W3C home > Mailing lists > Public > public-lod@w3.org > February 2010

Re: DBpedia-based entity recognition service / tool?

From: Matthias Samwald <samwald@gmx.at>
Date: Tue, 2 Feb 2010 17:37:39 +0100
Message-ID: <92308DA009354A57A139E784AE054FA0@ms>
To: "Dan Brickley" <danbri@danbri.org>, "Georgi Kobilarov" <georgi.kobilarov@gmx.de>
Cc: <public-lod@w3.org>
Danbri wrote:
> One idea I had on NoTube that is implemented experimentally in
> http://lupedia.ontotext.com/ is to use RDFa as an interop point. So
> one of the interfaces from the Ontotext demo there is to return RDFa
> markup - http://lupedia.ontotext.com/test-page4rdfa.html 

I had a similar idea when I created a text annotation service for the biomedical domain. For example:

http://whatizit.neurocommons.org/index.py/pmid?pipeline=whatizitEBIMedDiseaseChemicals&query=17477962&format=atag
(can be quite slow to respond)

This contains recognized entities embedded via RDFa. In this case, each sentence is represented as a sioc:Item and the entities within the sentence are represented as sioc:topics of the sentence.

The RDF of the example above looks like this:
http://tinyurl.com/ydayqr3

However, the entities are taken from ontologies in the Open Biomedical Ontologies (OBO) collection, and not DBpedia. Entity recognition is based on another bioinformatics server, and I cannot just add DBpedia as a vocabulary.

> however
> this doesn't leave much scope for including confidence measures etc in
> the output.

Why? You could include it as RDFa markup around pieces of the text that are not visible in the browser?

Cheers,
Matthias Samwald
Received on Tuesday, 2 February 2010 16:38:12 UTC

This archive was generated by hypermail 2.3.1 : Sunday, 31 March 2013 14:24:25 UTC