News Explore the world of Geo-information

Frans Knibbe

Space, the final frontier?

1 November 2016

Making space for space on the web

In our enlightened future there will be universal sharing of spatial data on the world wide web. Just how close or distant is that future? At this point in time, we are not there yet – there is no agreement on how to work with spatial data on the web. Perhaps the way to express space will be the last frontier that needs to be crossed before the world can enjoy a truly interoperable web of data? This blog explores the unifying potential of a new ontology for spatial data.

Recently I had the privilege of taking part of discussions on how to make the web a better place for data, at the yearly W3C conference TPAC. That goal is noble, because sharing data on the web can bring humanity the knowledge and wisdom that are still so much needed. Much progress has been made in finding good ways to share raw data on the world wide web, in such a way that data can be easily found, understood, processed and combined. Some recent evidence takes the form of the Data on the Web Best Practices, a candidate Recommendation from the W3C, and a guide from Google on how to annotate datasets with terms from to improve discoverability: Science Datasets. For spatial data in particular the OGC/W3C Spatial Data on the Web Working Group (SDWWG), in which Geodan proudly participates, is on its way to extend the Data on the Web Best Practices with Best Practices for Spatial Data on the Web.

There is something funny about spatial data on the web. There is no shortage of practices to evaluate when looking for best practices. One only has to look at how geographical data are put on the web to see that there are numerous ways of making spatial data available. But it does not end there, because working with spatial data is not the exclusive privilege of geographers. Space is present in almost all domains of human enterprise in one form or another, and because of that many other ways of expressing spatial information have been devised. Altogether, the amount of standards and practices for expressing spatial data is staggering.
Spatial data are everywhere, and it would be a shame for all those useful data to exist in isolated corners of the web, without the ability of linking and combining. Space, like time, is a universal phenomenon that is present everywhere and therefore has great integrative potential. Should we find a way to put all diverse spatial data in the same cyberspace, that would truly provide a dataverse worthy of bold exploration. But spatial data on the web nowadays use many standards, models and formats, with varying degrees of web-friendliness. Space, unfortunately, is not really a part of the web yet.

A spatial ontology

The challenge seems to be to come up with a universal model of space that is simple, powerful and ready for the web, something that everyone can use with as few adaptations as possible. As development of an agreed spatial ontology is part of its charter, the SDWWG has started to work on an ontology for spatial data on the web. It is intended to take the form of a new version of the GeoSPARQL ontology. The new ontology will extend GeoSPARQL with more rigorous definitions of the core concepts involved in spatial data. Such web definitions could provide solid foundations for things like

  • data exchange formats
  • data types for persistent storage
  • definitions of spatial functions and spatial filters
  • universal (i.e. not application-specific or domain-specific) APIs.

To avoid the trap of creating yet another standard that only adds one more alternative to a multitude of standards, the aim is to align with existing standards as much as possible. Such existing standards can be found both in the realm of OGC (or ISO/TC 211) and the web domain.
The spatial ontology is in development. You can get an idea of that development from this WebProtégé project. Interestingly enough, only two core concepts (or classes, if you like) seem to be needed for a functional multi-purpose spatial ontology. They are the notions of a spatial thing and a geometry.

Spatial thing

A spatial thing can be defined as something that has some kind of presence or extent in space. That space could be one, two or three dimensional and could be virtual or real. Examples of spatial things are the planet Saturn, the Voyager 2 spacecraft, the Eiffel tower, a drawing on paper, the Earth’s magnetic field, an amoeba, the lost island of Atlantis and you. Ultimately it is up to whomever makes data available to decide if it makes sense to see something as a spatial thing.

An important property of a spatial thing is that it can have a spatial relationship with another spatial thing. Different types of spatial relationship can be discerned. Topological relationships are one kind: spatial thing A can be within spatial thing B, or adjacent to spatial thing C. Another kind is distance: spatial thing A can be 524 metres from spatial thing B. Or, a bit more vague, it can be far away from spatial thing C. Lastly, there is the directional relationship: spatial thing A can be north of spatial thing B, or upstairs from spatial thing C.
It is not hard to see that the ability to express or infer spatial relationships between things can do a lot of good for data connectivity on the web.

A spatial thing can be identified by some sort of coding system, for example a toponym or a street address. And a spatial thing can be modelled by one or more geometries, which leads to the second core concept.


A geometry could be defined as an ordered set of n-dimensional points that can be used to model the shape or location of a spatial thing. Geometries can have the following properties, which may or may not be known:

  • A set of coordinates
  • A reference to a Coordinate Reference System (geographic or other)
  • Dimensionality (a geometry can be one, two or three dimensional)
  • A type (e.g. point, multipoint, line, polygon, multicurve, …)
  • A level of detail or resolution (when the geometry is meant as a model of a spatial thing).

Coordinates consist of numbers so computers can do a lot of useful things with geometries. Many kinds of calculations can be performed on them, and geometry is easy to display, for example on a map, or in Virtual Reality or Augmented Reality environments. Like spatial things, geometries can have spatial relationships too, but for geometries they are computable. With geometric data, computers can calculate the distance between geometry A and geometry B, or calculate whether geometry A is adjacent to geometry B.

Simple interoperability

It is easy to perceive spatial data as complex, and in truth many complicated spatial models exist. But working with spatial data on the web can be simple without making choices that hamper interoperability. The spatial ontology in development will probably contain much more than just definitions of the concepts of spatial thing and geometry. But as with all web vocabularies, you can choose which part of the model you want to use. Just saying that something is a spatial thing, or is a subclass of a spatial thing, can be a powerful statement, especially in a world where the business of automated data analysis is booming. Likewise, having a basic online definition of the concept of geometry can do a lot for possibilities of combining geometric data that are available in many different formats.

Mankind is on the brink of leaving the confines of our home world and truly begin human space exploration. Will we have been able to truly accommodate spatial data in our information space before that time?