Silk LinlType wrong

[Marco Scalone]

Hello!
I'm doing some testing with LDIF but have some problems in the linking step. I'm trying to generate links from entities (my books) to authors (foaf:Agent) from DBLP. I'm using a property dc:contributor that hast the author name (literal) and trying to compare to the foaf:name of the authors in DBLP. 

The problem is that I need to generate a link of type dc:contributor o dc:creator but silk always use owl:sameAs no matter what you put in LinkType config element. This is a case where owl:sameAs is not suitable at all. Ithought it was a silk problem but in the intermediate result of Silk the target links appears right! dc:contributor. When the integration process ends the link type is owl:sameAs witch is wrong because a book is not sameAs an foaf:Agent (from DBLP)

Any idea what is happening?
thanks.

[Marco Scalone]

Ok let me rectify my question.
nothing is wrong with the sameAs link. In the LDIF doc says that SILK is used to identity resolution and uses only sameAs, so thats the way its implemented/used.

The question is.. If Silk allows using  different link types why reduce that functionality to only sameAs in LDIF?

thanks

[andrea]

Hi,

good catch, thanks!

It should be solved now, in trunk [1].

Cheers,

Andrea

[1] https://github.com/wbsg/ldif/commit/ab0246c

--
You received this message because you are subscribed to the Google Groups "LDIF" group.
To unsubscribe from this group and stop receiving emails from it, send an email to ldif+uns...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.
 
 

[Marco Scalone]

thanks! I'll try it

[Marco Scalone]

It worked! thanks

[Marco Scalone]

That's me again :)
I thought it was working fine but I continued testing the trunk version and some issues appeared with the linking process.

The link seems to be done correctly but not writing down in the integration result. I also configured the integrationJob.xml to write down the intermediate results of the linkng process but nothing was written.

The strange thing is that as part of the output at the end of the process the linking quad is printed out (in the terminal) but the number of links genearated by silk is 0 (as you can see bellow in bold).

If I run the same process (the same config) with the "official" build ldif-0.5.1 (previous the modification) the integration result is written down (but with the fixed sameAs linktype bug) 

¿Any Idea?

OUTPUT WITH TRUNK VERSION ---------------------------------
...

<http://localhost:2020/resource/item/oai:cnx.org:m10672> <http://purl.org/dc/elements/1.1/contributor> <http://dblp.l3s.de/d2r/resource/authors/A._Algaba> <http://www4.wiwiss.fu-berlin.de/ldif/graph#generatedBySilk> .

<http://localhost:2020/resource/item/oai:cnx.org:m10668> <http://purl.org/dc/elements/1.1/contributor> <http://dblp.l3s.de/d2r/resource/authors/A._Algaba> <http://www4.wiwiss.fu-berlin.de/ldif/graph#generatedBySilk> .

ago 26, 2013 6:03:11 PM ldif.local.IntegrationJob executeLinkingPhase
INFO: Time needed to link data: 8.371s
ago 26, 2013 6:03:11 PM ldif.local.IntegrationJob executeLinkingPhase
INFO: Number of links generated by silk: 0
ago 26, 2013 6:03:11 PM ldif.local.runtime.URITranslator$ generateMintedUriMap
INFO: Minting URIs...
ago 26, 2013 6:03:11 PM ldif.local.runtime.URITranslator$ rewriteURIs
INFO: Start URI translation...
ago 26, 2013 6:03:12 PM ldif.local.runtime.URITranslator$ rewriteURIs
INFO: End URI translation: Processed 15343 quads
ago 26, 2013 6:03:12 PM ldif.local.IntegrationJob executeURITranslation
INFO: Time needed to translate URIs: 0.435s
ago 26, 2013 6:03:12 PM ldif.local.IntegrationJob writeOutput
INFO: Writing integration output...
ago 26, 2013 6:03:12 PM ldif.local.IntegrationJob writeOutput
INFO: 15343 Quads written
ago 26, 2013 6:03:12 PM ldif.local.Scheduler runIntegration
INFO: Integration Job completed
Press any key to exit

--------------------------------------------------------------------------------

OUTPUT WITH OFFICIAL ldif-0.5.1 VERSION ---------------------------------------

...
ago 26, 2013 6:12:07 PM de.fuberlin.wiwiss.silk.util.task.HasStatus$class updateStatus
INFO: Filtering finished in 4ms (100,0%)
ago 26, 2013 6:12:07 PM de.fuberlin.wiwiss.silk.util.task.HasStatus$class updateStatus
INFO: Writing output started (0,0%)
ago 26, 2013 6:12:07 PM de.fuberlin.wiwiss.silk.output.Output close
INFO: Wrote 131 links
ago 26, 2013 6:12:07 PM de.fuberlin.wiwiss.silk.util.task.HasStatus$class updateStatus
INFO: Writing output finished in 3ms (100,0%)
INFO  Time needed to link data: 8.086s
INFO  Number of links generated by silk: 263
INFO  Minting URIs...
INFO  Start URI translation...
INFO  End URI translation: Processed 15347 quads
INFO  Time needed to translate URIs: 0.559s
INFO  Writing integration output...
INFO  15480 Quads written
INFO  Integration Job completed
Press any key to exit

--------------------------------------------------------------------------------

[Marco Scalone]

I forgot it,

Another thing I notice (in the official build) is that the quads added as result of the link process (always sameAs) had the graph <http://www4.wiwiss.fu-berlin.de/ldif/graph#{sourceID}> instead of the  <http://www4.wiwiss.fu-berlin.de/ldif/graph#generatedBySilk> as i would expect.

the following  is an example of an link added quad using ldif-0.5.1 build

in the  intermediate silk result file:
<http://localhost:2020/resource/item/oai:cnx.org:m15557> <http://purl.org/dc/elements/1.1/contributor> <http://dblp.l3s.de/d2r/resource/authors/A._Algaba> <http://www4.wiwiss.fu-berlin.de/ldif/graph#generatedBySilk>

in the final integration result file:
<http://localhost:2020/resource/item/oai:cnx.org:m15557> <http://www.w3.org/2002/07/owl#sameAs> <http://dblp.l3s.de/d2r/resource/authors/A._Algaba> <http://www4.wiwiss.fu-berlin.de/ldif/graph#cnx2013> .

regards
Marco

[Marco Scalone]

Hello,
   Could you test some of these issues?

thanks,
Marco

[andrea]

Hey Marco, 

I just fixed the issue in trunk: it should now output the correct number of links generated by Silk.

I'll have a look at the other issues when I'm back from vacation, in about a week ;)

Thanks again for your feedback,

Andrea

[Marco Scalone]

Andrea,
       I didn't want to make pressure, just asking :)

I'll test it later,

thanks alot

2013/9/4 andrea <andrea....@gmail.com>

[andrea]

Hi Marco,

the issue with the provenance graphs is related to the 'rewriteURIs' option. When rewriting URIs, the provenance graph for sameAs statements is currently overwritten.

This is actually part of a wider issue I tracked here https://github.com/wbsg/ldif/issues/4

Thanks,

Andrea