A Semantic Web representation of biomedical event statements (ISMB Bio-Ont SIG 2009)

Colin Batchelor

It is straightforward to put molecules in RDF, e.g. as InChI URIs or as OBO HTTP URIs based on ChEBI. But events involving molecules are more complicated. But events cannot exist without participants. Previous work includes Schulz and Jansen, and a mail thread on the OBO tracker/mailing list.

Example: “Jones did it, deliberately, in the bathroom, with a knife, and midnight. What he did was butter a piece of toast.” (from Davidson 1967). The event, with a duration, is buttering. Instead of representing it with a single n-ary predicate (Buttering(jones, piece of toast, knife, midnight, bathroom), Davidson introduces an event variable and separate predicates for most of the details (event(e) [buttering(Jones, piece of toast,e) & with(knife,e) & at(e, midnight) & in(e,bathroom)]). But it’s not easy to go from a n-ary relationship to a small one. And that second version is still a 3-place predicate, and you really want a 2-place predicate.

The Neo-davidsonian semantics is to reduce all predicates to binary or unary and introduce semantic roles: exists(e)[buttering(e) & has_agent(e,Jones), has_patient(e,piece of toast)]. These FOL statements are all about variable/instances, but DL are set up differently, about TBoxes and ABoxes. What do we do with our predicates? Jones didn’t do all that with all toast with all knives etc… Therefore we need to specify instance_of, e.g. instance_of(e,buttering) for the appropriate parts of the statement.

What is the scope? Covered: PPIs, chemistry ‘recipe’ instructions. What’s not covered is nominalizations in general, and genetic variations. A nominalization is not necessarily an event, even though it sounds like one: it need not have temporal parts, and may be wholly present to the extent they exist at all. An example of this is the physical piece of paper that is a bill/reciept.

FriendFeed discussion: http://ff.im/4wTA2

Please note that this post is merely my notes on the presentation. They are not guaranteed to be correct, and unless explicitly stated are not my opinions. They do not reflect the opinions of my employers. Any errors you can happily assume to be mine and no-one else’s. I’m happy to correct any errors you may spot – just let me know!

Advertisements

Leave a Reply

Please log in using one of these methods to post your comment:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s