lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hoss Man (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (LUCENE-3013) I wish Lucene query explanations were easier to localise
Date Fri, 20 May 2011 20:22:47 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-3013?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13037062#comment-13037062
] 

Hoss Man commented on LUCENE-3013:
----------------------------------

bq. 2. Add subclasses of Explanation with enough methods for callers to interrogate the individual
details of the explanation instead of outputting it as a monolithic string.

I think this is definitely the way to go ... most Explanation objects tend to fall into two
fairly basic categories...

* atomic value (ie: the norm for a field, boost multiplier, etc...)
* mathmatic operation on sub explanations (ie: product of all child explanations; addition
of all child explanations; etc..)

I think each of those should be concrete subclasses of ComplexExplanation with methods for
some of the fixed metadata (ie: the type of math operator), and i think we should then add
more additional subclasses of each of those for the things that are really common in Lucene
with additional methods where appropriate...

* QueryNormExplanation extends AtomicValueExplanation (no special methods)
* IDFExplanation extends AtomicValueExplanation (getDocFreq, getMaxDocs)
* QueryWeightExplanation extends ProductExplanation (getTerm)
* FieldNormExplanation extends AtomicValueExplanation (getField, getDocId)
* TFExplanation extends AtomicValueExplanation (getTerm, getFreq)
* etc...

...simple clients could then continue to use the simple getValue/getDescription methods, but
more complex clients could use introspection to see if it's a ttype they want to generate
a more user freindly "explanation" our of.

> I wish Lucene query explanations were easier to localise
> --------------------------------------------------------
>
>                 Key: LUCENE-3013
>                 URL: https://issues.apache.org/jira/browse/LUCENE-3013
>             Project: Lucene - Java
>          Issue Type: Wish
>          Components: core/query/scoring
>            Reporter: Trejkaz
>
> Often users ask us to provide a nice UI to explain why a document matched their query.
 Currently the strings output by Explanation are very advanced, and probably only understandable
to those who have worked on Lucene.  I took a shot at trying to make them friendlier, but
it basically came down to parsing the strings it output and trying to figure out what kind
of query was at each point (the inability to get to a Query from the Explanation is a small
part of the problem here), formulating the result into readable English.  In the end it seems
a bit too hard.
> The solution to this could be done in at least two ways:
> 1. Add getLocalizedSummary() / getLocalizedDescription() method(s) and use resource bundles
internally.  Projects wishing to localise these could add their own resource bundles to the
classpath and/or get them contributed to Lucene.
> 2. Add subclasses of Explanation with enough methods for callers to interrogate the individual
details of the explanation instead of outputting it as a monolithic string.
> I do like the tree structure of explanations a lot (as it resembles the query tree),
I just think there is work to be done splitting up the strings into usable fragments of information.

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira

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


Mime
View raw message