lucene-lucene-net-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Trond Aksel Myklebust" <Trond.Aksel.Mykleb...@idi.ntnu.no>
Subject SV: Missing records from searchresult
Date Thu, 20 Apr 2006 18:38:25 GMT
I've tried:
DotLucene 1.4.3 build 004
DotLucene 1.9 RC1 build 002 alpha

Since Luke and PyLucene returns the correct answers using the index
dotLucene creates I assume that the problem lies in searching using
dotLucene.
The best I can do atm regarding a testcase is a copy of my test Lucene index
together with a small app that returns the wrong number of results compared
to Luke and PyLucene.
http://folk.ntnu.no/trondaks/uploads/LuceneTest.zip
If I reduce the number of indexed documents the difference between the
number of hits from dotLucene and Luke gets smaller until they are the same,
almost like the index for "a" is overpopulated or there is a score threshold
somewhere. According to Luke "a" is the among the most used terms, unsure if
there is a connection between that and my "text:a*" search. 

Not exactly what you ask for but hope it can help someway. The index data is
extracted from a database:
IndexWriter writer = new IndexWriter(@"LuceneIndex", new
WhitespaceAnalyzer(), true);
//Get data from database
Document doc = new Document();
doc.Add(Field.Text("text", document.toLower()));
writer.AddDocument(doc);
writer.Close()


Trond


-----Opprinnelig melding-----
Fra: Jeff Rodenburg [mailto:jeff.rodenburg@gmail.com] 
Sendt: 20. april 2006 19:25
Til: lucene-net-user@incubator.apache.org
Emne: Re: Missing records from searchresult

Trond -

We have several assumptions that have to be made about how your index was
created, i.e. analyzer in use, contents of fields, etc.  Normally, if you
provide a unit test that mimics the behavior you're seeing, it's easier for
all of us to provide feedback.

Can you provide some information about the version of dotlucene you're
using?

-- j

On 4/20/06, Trond Aksel Myklebust <Trond.Aksel.Myklebust@idi.ntnu.no> wrote:
>
> Hello,
> Been using PyLucene, but for a small project I want to use dotLucene, 
> though having some difficulties.
>
> I have indexed 732 files and want to do a "text:a*" search against 
> these indexes files, but dotLucene only returns 530 of them. If I do a
"text:s*"
> search 732 files are returned as expected. Running the "text:a*" query 
> using Luke returns 731 matches, running the query using PyLucene 
> returns the same.
> Why am I missing 201 matches when using dotLucene, any issues I am not 
> aware of? If I do a search "text:ar*" files that was missing from the 
> first search shows up.
>
> dotLucene code that doesnt work:
> IndexSearcher searcher = new IndexSearcher(@"C:\Lucene"); Query query 
> = QueryParser.Parse("a*", "text", new WhitespaceAnalyzer()); Hits hits 
> = searcher.Search(query);
> searcher.Close()
>
>
> PyLucene code that works:
> import sys
> from PyLucene import *
> searcher = IndexSearcher("C:\\Lucene") query = QueryParser.parse("a*", 
> "text", WhitespaceAnalyzer()) hits = searcher.search(query)
> searcher.close()
>
> Hope someone got an answer/fix.
>
> Regards
> Trond A Myklebust
>
>
>



Mime
View raw message