Beyon Here Be Dragons: Ontology and Semantics


On old maps, when a place was unknown and could not be drawn, the cartographer would write “Beyond Here Be Dragons” and draw in a picture of the supposed dragons of that place. Unknown places were dangerous and best avoided. Mapping data into models has its dragons, unfortunately I know just where most of them are lurking; I just don’t know when and how hard they will bite. The particular dragons I am staring at are the ones that lie at the philosophical roots of science, particularly the dragons of epistemology and ontology.
Barry Smith, who is the philosopher behind the OBO ontology firmly believes that a scientific ontology must be rooted in Aristotilian realism; something is actually “out there” and the ontological categories we use to describe it exist ‘out there’; we discover them, we do not create the categories. He wants ontologies to be objective, and not subjective structures that exist only in our minds. Our categories may be wrong, and need to be corrected when new scientific evidence comes in, but I am describing something that exists not just my personal mental state. Barry wants to say “the temperature is 24C”, not just “I perceive that the temperature is 24C”.
A big issue here is stability, science should be stable for research data to be usable over time. Subjective ontologies, like departmental organizational charts, or research procedures are constantly changing, and require massive data updates to accommodate them. Or even worse, if the data cannot be changed, it all ends up being a wasted effort.
These issues have consequences. There have been two main schools of ontology; that proposed by the W3C and the OBO. The W3C approach was to let everyone write their own ontology piece, and put them all together later (“Let a thousand ontologies bloom”)*. To my knowledge, nothing useful has come out of this. The OBO is painfully disciplined and still not complete (the Plant Ontology:PO has not be officially approved), however it has had big bucks put into it, and significant pay-off in tagging medical documents and molecular genetics.
To create an objective, scientific ontology is a noble goal; but I am not sure it is entirely honest and achievable. Saying that the purpose of a hammer is to drive nails (and not crack walnuts) is moving toward the intention of the creator’s design, or at least the agreed intention of a group of carpenters. The boundaries between objective and subjective are not as clear as some would like.

So, where are the dragons?
There are the technical dragons (see: Is it Mature Yet?) such as defining growth stages and how to compare lettuce to tomatoes. These are a nuisance, and while difficult to deal with they are not dangerous. The political ones are what scare me, and why I almost did not write this. While Barry Smith talks about a return to Aristotilian Realism, I don’t actually think that he philosophically wants to go that far back, I actually think he is more of a rationalist modernist; believing that through careful definitions we can build a foundation for knowledge (at least objective, scientific knowledge); unlike Aristotle, he carefully wants to avoid values and ethics in the ontology. The danger is the current war between the philosophy of Modernism (and modern science) and Post-Modernism, and it is precisely a battle over semantics and ontology. Are there ‘real’ categories, or are all semantics just manipulations for power? To put my foot in the middle of the dragon droppings, this is the issue of Post-Modern gender identity; are there objective ‘normative’ categories (regardless of now many exception cases we may allow), or is everyone’s personal perception the only reality (everything is a ‘category of one’)? If we agree that we will restrict our travels to growing plants in controlled environments, and the scientific assumptions needed for that; I think we can avoid waking on the dragons that sleep outside.

Just as we want to avoid flame wars over organic and GMO issues, we need to be aware that we need to tread carefully in the ontology space. I am not wanting to bring politics into the PFC, and I am not going to publically respond to any replies to this post; but as we get into the philosophical roots of science (and especially ontology categories) we need to be aware of our assumptions.

*For a great (fiction) read on this see: “How Google Beat Amazon and EBay to the Semantic Web”. The subjective flaw is also why this never came about.