Hi Hugh,
Thank you for your suggestion -- yes, I was able to replicate the problem on one
of your servers. Actually, I've been experimenting a bit more with my problem
and observed that if I used a triple using the "a" keyword (or rdf:type) with
a specific object resource in my query, the hit scores were returned all the
same.
Here is an example query I ran against your live endpoint
(lod.openlinksw.com/sparql):
select ?sc ?entity (bif:search_excerpt(bif:vector ('The Who'), ?lit)) as ?value
where {
?entity a <http://schema.org/MusicGroup> .
?entity ?prop ?lit .
?lit bif:contains "'The Who'" option (score ?sc)
}
order by desc(?sc)
limit 200
I ran the above query a few times and each time I would get the same values for
?sc
(they actually varied from run to run, but they were all the same for each run).
Occasionally, I would even get all zeros! I can email you a couple of
resulting csv
files that I saved, if you'd like. Here is the command I used to run the query:
curl -u demo:demo -H "Accept:text/csv" --data-urlencode qu...@q7.spql
http://lod.openlinksw.com/sparql > output.csv
BTW, I was getting similar results when using your endpoint interface from a
browser.
However, a similar query, but without a specific object resource for the "a"
predicate
triple, produced reasonable scores, every time. For example:
select ?sc ?cls ?entity (bif:search_excerpt(bif:vector ('The Who'), ?lit)) as
?value
where {
?entity a ?cls .
?entity ?prop ?lit .
?lit bif:contains "'The Who'" option (score ?sc)
}
order by desc(?sc)
limit 200
Furthermore, what I also observed was that when I executed such a query (with
the "a"
predicate and a specific object) immediately after I loaded my data set into an
empty db,
the scores were reported different (reasonable values). However, as soon as I
would
shutdown the server and bring it back up, the same query returned scores that
were all
equal.
Thank you,
Krys
On 3/11/15 12:41 AM, Hugh Williams wrote:
> HI Krys,
>
> Note that I am aware of. Are you able to provide a demonstration of the
> problem being accounted against one of our live
> server like http://lod.openlinksw.com/sparql or
> http://demo.openlinksw.com/sparql , which run the 7.2 code base and do
> provide variable scores//ranks from my testing …
>
> Best Regards
> Hugh Williams
> Professional Services
> OpenLink Software, Inc. // http://www.openlinksw.com/
> Weblog -- http://www.openlinksw.com/blogs/
> LinkedIn -- http://www.linkedin.com/company/openlink-software/
> Twitter -- http://twitter.com/OpenLink
> Google+ -- http://plus.google.com/100570109519069333827/
> Facebook -- http://www.facebook.com/OpenLinkSoftware
> Universal Data Access, Integration, and Management Technology Providers
>
>> On 9 Mar 2015, at 04:16, Krys Kochut <koc...@cs.uga.edu
>> <mailto:koc...@cs.uga.edu>> wrote:
>>
>> Hello,
>>
>> I just upgraded from open source Virtuoso 7.1 to 7.2.0_p1 (from sourceforge).
>> All is fine except that I have a problem with the full text search using
>> bif:contains and the score option. What happens is that it returns the same
>> score values to all hits. For example, a query including the following
>> triple:
>>
>> ?lit bif:contains "'melanoma'" option (score ?score)
>>
>> returns all correct matches, but the values of the ?score variable are all
>> the same for all matches.
>>
>> Has something changed in the algorithm computing the text matching scores
>> in 7.2 that I may not be aware of? In VOS 7.1, the same exact queries
>> involving
>> bif:contains and the option score returned different scores, but now in 7.2
>> they are all the same. I created the full text index in the same way as
>> before.
>>
>> Thank you.
>> Krys
>>
>>
>>
>> ------------------------------------------------------------------------------
>> Dive into the World of Parallel Programming The Go Parallel Website,
>> sponsored
>> by Intel and developed in partnership with Slashdot Media, is your hub for
>> all
>> things parallel software development, from weekly thought leadership blogs to
>> news, videos, case studies, tutorials and more. Take a look and join the
>> conversation now. http://goparallel.sourceforge.net/
>> _______________________________________________
>> Virtuoso-users mailing list
>> Virtuoso-users@lists.sourceforge.net
>> <mailto:Virtuoso-users@lists.sourceforge.net>
>> https://lists.sourceforge.net/lists/listinfo/virtuoso-users
>
------------------------------------------------------------------------------
Dive into the World of Parallel Programming The Go Parallel Website, sponsored
by Intel and developed in partnership with Slashdot Media, is your hub for all
things parallel software development, from weekly thought leadership blogs to
news, videos, case studies, tutorials and more. Take a look and join the
conversation now. http://goparallel.sourceforge.net/
_______________________________________________
Virtuoso-users mailing list
Virtuoso-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/virtuoso-users
