[LLD] Fwd: Michel's semantic interoperability definitions

In discussions about finishing touches on the Emerging Practices IG
Note, I brought up wanting to integrate some definitions that Michel
had brought up. I have finally located the text that I keep referring
to. :) See below.

Cheers, Scott

---------- Forwarded message ----------
From: Michel Dumontier <michel.dumontier@gmail.com>
Date: Mon, Jul 9, 2012 at 7:32 PM
Subject: Re: GXA RDF
To: expressionrdf@googlegroups.com
Cc: jupp@ebi.ac.uk


Ok,
 so, i'm planning to add a mapping table - from a generalized concept
to the specific vocabularies in which that concept is defined. Now,
the list of terms immediately supports the goal of semantic
annotation, but it falls short in terms of semantic interoperability.
One of the things that we can do is mimic what we did for the TMO work
- we added terms to the TMO with formalized mappings (rdfs:subClassOf,
owl:equivalentClass) to those target terminologies, and could then use
the TMO classes to query data that was annotated with *any* of the
terminologies.  This is fine for simple concept-based data retrieval,
but is very poor for structured queries, where each of the source data
has a different topology.  Approaches like that of BioPAX helps
marshal structurally and semantically varying data into one coherent
form (FWIW). We can, however, facilitate this "global schema mapping"
through SPARQL construct queries that lifts the data from one source
to the target.  So here's the list of options with increasing work:

1. semantic annotation - list terms in use and/or corresponding terms
using BioPortal
 * benefit: users know where to look for terms of interest to them and
will provide some kind of semantically annotated data
2. terminological mapping - formalized the equivalance/subclass
relations among terms
 * benefit: users know that any of the terms can be used to annotate
and query their data (at the type level)
3. semantic interoperability - establish SPARQL-based transformations
of each source to a global schema
* benefit: users can query any data using a common formalization of those data

m.

Received on Monday, 10 September 2012 14:58:04 UTC