urban regions (in environmental, social and fiscal terms), and their resilience to both man-made and natural large-scale events, are thus of first-order national importance to our well-being and quality of life. Dealing intelligently and effectively with the complex problems of urban management and design requires use of the best information, analysis and decision-support tools available. The representation and analysis of system performance metrics is a critical
This project will make both theoretical and applied contributions. Its theoretical contributions are in the areas of 1) ontologies for system metrics that integrate and extend foundational ontologies, and 2) automating comparative analysis. By design, foundational ontologies are internally consistent. Our work will discover the consistency axioms necessary to integrate them into a system, and extend them to include missing concepts. Secondly, we will extend the theory of comparative analysis to heterogeneous models that capture the richness of the underlying domain’s Metrics. Hence, the project’s contributions include:
- The formal definition of Metrics that provide precise, unambiguous and machine interpretable definitions of system performance,
- The ability to perform consistency and comparative analyses of Metrics based upon their definitions (as opposed to ad hoc softwareimplementations), and
- The standardization of terminology thereby enabling cities to publish their open data using common vocabularies or ontologies.