FAQ
Found the reason, it is a bug IMHO.

The example should be:

A: term1^5 term2^6 term3^7
B: term1^5E-4 term2^6E-4 term3^7E-4
C: term1^0.0006 term2^0.0006 term3^0.0007

A & C suppose return the same rank
B is different

Since B will be parsed as: term1^5 E-4 term2^6 E-4 term3^7 E-4

The parser takes E-4 as another term.

Can someone help to log it as a "parsing bug" if we consider it as a
bug. Thanks.

Charlie

---
Wednesday, September 27, 2006, 4:49:11 PM, you wrote:

I assume you mean that the set of matches is hte same, but the scores (and
possibly the order) are different correct?
The IndexSearcher.explain methods should help make the reason clear --
compare the output for each query when looking at the same docIds.
I suspect what you'll find is that with the low boost values, the
fieldNorms become more significant resulting in a change of order.
: Date: Wed, 27 Sep 2006 16:41:52 -0500
: From: charliecmo@gmail.com
: Reply-To: java-user@lucene.apache.org, Charlie Zhao <CharlieCMO@gmail.com>
: To: java-user@lucene.apache.org
: Subject: strange behavior 4 query term boost
:
: Hello,
:
: I don't understand why the following two queries give totally different results.
:
: term1^5 term2^6 term3^7
:
: term1^0.0005 term2^0.0006 term3^0.0007
:
: Can anyone explain? Thanks.
:
: (lucene2.0, using TopDocs)
:
: --
: Thanks,
: Charlie




---------------------------------------------------------------------
To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
For additional commands, e-mail: java-user-help@lucene.apache.org

Search Discussions

  • Mike Klaas at Sep 28, 2006 at 6:40 pm

    On 9/27/06, charliecmo@gmail.com wrote:
    Found the reason, it is a bug IMHO.

    The example should be:

    A: term1^5 term2^6 term3^7
    B: term1^5E-4 term2^6E-4 term3^7E-4
    C: term1^0.0006 term2^0.0006 term3^0.0007

    A & C suppose return the same rank
    B is different

    Since B will be parsed as: term1^5 E-4 term2^6 E-4 term3^7 E-4

    The parser takes E-4 as another term.

    Can someone help to log it as a "parsing bug" if we consider it as a
    bug. Thanks.
    A "bug" exists when code does not perform to
    specifications/documentation, not when it doesn't perform to
    everybody's expectations. Supporting scientific notation would be a
    new feature.

    -Mike

    ---------------------------------------------------------------------
    To unsubscribe, e-mail: java-user-unsubscribe@lucene.apache.org
    For additional commands, e-mail: java-user-help@lucene.apache.org

Related Discussions

Discussion Navigation
viewthread | post
Discussion Overview
groupjava-user @
categorieslucene
postedSep 28, '06 at 2:51a
activeSep 28, '06 at 6:40p
posts2
users2
websitelucene.apache.org

2 users in discussion

Charliecmo: 1 post Mike Klaas: 1 post

People

Translate

site design / logo © 2022 Grokbase