I'm not clear exactly what you are asking but I think you will have to
build your TermQuery instances one at a time and that sounds fine, if
it does what you want and is sufficiently fast.
--
Ian.
On Tue, Jan 31, 2012 at 1:34 PM, Pedro Lacerda <pslace...@gmail.com> wrote:
> For the first strategy i'm using MoreLikeThis to generate one query (from
> Doc terms) for each analyzed field (from type1 and type2), applying boosts
> and searching with TermsFilter to select only documents of type2.
>
> For the second I construct an map <termString, boost> where boost is the
> tf-idf of Doc (using searcher and similarity). I failed in using this map
> to construct an query because I'm finding something like TermQuery("*",
> termStr), or building one TermQuery by field by termStr is ok?
>
> Sorry if i'm not sufficiently explicit about what I mean, I'm on basic
> level English course.
>
> Pedro Lacerda
>
>
>
> 2012/1/26 Pedro Lacerda <pslace...@gmail.com>
>
>> Hi list,
>>
>> We have two different document types with different fields each. My
>> problem is given one document (Doc) from type1, find similar ones of type2.
>> Initially I thought two strategies to do it:
>>
>> - index all documents together; build my query with terms from Doc and
>> fields of type2; and filter out documents of type1.
>> - index type1 and type2 documents separately; compute scores (like
>> tf-idf) for each term of Doc on type1 index; build my query with terms
>> from
>> Doc and apply the scores as boosts; search on type2 index.
>>
>> I hope you have nice suggestions for me, because I started to learn Lucene
>> but she is giving me a lot of headache!
>>
>> Pedro Lacerda
>>
>>
---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscr...@lucene.apache.org
For additional commands, e-mail: java-user-h...@lucene.apache.org
