Everything You Wanted To Know About Ontology But Were Afraid To Ask

Image Courtesy http://www.opte.org/

Image Courtsey http://www.opte.org/

In the coming year, you will be hearing a lot more about ontology. If you’re in the search & discovery or local media business you probably already are and if not, then read on so you are the one ahead of the curve and not the one behind it.

Ontology is about understanding relationships. For search & discovery, it’s about providing increasingly relevant results by understanding your relationships and context around whatever it is you’re looking for.

According to Wikipedia

An ontology renders shared vocabulary and taxonomy which models a domain with the definition of objects and/or concepts and their properties and relations.

More simply put, an ontology allows us to create a virtual model of relationships within a domain. A domain is just about anything you want to build a formalized relationship map of and may be specific or very general. For example, you could create an ontology of:

Building ontologies can be complex and often falls to those with a deep understanding of library sciences, qualitative modeling or database design, though this isn’t your fathers normalization!

We build ontologies because it allows us to formalize and reuse knowledge about a specific domain making it much easier for various applications to work together and for larger more complex relationships to be understood, so you get the best possible search results.

Ontologies are gaining attention because of the need to deliver relevant results while sifting through an ever-increasing universe of information. The recently released facebook graph is an example of a company trying to meet that need by building a social ontology.

Ontologies also play a part in what’s become known as the semantic web.

Let’s take a look at an example to clarify things.

When searching for “cappuccino” most humans would know that a coffee shop is a good place to go. But for a computer that distinction is not so simple and if taken in completely literal terms, you might not get any search results when looking for “cappuccino.”

Through ontology driven search “cappuccino” is one of the many taxonomy terms related to coffee (as well as milk, caffeine  and a host of other things) so you can get your caffeine fix.

As our graph continues to expand our possible search results get ever more complex. But through the use of ontologies, the search engine is also able to return relevant search results by combining the ontology data with things like location, the time of the day (is the coffee shop open?) and even who my friends are (a social ontology); because I’d like to know which coffee shop my over-caffeinated coffee junkie friends say has the best cappuccino.

In this scenario, we’re combining ontologies into a larger graph that allows us to create a highly relevant set of search results for you.

To get a more humorous example of why the ontologies and the semantic web are so important just consider the case of “semi skimmed-milk” vs “milk skimmed-semi” (see below).

One Response to “Everything You Wanted To Know About Ontology But Were Afraid To Ask”

  1. Ian Oliver says:

    Great stuff, and for those of you working with ontologies, modelling and visualisation is often the hard part. There’s some links here for those of you who are interested: http://ijosblog.blogspot.fi/2011/11/visual-modelling-and-ontologies.html and I recommend you check out some of the work by John Howse and Gem Stapleton et al, from this year’s ISWC conference.

    We can actually go quite far beyond just simple hierarchical classification systems to expressing relationship rules and other kinds of constraints and situations and thus take advantage of some of the more expressive and interesting description logic foundations for building ontologies, for example, we have used this to express and communicate an ontology of “data collection” types for privacy: http://ijosblog.blogspot.fi/2013/06/data-collection.html

Leave a Reply