Can't output the generate links

57 views
Skip to first unread message

maatar...@gmail.com

unread,
Jan 17, 2016, 5:18:58 PM1/17/16
to Silk

Hi,

I try to read here and there, and honestly you guys really don't make it easy. Don't you have one simple guide that one should follow to retrieve his results (output). This is the main point of the software, getting the links at the end, and this functionality is simply not obvious at all.

I am using the latest release. I spend a full day on it, and i can't find a way to retrieve the output.

I am never able to specify an output file, other than the one from the source and target.

It appears that when i choose the target file/endpoint to be also the output, then i get a plugin not present bug.

This is deeply confusing.

Please a quick help here would be great and benefit anyone who comes to use the tool. 




Ghislain ATEMEZING

unread,
Jan 18, 2016, 4:23:38 AM1/18/16
to silk-di...@googlegroups.com
Hi,
You can find this tutorial useful in your usecase [1] as it uses SKOS datasets at Europeana.

HTH

Ghislain

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



--

"Love all, trust a few, do wrong to none" (W. Shakespeare)

Robert Isele

unread,
Jan 18, 2016, 8:26:02 AM1/18/16
to silk-discussion
Hi,

are you using Silk Single Machine on the command line? In that case, there is a working example included in the current 2.7.0 release candidate. In case your are using the Workbench, there is also a simple example included.

Cheers,
Robert

maatar...@gmail.com

unread,
Jan 18, 2016, 9:51:36 AM1/18/16
to Silk
Where can I download the 2.7.0 release candidate ? I don't see a Tag for it in github ? 

Robert Isele

unread,
Jan 18, 2016, 9:59:31 AM1/18/16
to silk-discussion
The release candidate is tagged as release-2.7.0-RC1 . Binaries are available in the release section

maatar...@gmail.com

unread,
Jan 18, 2016, 10:29:23 AM1/18/16
to Silk
Robert,

 i have downloaded the version 2.7.0 release candidate. The source. 

I am running with OSX. Java 8. I don't have java 7 on my machine. I used the command sbt "project workbench" run

I have loaded the example. However as explain in the other post, that was started here https://groups.google.com/forum/#!topic/silk-discussion/mcRrbcn8GVs

I can't select any output file. I create them as resource of the project. However when i try to select them when specifying the properties of the linking it just does not work at all. 

I have tried with the examples that comes with the project and the result is the same i can't select an output file. 


I have attached the screenshot for your convenience. 


Is Java 8 the issue? Have you tested on OSX yourself ? If there is a workaround i would appreciate. 

Thanks

M
Screen Shot 2016-01-18 at 10.05.01 AM.png
Screen Shot 2016-01-18 at 10.17.10 AM.png
Screen Shot 2016-01-18 at 10.04.47 AM.png

Robert Isele

unread,
Jan 18, 2016, 10:35:00 AM1/18/16
to silk-discussion
In the output row of the Linking Task dialog you can select any dataset that you added to the project. I suspect you did create a resource to be used as output, but no data set for it yet? In that case, you need to add a file based data set to the project, which writes to the output resource you have created.

I understand that the current steps required for writing to files are not very intuitive and need to be improved in the future.

maatar...@gmail.com

unread,
Jan 18, 2016, 12:48:02 PM1/18/16
to Silk
Robert it worked. 

thanks for the support. 

I actually did that myself but for sparql, because the endpoint in stardog is different. DB/Query and DB/Update. So i'm going to also try to figure out how to write back in the DB directly. 

Unfortunately at this point it seems like the query that is generated by silk is causing trouble with stardog. 

Unfortunately i have no way to see the exact form of that query. 

Robert Isele

unread,
Jan 20, 2016, 5:54:43 AM1/20/16
to silk-discussion
I understand you are having problems when writing data to the store (but not when loading)? The SPARQL/Update query that is executed that the following form (assuming that you specified a graph):

INSERT DATA { GRAPH <" + graph + "> {
 Triples
}
Could you please try if such as query is working if you fire it manually?

Maatary Okouya

unread,
Jan 20, 2016, 6:25:11 AM1/20/16
to silk-discussion
Hummm that is indeed strange. This kind of query works perfect. The problem I have seem to be that at writing time, there is some query that use the load command.

Can u tell me if the following would be a problem :

In stardog, for a database A there are two end points:

- query endpoint: A/query (no update operation possible)
-update endpoint:A/update (not query operation possible)


So if you need to query A you have to use something of the form
A/query?query=.....

And for update

A/update?query=.....


I always thought that this might be a bit different than the usual so maybe that could cause an issue ?

I have indeed specified a third sparql endpoint in my spec which is the same database but only the update endpoint form and a graph where to write. So if on an Output sparql endpoint, you only do write operation, then everything is fine. However if you do query operation it will break your code I think.

Best,

Daniel


You received this message because you are subscribed to a topic in the Google Groups "Silk" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/silk-discussion/ksnbRYseW80/unsubscribe.
To unsubscribe from this group and all its topics, send an email to silk-discussi...@googlegroups.com.

Robert Isele

unread,
Jan 20, 2016, 9:24:29 AM1/20/16
to silk-discussion
Yes, currently the SPARQL endpoints do not support separate URLs for reading and writing. But that can be worked around by just adding separate datasets for reading and writing from/to the store and setting the URLs accordingly.

I will add a possibility to log failed queries for the next RC, which will make it possible to trace the problem back to the exact query that failed [1].

[1] https://github.com/silk-framework/silk/issues/44

Robert Isele

unread,
Jan 20, 2016, 9:40:41 AM1/20/16
to silk-discussion
I have uploaded a SNAPSHOT of the Workbench that will log failed queries here [1].

This also includes a parameter in the Generate Links tab "includeReferenceLinks" that assures that no negative reference links are generated, while always generating positive reference links. The config paramters are currently not preserved beyond restarts, but that is a planned extension.

Maatary Okouya

unread,
Jan 20, 2016, 7:14:47 PM1/20/16
to silk-discussion
Thanks, will double confirm what happens tomorrow.
Reply all
Reply to author
Forward
0 new messages