lucene-general mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Ted Dunning <ted.dunn...@gmail.com>
Subject Re: Cluster Retrieval in Lucene
Date Thu, 25 Nov 2010 22:23:32 GMT
Can you provide a citation?  Citeseer is down at the moment.

On Thu, Nov 25, 2010 at 2:09 PM, vermansi <vermansi@gmail.com> wrote:

>
> Im sorry if my post misled you in any way but by cluster retrieval i mean
> something like this :
>
> http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.83.4177&rep=rep1&type=pdf
>
>
> On Fri, Nov 26, 2010 at 12:55 AM, Ted Dunning [via Lucene] <
> ml-node+1968974-40095033-277119@n3.nabble.com<ml-node%2B1968974-40095033-277119@n3.nabble.com>
> <ml-node%2B1968974-40095033-277119@n3.nabble.com<ml-node%252B1968974-40095033-277119@n3.nabble.com>
> >
> > wrote:
>
> > This is generally referred to as sharding.
> >
> > Solr can do this.
> >
> > Katta does this as well in a bit more flexible approach.  Solr Cloud is
> > retrofitting a similar approach into Solr.
> >
> > On Thu, Nov 25, 2010 at 9:49 AM, vermansi <[hidden email]<
> http://user/SendEmail.jtp?type=node&node=1968974&i=0>>
> > wrote:
> >
> > >
> > > Hello
> > > I wish to implement cluster based retrieval model in lucene. I havent
> > gone
> > > through the code fully and am unaware of any existing implementations
> for
> >
> > > it
> > > based on lucene.
> > > Could someone give me a heads up on where to begin .. as there is too
> > much
> > > of code to go through and I have very little time.
> > >
> > > Now my idea is ..
> > >
> > > Lucene index should be created in form of clusters . Ie At indexing
> time
> > > each Document (D) could belong to a cluster.
> > > On Query (Q) submission the each cluster is searched for relevant
> > > documents.
> > > And the documents from that cluster as well as other clusters are
> ranked.
> >
> > >
> > > A brute force way of implementing it could be
> > > 1. Clusters are denoted by a field Name -- cluster (C).
> > > 2. the words are search in cluster field.
> > > 3. The scoring functions are changed to incorporate the math used in
> > > cluster
> > > retrieval
> > > 4. Documents in each cluster ranked seperately. And then merged
> > >
> > > Now the problem with this approach is many queries will have to be
> > created
> > > and the result processing will increase considerably...
> > > If there are more ways to doing it please lemme know .
> > >
> > > Regards
> > > Mansi
> > >
> > > --
> > > View this message in context:
> > >
> >
> http://lucene.472066.n3.nabble.com/Cluster-Retrieval-in-Lucene-tp1968500p1968500.html
> <
> http://lucene.472066.n3.nabble.com/Cluster-Retrieval-in-Lucene-tp1968500p1968500.html?by-user=t
> >
> > > Sent from the Lucene - General mailing list archive at Nabble.com.
> > >
> >
> >
> > ------------------------------
> >  View message @
> >
> http://lucene.472066.n3.nabble.com/Cluster-Retrieval-in-Lucene-tp1968500p1968974.html
> > To unsubscribe from Cluster Retrieval in Lucene, click here<
> http://lucene.472066.n3.nabble.com/template/NamlServlet.jtp?macro=unsubscribe_by_code&node=1968500&code=dmVybWFuc2lAZ21haWwuY29tfDE5Njg1MDB8NDAzMzYxNzU4
> >.
> >
> >
>
> --
> View this message in context:
> http://lucene.472066.n3.nabble.com/Cluster-Retrieval-in-Lucene-tp1968500p1969844.html
> Sent from the Lucene - General mailing list archive at Nabble.com.
>

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message