Geospatial Semantics

Author: John Forne
Date: 10 September 2009

Kia ora tatou

Yesterday's workshop at the Auckland University on Geospatial Semantics run by the Centre for eResearch was very interesting and highlights the need for a co-ordinated approach to achieve interoperability.

If "interoperability" isn't enough of a mouthful - try "semantic interoperability". This was the theme of yesterday's workshop. So what is semantic interoperability? Let's unravel the term...

Broadly, interoperability is a concept that describes a state when things, such as spatial information, are able to be used in different systems (to inter-operate). For example, in a computing context, interoperability would be achieved when data is able to be shared across different systems, such as between government agencies. There are different levels of interoperability. For example, we might both be able to display a PowerPoint presentation on screen (data services)... But the presentation is in French and I can't read French (data content). Semantics is at this data content level.

Semantics is harder to define neatly, but refers to the study of meaning, particularly the study of signs and symbols to represent other concepts and entities as used by agents or communities within particular contexts. Semantics studies how we describe, represent, or structure what we're talking about. For example, whether we talk about volcanic rocks or igneous extrusive rocks.

Ok... having got that semantic task cleared up... what were some of the key things that I took away from the day?

• The study of semantics is highly esoteric - but is very interesting and is important. We need more presenters like Boyan Brodaric and Mark Gahegan to (ironically) demystify what is a potentially inaccessible field with specialised language!
• Developing a semantic structure is vital for interoperability - but not sufficient. We need to co-ordinate our semantics. A standard is of limited value - unless we all use the same standard. We have a challenge to co-ordinate our efforts.
• Semantic interoperability offers great potential - but we need to recognise that there are limits to trying to impose a structure on the infinite complexity of the real world and, particularly our subjective representation of this world. We need to develop semantic models for data that somehow reflect the diversity of models that different people create to represent their worlds.
• The divide between a metadata-based approach (like an old fashioned library catalogue) to finding, organising, structuring data and a semantic-based (like tagging meaning to data) approach is not black and white. Metadata and semantics are complimentary. It's not one approach or the other, certainly given the current state of technological developments and uptake of semantic approach. Metadata is useful for cataloguing data.

A copy of Boyan Brodaric's keynote presentation is available below. Boyan's presentation provides a nice introduction to semantics and provides some good examples of semantics in geology.

• Semantics and the Geosciences: an introduction with examples from geology (PDF 2.2MB)

For a full list of speakers, see the workshop agenda (PDF 76KB). I'll update this post with a link to the other presentations once copies of them are made available.