lucene-java-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Erick Erickson" <erickerick...@gmail.com>
Subject Re: Lucene shows parts of search query as a HIT
Date Thu, 19 Jul 2007 14:41:00 GMT
QueryParser.setDefaultOperator

On 7/19/07, Askar Zaidi <askar.zaidi@gmail.com> wrote:
>
> Hey Erik,
>
> How can I change the default Lucene OR property to AND.
>
> When I tried query.toString(), I got
>
> contents:w contents:chan contents: kim
>
> Thats fine, but its doing OR, how can I make it AND so that it shows:
>
> contents: W Chan Kim ??
>
> thanks a ton !
>
> AZ
>
> On 7/19/07, Erick Erickson <erickerickson@gmail.com> wrote:
> >
> > You say there's only one document and you added many. The line
> >
> > IndexWriter writer = new IndexWriter(indexPath, new StandardAnalyzer(),
> > true);
> >
> > blows away any existing index data and starts over. If you're calling
> > this fragment for each document, you'll always have only one doc. Try
> > changing the 'true' to 'false'. Or better yet, open the writer outside
> the
> > document add routines and pass it in...
> >
> > I still have no clue why you're getting the results you are.
> > StandardAnalyzer
> > doesn't stem, so that's probably not it. Try query.toString() to see
> what
> > the actual query you're submitting is. Then, I'd think about playing
> with
> > that
> > query in Luke to see what's happening.
> >
> > Best
> > ErickOn 7/19/07, Ard Schrijvers <a.schrijvers@hippo.nl> wrote:
> > >
> > > Hello Askar,
> > >
> > > Which analyzer are you using for indexing and searching? If you use an
> > > analyzer that uses stemming, you might see that "change", "changing",
> > > "changed", "chan" etc al get reduced to the same word "chan".
> > >
> > > In luke you can test with plugins that show you what tokens are
> created
> > > from your analyzer,
> > >
> > > Regards Ard
> > >
> > >
> > > >
> > > >
> > > > Hey Guys,
> > > >
> > > > I just checked my Lucene results. It shows a document with
> > > > the word hit
> > > > "change" when I am searching for "Chan", and it considers
> > > > that as a hit. Is
> > > > there a way to stop this and show just the exact word match ?
> > > >
> > > > I started using Lucene yesterday, so I am fairly new !
> > > >
> > > > thanks
> > > > AZ
> > > >
> > > > On 7/18/07, Erick Erickson <erickerickson@gmail.com> wrote:
> > > > >
> > > > > Are you sure that the hit wasn't on "w" or "kim"? The
> > > > > default for searching is OR...
> > > > >
> > > > > I recommend that you get a copy of Luke (google lucene luke)
> > > > > which allows you to examine your index as well as see how
> > > > > queries parse using various analyzers. It's an invaluable tool...
> > > > >
> > > > > Best
> > > > > Erick
> > > > >
> > > > > On 7/18/07, Askar Zaidi <askar.zaidi@gmail.com> wrote:
> > > > > >
> > > > > > Hey folks,
> > > > > >
> > > > > > I am a new Lucene user , I used the following after indexing:
> > > > > >
> > > > > > search(searcher, "W. Chan Kim");
> > > > > >
> > > > > > Lucene showed me hits of documents where "channel" word
> > > > existed. Notice
> > > > > > that
> > > > > > "Chan" is a part of "Channel" . How do I stop this ?
> > > > > >
> > > > > > I am keen to find the exact word.
> > > > > >
> > > > > > I used the following, before the search method:
> > > > > >
> > > > > > IndexWriter writer = new IndexWriter(indexPath, new
> > > > StandardAnalyzer(),
> > > > > > true);
> > > > > >
> > > > > >                                 writer.addDocument
> > > > > > (createDocument(item,words));
> > > > > >                                 writer.optimize();
> > > > > >                                 writer.close();
> > > > > >                                 searcher = new
> > > > IndexSearcher(indexPath);
> > > > > >
> > > > > > thanks !
> > > > > >
> > > > > > AZ
> > > > > >
> > > > >
> > > >
> > >
> > > ---------------------------------------------------------------------
> > > To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
> > > For additional commands, e-mail: java-user-help@lucene.apache.org
> > >
> > >
> >
>

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