Re: OpenRefine

--------------------------------------------------
Fadi Maali
PhD student @ Insight Galway (formerly DERI)
Irish Research Council Embark Scholarship holder
http://www.deri.ie/users/fadi-maali

On 28 Oct 2013, at 20:37, Jeff Mixter <jeffmixter@gmail.com> wrote:

> A Response and a follow-up question,
> 
> As Kingsley mentioned suggested, I have been able to get the reconciliation service to work using LOD Refine with DBpedia and Freebase.  If you want, I would be more than happy to set up a Google Hangouts and walk you through the process.
> 
> I have experienced some problems, but it was when I tried to reconcile against my own 4Store instance.  Which leads me to my own question.  LOD Refine's standard SPARQL query is extremely poorly constructed and when querying large datasets (using 4Store) my requests always time out.  The dataset is a VIAF.nt data dump and consists of ~476 Million triples.  I have tried increasing the "soft limit" of the 4store server, but all this seems to do is cause them to take longer to fail.  I do not want to spend the money on a Virtuoso instance, since I am just using this dataset for experimentation.  Using Fiddler I was able to track the actual http request made by the Reconciliation and noticed an oddity in the standard query.  The issue with the standard LOD query is that they declare that the parameter being sent in the query is a string (str) even though that should be obvious (see below):
> 
> 
> Exact Match Search
> PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
> PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
> 
> SELECT ?entity 
> WHERE {?entity <http://xmlns.com/foaf/0.1/name> ?label 
>       FILTER ( str(?label) = "Hunter, Virginia, 1921-2011")
> } 
> 

This is needed to match language-tagged and type-tagged literals.

> Character by Character string match
> PREFIX rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#>
>     PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#>
>     
>     SELECT ?entity ?label1
>     WHERE {?entity <http://xmlns.com/foaf/0.1/name> ?label1 
>           FILTER regex(str(?label1), "Hunter, Virginia, 1921-2011", "i")
>     }   
> 
> When using the http SPARQL endpoint, I can remove the str parameter and retrieve back results.  This is true both for direct queries (exact matches) as well as string matches (close match).  Does anyone have any way to easily change the syntax of the SPARQL query that is sent out?

This seems as a problem with 4store to me. 
To change the SPARQL you can edit this:
https://github.com/fadmaa/grefine-rdf-extension/blob/master/src/org/deri/grefine/reconcile/rdf/factories/PlainSparqlQueryFactory.java

change the static strings at the bottom of the file.

Regards,
Fadi
> 
> Thanks,
> 
> Jeff Mixter
> jeffmixter@gmail.com
> mixterj@oclc.org
> 440-773-9079
> 
> 
> On Mon, Oct 28, 2013 at 3:44 PM, Kingsley Idehen <kidehen@openlinksw.com> wrote:
> On 10/28/13 2:59 PM, Hugh Glaser wrote:
> Unfortunately I’ve not been a regular user, so it is probably my stupidity.
> Basically, I go through the Reconcile process using the Freebase Reconcile service, but it doesn’t find anything to reconcile, even though I have fixed it so that there is an entry that has exactly the same text as the Freebase entry title.
> It just shows as if there are no positive results.
> I try clicking on the search for match after that, but it never comes back, which makes me wonder.
> 
> I've played with LOD Refine and successfully used reconciliation via Sindice, DBpedia and a variety of Entity Extraction Services (e.g., DBpedia Spotlight, Spaziodati, and Zemanta). By reconciliation I mean:
> 
> 1. selecting a column
> 2. choosing the reconciliation or entity extraction menu options
> 3. getting back Linked Data URIs .
> 
> With the reconciliation in place, I can then:
> 1. leverage the Linked Data URIs from the reconciliation process when constructing RDF skeletons (mappings from tabular to property graph based data representation)
> 5. use the RDF skeleton to generate a Turtle document.
> 
> In my recent experience, there are times when the system has just stopped working without any specific indication as to why e.g., imported (or bound) prefixes aren't recognized when constructing RDF mapping skeletons etc..
> 
> Maybe you could share your project file (via their project export option) and others could shed light on what might be going wrong etc..
> 
> 
> Kingsley
> 
> 
> On 28 Oct 2013, at 18:53, John Erickson <olyerickson@gmail.com> wrote:
> 
> Hugh, I wonder if you could be more specific regarding the troubles
> you had with OpenRefine?
> 
> One of our students also had trouble, and I'm wondering if it might be
> the same problem.
> 
> Like you, reconciliation with Refine has worked for me in the past but
> I haven't tried the same process using OpenRefine...
> 
> On Mon, Oct 28, 2013 at 2:41 PM, Hugh Glaser <hg@ecs.soton.ac.uk> wrote:
> Hi.
> I’m not sure where to ask, so I’ll try my friends here.
> I was having a go at OpenRefine yesterday, and I can’t get it to reconcile, try as I might - I have even watched the videos again.
> I’m doing what I remember, but it is a while ago.
> Are there others currently using it successfully?
> Or is it possibly a Mavericks (OSX) upgrade thing, which I did recently.
> Cheers
> --
> Hugh
> 
> 
> 
> -- 
> John S. Erickson, Ph.D.
> Director, Web Science Operations
> Tetherless World Constellation (RPI)
> <http://tw.rpi.edu> <olyerickson@gmail.com>
> Twitter & Skype: olyerickson
> --
> Hugh
> 023 8061 5652
> 
> 
> 
> 
> 
> 
> -- 
> 
> Regards,
> 
> Kingsley Idehen 
> Founder & CEO
> OpenLink Software
> Company Web: http://www.openlinksw.com
> Personal Weblog: http://www.openlinksw.com/blog/~kidehen
> Twitter/Identi.ca handle: @kidehen
> Google+ Profile: https://plus.google.com/112399767740508618350/about
> LinkedIn Profile: http://www.linkedin.com/in/kidehen
> 
> 
> 
> 
> 
> 
> 
> 
> -- 
> Jeff Mixter
> jeffmixter@gmail.com
> 440-773-9079

Received on Wednesday, 6 November 2013 10:50:17 UTC