[
https://issues.apache.org/jira/browse/SOLR-3667?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13421404#comment-13421404
]
Tim Kornhammar commented on SOLR-3667:
--------------------------------------
I've searched a lot for that information and all I needed where the keyword
lossy...
If it isn't a bug, how come the score can be a float with precision but the
boost which would be a part of the score cannot?
Also the docs about encodeNormValue() says "The encoding uses a three-bit
mantissa, a five-bit exponent, and the zero-exponent point at 15, thus
representing values from around 7x10^9 to 2x10^-9 with about one significant
decimal digit of accuracy." so how is it lossy?
My current solution is to use:
bq. q={!boost b=ranking}keyword:test
And by letting computeNorm() return 1.
But that should lead be a performance loss, to change the scoring at search
time so not really the best solution..?
> Solr::fieldNorm() gives weired values with CustomSimilarity.
> ------------------------------------------------------------
>
> Key: SOLR-3667
> URL: https://issues.apache.org/jira/browse/SOLR-3667
> Project: Solr
> Issue Type: Bug
> Components: update
> Affects Versions: 3.6
> Environment: Ubuntu 12.04
> Reporter: Tim Kornhammar
>
> I have only seen floats being "smaller" int values like "13.33" becoming
> "12.00", it is a very strange behavior. It behaves the same if the boost is
> on the keyword field...
> Example of update data:
> {code:xml|borderStyle=solid}
> <add>
> <doc boost="13.33">
> <field name="keyword">test one</field>
> <field name="ranking">13.33</field>
> </doc>
> <doc boost="10.00">
> <field name="keyword">test two</field>
> <field name="ranking">10.00</field>
> </doc>
> </add>
> {code}
> Example of solr response, ranking is what I have tried to give both the
> document boost and field boost individualy.
> {code:xml|borderStyle=solid}
> <result>
> <doc>
> <float name="score">12.0</float>
> <str name="keyword">test one</str>
> <float name="ranking">12.0</float>
> </doc>
> <doc>
> <float name="score">10.0</float>
> <str name="keyword">test two</str>
> <float name="ranking">10.0</float>
> </doc>
> </result>
> <lst name="debug">
> <lst name="explain">
> <str name="1">
> 12.0 = (MATCH) fieldWeight(keyword:test in 0), product of:
> 1.0 = tf(termFreq(keyword:test)=1)
> 1.0 = idf(docFreq=2, maxDocs=2)
> 12.0 = fieldNorm(field=keyword, doc=0)
> </str>
> <str name="2">
> 10.0 = (MATCH) fieldWeight(keyword:test in 0), product of:
> 1.0 = tf(termFreq(keyword:test)=1)
> 1.0 = idf(docFreq=2, maxDocs=2)
> 10.0 = fieldNorm(field=keyword, doc=0)
> </str>
> </lst>
> </lst>
> {code}
> My CustomSimilarity class:
> {code:title=CustomSimilarity.java|borderStyle=solid}
> import org.apache.lucene.index.FieldInvertState;
> import org.apache.lucene.search.DefaultSimilarity;
> public class CustomSimilarity extends DefaultSimilarity {
> public CustomSimilarity() {
> super();
> }
> public float tf(int freq) {
> return (float) 1.0;
> }
> public float idf(int docFreq, int numDocs) {
> return (float) 1.0;
> }
> public float computeNorm(String field, FieldInvertState state) {
> return state.getBoost();
> }
> }
> {code}
--
This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators:
https://issues.apache.org/jira/secure/ContactAdministrators!default.jspa
For more information on JIRA, see: http://www.atlassian.com/software/jira
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
