Kingsley Idehen wrote:
Nathan wrote:
Kingsley Idehen wrote:
[email protected] wrote:
Not sure if that will help do what I want. I'd like something that
can be used with a search box that can be used for free-text queries
to find items of a specific class.
For example to look for movies it could set to find items of
rdf:type http://freebase.com/film/film, where the search terms could
be from something like rdfs:label from the movie and rdfs:label from
http://freebase.com/film/film/directed_by,
http://freebase.com/film/film/produced_by and
http://freebase.com/film/film/starring.
A query with the terms "Scorsese DiCaprio gangs" might then return,
http://freebase.com/guid/9202a8c04000641f800000000014e9f1, the movie
"Gangs of New York" which has the actor "Leonardo DiCaprio",
director "Martin Scorsese" and "gangs" in the title. The point here
though is that when the query is executed, it won't know that
Scorsese is the director, DiCaprio is an actor and gangs is from the
title, just that each of the terms should be in either the movie
label, directed_by, produced_by or starring.
Are you telling me that via Solr you can perform a disambiguated full
text search?
Did you try putting: Gangs of New York, into the "URI Lookup (By
Label)" tab?
If you have a URL for a Solr instance that performs disambiguated
search against a full text pattern, please send it to me.
possibly worth noting that I moved *from* SOLR to Virtuoso for
precisely this reason (initially) all though now I could give a
hundred more (but please don't ask me to list a full 100 reasons).
Best,
Nathan
Nathan,
List the top 5. Or the Top 1 that would trigger a Solr retort URL re
claim invalidation :-)
I ask because I am yet to grok how Solr offers any kind of solution to:
disambiguated precision search & find, in a Linked Data realm, that
comes close to what Virtuoso offers.
SOLR doesn't come close and can't do the disambiguation; moreover its
key-value not triple so simply can't hold the eav structures needed to
deal with linked data (nor can it "follow the links" which is just
slightly critical)
in short I was convinced we needed SOLR, had been using it for ages,
imported the data in virtuoso and replaced solr with infinite
functionality and ways of querying the data, and didn't loose the search
because we just bif:contains - it's also worth noting though that we
don't really need search now other than the odd occasion, because
everything is LINKed ;)
Best,
Nathan
