W3C home > Mailing lists > Public > public-semweb-lifesci@w3.org > August 2007

Re: Paper: URI Identity Management for Semantic Web Data Integration and Linkage

From: Mark Montgomery <markm@kyield.com>
Date: Wed, 1 Aug 2007 13:25:37 -0700
Message-ID: <001301c7d47a$1f517840$a100a8c0@Inspiron>
To: "Andreas Harth" <andreas.harth@deri.org>, "Eric Neumann" <eneumann@teranode.com>
Cc: <public-semweb-lifesci@w3.org>

Unfortunately, none of these or any other I recall seeing to date favors 
innovation, but rather reinforces group think. The least empowering to group 
think and automatic replication of either outdated or false information is 
random ordering, perhaps, but the combination of all of these won't get us 
to a semantic web. I think it's essential to always consider the weaknesses 
of a medium in working towards improving its strengths, and error in that 
direction if we must.

.02 - MM

----- Original Message ----- 
From: "Andreas Harth" <andreas.harth@deri.org>
To: "Eric Neumann" <eneumann@teranode.com>
Cc: <public-semweb-lifesci@w3.org>
Sent: Wednesday, August 01, 2007 11:28 AM
Subject: Re: Paper: URI Identity Management for Semantic Web Data 
Integration and Linkage


>
> Hi Eric,
>
> you might find [1] interesting in relation to the one thing/multiple URI
> problem you're mentioning.
>
> We distinguish four approaches to chosing a URI over multiple
> alternatives in a Web data aggregation scenario:
>
> - random or lexical ordering
> - most agreed upon across data sources
> - count of occurrence
> - links analysis
>
> Regards,
> Andreas.
>
> [1] http://sw.deri.org/2007/02/objcon/paper.pdf
>
> Eric Neumann wrote:
>>
>> A recent paper has been published that addresses some of the same URI
>> issues we've been discussing:
>>
>> "URI Identity Management for Semantic Web Data Integration and Linkage"
>> http://eprints.ecs.soton.ac.uk/14361/
>>
>> They focus on Coreference aka Record Linkages, which deals with the
>> problem of having more than one URI for the "same thing". There's a
>> probabilistic theory behind it, but it offers an approach different
>> from using "owl:sameAs' between possible similar things.
>>
>> My take-- coming from a life sciences perspective-- is that we still
>> need to standardize URI's to specific datarecords (e.g., uniprot,
>> entrez, ensemble, etc) as well as concepts, but when we need to
>> "cross-bundle" different records that are supposedly referring to the
>> same bio-entity (uniprot/p12345 ~ entrez/g6422 ~ ensmeble/s47721),
>> then this approach may be worth considering.
>>
>> I leave it to the group to discuss the possible value of this paper to
>> our ongoing URI activity...
>>
>> Eric
>>
> 
Received on Wednesday, 1 August 2007 20:26:17 GMT

This archive was generated by hypermail 2.3.1 : Tuesday, 26 March 2013 18:00:49 GMT