lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Alan Woodward <alan.woodw...@romseysoftware.co.uk>
Subject Re: machine learned ranking of documents
Date Tue, 09 Oct 2012 19:29:10 GMT
Hi parnab,

You want to look at the similarities package:
http://lucene.apache.org/core/4_0_0-ALPHA/core/org/apache/lucene/search/similarities/package-summary.html

Alan Woodward

On 9 Oct 2012, at 20:04, parnab kumar wrote:

> Hi All,
>        How do i incorporate machine learned ranking components  into
> lucene. I donot want to do re-ranking of documents(i.e I donot want to
> pre-fetch documents and then look into their features ). i want to do it on
> the fly as lucene does it . The various parameters (tf ,idf, norm ,boost
> in different fields ) that lucene uses for scoring will be used as one
> among the several features i have . I want to supply these features to a
> model which will compute a score and return me back a score which i want to
> incorporate into the lucene scoring components. Where should i look into ?
> which modules ,class i should modify,implement etc.
>                              Please advice/recommend or give any pointers
> !!
> 
> Thanks,
> Parnab


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