lucene-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Robert Muir (JIRA)" <j...@apache.org>
Subject [jira] Commented: (LUCENE-2458) queryparser shouldn't generate phrasequeries based on term count
Date Wed, 12 May 2010 13:02:41 GMT

    [ https://issues.apache.org/jira/browse/LUCENE-2458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12866557#action_12866557
] 

Robert Muir commented on LUCENE-2458:
-------------------------------------

{quote}
What are some real use-cases where this is "good"? WordDelmiterFilter seems like a good example
(eg, Wi-Fi -> Wi Fi).
It sounds like it's a very bad default for non-whitespace languages.
{quote}

Its a horrible bug! And to boot, i don't think it helps english much as a default either.
Here's a comparison on an english test collection (Telegraph collection with standardAnalyzer
+ porter):

||measure||T||TD||TDN||
|% of queries affected|6%|14%|32%|
|positionfilter improvement|+1.704%|+0.213%|+0.805%|

So, turning it off certainly doesn't hurt (I won't try to argue that this small "improvement"
by turning it off means anything).
For chinese, its a 10x improvement on TREC5/TREC6: obviously the bug is horrible there because
its generating phrase queries all the time.

{quote}
This seems like a good idea (since we seem to have real-world cases where it's very useful
and others where it's very bad)? Could/should it be per-analyzer? (ie, WDF would always do
this but, say, ICUAnalyzer would never). Or, per-token created?
{quote}

I am strongly opposed to this. My tibetan example with WDF or whatever above is an easy example.
I haven't seen any measured real-world example where this helps, subjectively saying "I like
this bug" isnt convincing me.

We don't need to push "what should be phrase query" onto analysis, it doesn't know from unicode
properties etc, what the user wanted.
We don't need to put hairy logic into things like StandardTokenizer, to determine if "the
user wanted a phrase query" or not in certain contexts.

Instead we should just do what the documentation says, and only issue phrase queries when
the user asks for one!!!!!!


> queryparser shouldn't generate phrasequeries based on term count
> ----------------------------------------------------------------
>
>                 Key: LUCENE-2458
>                 URL: https://issues.apache.org/jira/browse/LUCENE-2458
>             Project: Lucene - Java
>          Issue Type: Bug
>          Components: QueryParser
>            Reporter: Robert Muir
>            Priority: Critical
>
> The current method in the queryparser to generate phrasequeries is wrong:
> The Query Syntax documentation (http://lucene.apache.org/java/3_0_1/queryparsersyntax.html)
states:
> {noformat}
> A Phrase is a group of words surrounded by double quotes such as "hello dolly".
> {noformat}
> But as we know, this isn't actually true.
> Instead the terms are first divided on whitespace, then the analyzer term count is used
as some sort of "heuristic" to determine if its a phrase query or not.
> This assumption is a disaster for languages that don't use whitespace separation: CJK,
compounding European languages like German, Finnish, etc. It also
> makes it difficult for people to use n-gram analysis techniques. In these cases you get
bad relevance (MAP improves nearly *10x* if you use a PositionFilter at query-time to "turn
this off" for chinese).
> For even english, this undocumented behavior is bad. Perhaps in some cases its being
abused as some heuristic to "second guess" the tokenizer and piece back things it shouldn't
have split, but for large collections, doing things like generating phrasequeries because
StandardTokenizer split a compound on a dash can cause serious performance problems. Instead
people should analyze their text with the appropriate methods, and QueryParser should only
generate phrase queries when the syntax asks for one.
> The PositionFilter in contrib can be seen as a workaround, but its pretty obscure and
people are not familiar with it. The result is we have bad out-of-box behavior for many languages,
and bad performance for others on some inputs.
> I propose instead that we change the grammar to actually look for double quotes to determine
when to generate a phrase query, consistent with the documentation.

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.


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


Mime
View raw message