Tagging using SOLR

Dear all,

We are running an appalication built using SOLR, now we are
trying to build
a tagging system using the existing SOLR indexed field
called
"tag_keywords", this field has different keywords
seperated by comma, please
give suggestions on how can we build tagging system using
this field?

Thanks,
Mohandoss.

On Thu, 2007-09-06 at 12:59 +0530, Doss wrote:
> Dear all,
> 
> We are running an appalication built using SOLR, now we
are trying to build
> a tagging system using the existing SOLR indexed field
called
> "tag_keywords", this field has different
keywords seperated by comma, please
> give suggestions on how can we build tagging system
using this field?

http://wi
ki.apache.org/solr/ConfiguringSolr

http://wiki.apa
che.org/solr/SchemaXml
Define a new field named keyword and use the
"text_ws" as type. Instead
of comma use whitespaces instead.
...
<!-- A text field that only splits on whitespace for
exact matching of
words -->
    <fieldtype name="text_ws"
class="solr.TextField"
positionIncrementGap="100">
      <analyzer>
        <tokenizer
class="solr.WhitespaceTokenizerFactory"/>
      </analyzer>
    </fieldtype>
...
<field name="id" type="text_ws"
indexed="true" stored="true"/>

HTH

salu2
-- 
Thorsten Scherler                                
thorsten.at.apache.org
Open Source Java                      consulting, training
and solutions

On Sep 6, 2007, at 3:29 AM, Doss wrote:
> We are running an appalication built using SOLR, now we
are trying  
> to build
> a tagging system using the existing SOLR indexed field
called
> "tag_keywords", this field has different
keywords seperated by  
> comma, please
> give suggestions on how can we build tagging system
using this field?

There is also a wiki page on some brainstorming on how to
implement  
tagging within Solr: <http://
wiki.apache.org/solr/UserTagDesign>

It's easy enough to have a tag_keywords field, but updating
a single  
tag_keywords field is not so straightforward without sending
the  
entire document to Solr every time it is tagged.  See
SOLR-139's  
extensive comments and patches to see what you're getting
into.

	Erik

Dear Thorsten, Erik,

Thanks for the guidelines, but basically our idea is to
build a system like
http://del.icio.us/tag/,
is it possible to take counts of similar words from
a solr indexed field?

Thanks,
Mohandoss


On 9/6/07, Erik Hatcher <erikehatchersolutions.com>
wrote:
>
>
> On Sep 6, 2007, at 3:29 AM, Doss wrote:
> > We are running an appalication built using SOLR,
now we are trying
> > to build
> > a tagging system using the existing SOLR indexed
field called
> > "tag_keywords", this field has different
keywords seperated by
> > comma, please
> > give suggestions on how can we build tagging
system using this field?
>
> There is also a wiki page on some brainstorming on how
to implement
> tagging within Solr: <http://
wiki.apache.org/solr/UserTagDesign>
>
> It's easy enough to have a tag_keywords field, but
updating a single
> tag_keywords field is not so straightforward without
sending the
> entire document to Solr every time it is tagged.  See
SOLR-139's
> extensive comments and patches to see what you're
getting into.
>
>        Erik
>
>

On Sep 7, 2007, at 3:09 AM, Doss wrote:
> Thanks for the guidelines, but basically our idea is to
build a  
> system like
> http://del.icio.us/tag/,
is it possible to take counts of similar  
> words from
> a solr indexed field?

How do you define similar words?

Here's a tag cloud for a single user in Collex, a system I
built:

	<http://www.nines.org/permalink/cloud/tag/nowviskie>


Collex is using SOLR-139 now for tagging/annotating.

You will need to think through handling updates to your
documents,  
and whether you will have user-specific tags too.  Reading
up on the  
links I sent you and doing some experimenting is highly
recommended.   
It's a non-trivial  scenario with Solr at this time.

By the way, a del.icio.us competitor, Simpy, is built on
Lucene - so  
it is quite possible to build a heavy duty tagging system,
but the  
devil is in the details.

	Erik



>
> Thanks,
> Mohandoss
>
>
> On 9/6/07, Erik Hatcher <erikehatchersolutions.com>
wrote:
>>
>>
>> On Sep 6, 2007, at 3:29 AM, Doss wrote:
>>> We are running an appalication built using
SOLR, now we are trying
>>> to build
>>> a tagging system using the existing SOLR
indexed field called
>>> "tag_keywords", this field has
different keywords seperated by
>>> comma, please
>>> give suggestions on how can we build tagging
system using this  
>>> field?
>>
>> There is also a wiki page on some brainstorming on
how to implement
>> tagging within Solr: <http://
wiki.apache.org/solr/UserTagDesign>
>>
>> It's easy enough to have a tag_keywords field, but
updating a single
>> tag_keywords field is not so straightforward
without sending the
>> entire document to Solr every time it is tagged. 
See SOLR-139's
>> extensive comments and patches to see what you're
getting into.
>>
>>        Erik
>>
>>