Hello, guys.
How are you? MoreLikeThis doesn´t support Distributed searchs, so I think
that I could emulate a similar behaviour with eDismax features. What I´ll
like to do is:
*1. *I have two fields that contains important information from the
document. This two fields will be used to relate content between other Solr
docs.
*2.* I defined an eDismax RequestHandler as follows:
<!-- Request handler that find similar documents by it's i_related and
i_text fields -->
<requestHandler name="/similar" class="solr.SearchHandler">
<lst name="defaults">
<str name="echoParams">explicit</str>
<!-- Query settings -->
<str name="defType">edismax</str>
<str name="qf">title^25.0 description^5.0</str>
<str name="mm">50%</str>
<str name="fl">*,score</str>
<str name="wt">json</str>
</lst>
</requestHandler>
*3. *Supose that we have *doc1 *with *description field*, which is
multivalued, containing*: { *big, yellow... and other stuff *}. *
*
*
*4. The objective: * find those similar documents in base of the values
contained in *doc1 description field*.
EDismax is nice, because you can even find documents whose description
field has those terms but also find those which have those terms inside
title field. Here boosting fields makes this solution more flexible.
However, the behavior is not de desired. The reasons are:
- Edismax uses the natural and default relevance scoring, so is expected
that some response documents appears at the top of the response even if
they only share one or two description terms. MoreLikeThis boost and order
those response documents by the number of terms that they share, so I would
like to do something like that (more or less). Does anyone know if sorting
with a function query could help me? Any tips?
- Edismax is as quick as desired. I finally I bet for that solution, I
would like to improve performance. Again I´m opened to hear any suggestion,
recommendation... to improve it.
Thank you very much
--
- Luis Cappa
