[
https://issues.apache.org/jira/browse/LUCENE-1606?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12782173#action_12782173
]
Robert Muir commented on LUCENE-1606:
-------------------------------------
bq. But, take the abcd*1234 case - you first seek to abcd, the terms enum finds
(say) abcdX1234 - don't you (DFA) know at this point that next possible
candidate is abcdY1234? Ie, you should seek to that term? (Doing next() at that
point is most likely a waste, and anyway the enum will turn your seek into a
next if it's "close").
pretty sure I know this information, but in general i only seek on mismatches,
I think for the reason that there can be a lot of seeks for AB* (say 1 million
terms).
so instead i wait for a mismatch until i seek again, I think tests showed this
due to what you mentioned below?
bq. That said, seeking on trunk is alot more costly than seeking on flex,
because trunk has to make a new [cloned] SegmentTermEnum for each seek.
I think that might be whats killing me, when i ran tests on lucene 2.9 or
whatever.
We should retest performance on flex.
This is why i said, significant rework of this maybe should take place in flex
(although I still think this is an improvement for trunk already), to fully
take advantage of it.
> Automaton Query/Filter (scalable regex)
> ---------------------------------------
>
> Key: LUCENE-1606
> URL: https://issues.apache.org/jira/browse/LUCENE-1606
> Project: Lucene - Java
> Issue Type: New Feature
> Components: Search
> Reporter: Robert Muir
> Assignee: Robert Muir
> Priority: Minor
> Fix For: 3.1
>
> Attachments: automaton.patch, automatonMultiQuery.patch,
> automatonmultiqueryfuzzy.patch, automatonMultiQuerySmart.patch,
> automatonWithWildCard.patch, automatonWithWildCard2.patch,
> BenchWildcard.java, LUCENE-1606-flex.patch, LUCENE-1606-flex.patch,
> LUCENE-1606.patch, LUCENE-1606.patch, LUCENE-1606.patch, LUCENE-1606.patch,
> LUCENE-1606.patch, LUCENE-1606.patch, LUCENE-1606.patch, LUCENE-1606.patch,
> LUCENE-1606.patch, LUCENE-1606.patch, LUCENE-1606_nodep.patch
>
>
> Attached is a patch for an AutomatonQuery/Filter (name can change if its not
> suitable).
> Whereas the out-of-box contrib RegexQuery is nice, I have some very large
> indexes (100M+ unique tokens) where queries are quite slow, 2 minutes, etc.
> Additionally all of the existing RegexQuery implementations in Lucene are
> really slow if there is no constant prefix. This implementation does not
> depend upon constant prefix, and runs the same query in 640ms.
> Some use cases I envision:
> 1. lexicography/etc on large text corpora
> 2. looking for things such as urls where the prefix is not constant (http://
> or ftp://)
> The Filter uses the BRICS package (http://www.brics.dk/automaton/) to convert
> regular expressions into a DFA. Then, the filter "enumerates" terms in a
> special way, by using the underlying state machine. Here is my short
> description from the comments:
> The algorithm here is pretty basic. Enumerate terms but instead of a
> binary accept/reject do:
>
> 1. Look at the portion that is OK (did not enter a reject state in the
> DFA)
> 2. Generate the next possible String and seek to that.
> the Query simply wraps the filter with ConstantScoreQuery.
> I did not include the automaton.jar inside the patch but it can be downloaded
> from http://www.brics.dk/automaton/ and is BSD-licensed.
--
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: java-dev-unsubscr...@lucene.apache.org
For additional commands, e-mail: java-dev-h...@lucene.apache.org
