search returns 'categories' instead of url

[remi tassing <ta...@gmail.com>]

Hi,

Let's say Solr is setup and can return relevant urls. What if I wanted to
get the most cited terms from a predefined list, instead? It could be from
a list of products, names, cities...

Any ideas?

Remi

Re: search returns 'categories' instead of url

[Ahmet Arslan <io...@yahoo.com>]

> Ok let's assume we have the basic solr engine that's able to
> search and
> give result of urls...now from those pages, I would like to
> know which
> terms are the most mentioned, e.g. iPad, Samsung,
> Candy...the list can be
> long but we could decide to only output the top#20 or so.

I use mlt.interestingTerms to generate meta name="keywords" for new-story pages. 

"...will show what "interesting" terms are used for the MoreLikeThis query. These are the top tf/idf terms."

http://wiki.apache.org/solr/MoreLikeThisHandler 

Re: search returns 'categories' instead of url

[remi tassing <ta...@gmail.com>]

Sincere apologies for the unclarity! I'm probably misusing technical terms
such 'category' ...

Ok let's assume we have the basic solr engine that's able to search and
give result of urls...now from those pages, I would like to know which
terms are the most mentioned, e.g. iPad, Samsung, Candy...the list can be
long but we could decide to only output the top#20 or so.

I'm not sure if this a more 'facet' or 'category' or 'cluster' job in Solr
terminology.

Remi

On Thursday, February 2, 2012, Chris Hostetter <ho...@fucit.org>
wrote:
>
> : > Another alternative solution would be to add a category field to the
> : > already crawled content.
>
> : > >> Let's say Solr is setup and can return relevant urls. What if I
wanted
> : > to get the most cited terms from a predefined list, instead? It could
be
> : > from a list of products, names, cities...
>
> You relaly need to explain your problem more -- i'm having a hard time
> understanding what type of usecase/situation you might be describing.
>
> based on your initial description it seems like you are just asking about
> something like using facet.query to get counts for specific terms; but
> them in your followup the idea of adding categorization to your existing
> index almost smells like a machine learning type problem.
>
> the question is just really too vague to make any guesses at.
>
> please give a specific examples of the type of data you are working with,
> the types of requests you want to send, the types of results you want to
> give back from those requests, and the types of results you do *NOT*
> wantto get back. so we can understand the boundaries of your probem.
>
>
> -Hoss
>

Re: search returns 'categories' instead of url

[Chris Hostetter <ho...@fucit.org>]

: > Another alternative solution would be to add a category field to the
: > already crawled content.

: > >> Let's say Solr is setup and can return relevant urls. What if I wanted
: > to get the most cited terms from a predefined list, instead? It could be
: > from a list of products, names, cities...

You relaly need to explain your problem more -- i'm having a hard time 
understanding what type of usecase/situation you might be describing.

based on your initial description it seems like you are just asking about 
something like using facet.query to get counts for specific terms; but 
them in your followup the idea of adding categorization to your existing 
index almost smells like a machine learning type problem.

the question is just really too vague to make any guesses at.

please give a specific examples of the type of data you are working with, 
the types of requests you want to send, the types of results you want to 
give back from those requests, and the types of results you do *NOT* 
wantto get back. so we can understand the boundaries of your probem.


-Hoss

Re: search returns 'categories' instead of url

[Dmitry Kan <dm...@gmail.com>]

An offtopic: as some of my questions went unnoticed too, I could recommend
asking them somewhere else in parallel, for example: stackoverflow.com.
But as SOLR and its "ecosystem" sometimes pose tough questions and
problems, stackoverflow can ignore them as well. Anyhow, just another
opportunity..

Dmitry

On Wed, Feb 1, 2012 at 8:26 PM, remi tassing <ta...@gmail.com> wrote:

> This topic is either boring or not clear enough...
>
> Another alternative solution would be to add a category field to the
> already crawled content.
>
> Any idea how to do it?
>
> Remi
>
> On Tuesday, January 31, 2012, remi tassing <ta...@gmail.com> wrote:
> > After looking at the Carrot2 introduction, it seems this can be solved
> with clustering but with pre-defined categories.
> > Does that make sense?
> > Remi
> >
> > On Sun, Jan 29, 2012 at 8:42 PM, remi tassing <ta...@gmail.com>
> wrote:
> >>
> >> Hi,
> >>
> >> Let's say Solr is setup and can return relevant urls. What if I wanted
> to get the most cited terms from a predefined list, instead? It could be
> from a list of products, names, cities...
> >>
> >> Any ideas?
> >>
> >> Remi
> >
>

Re: search returns 'categories' instead of url

[remi tassing <ta...@gmail.com>]

This topic is either boring or not clear enough...

Another alternative solution would be to add a category field to the
already crawled content.

Any idea how to do it?

Remi

On Tuesday, January 31, 2012, remi tassing <ta...@gmail.com> wrote:
> After looking at the Carrot2 introduction, it seems this can be solved
with clustering but with pre-defined categories.
> Does that make sense?
> Remi
>
> On Sun, Jan 29, 2012 at 8:42 PM, remi tassing <ta...@gmail.com>
wrote:
>>
>> Hi,
>>
>> Let's say Solr is setup and can return relevant urls. What if I wanted
to get the most cited terms from a predefined list, instead? It could be
from a list of products, names, cities...
>>
>> Any ideas?
>>
>> Remi
>

Re: search returns 'categories' instead of url

[remi tassing <ta...@gmail.com>]

After looking at the Carrot2 introduction, it seems this can be solved with
clustering but with pre-defined categories.

Does that make sense?

Remi

On Sun, Jan 29, 2012 at 8:42 PM, remi tassing <ta...@gmail.com> wrote:

> Hi,
>
> Let's say Solr is setup and can return relevant urls. What if I wanted to
> get the most cited terms from a predefined list, instead? It could be from
> a list of products, names, cities...
>
> Any ideas?
>
> Remi