You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@lucene.apache.org by "Dawid Weiss (JIRA)" <ji...@apache.org> on 2011/02/21 14:51:38 UTC
[jira] Updated: (LUCENE-2933) Two-stage state expansion for the
FST: distance-from-root and child-count criteria.
[ https://issues.apache.org/jira/browse/LUCENE-2933?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Dawid Weiss updated LUCENE-2933:
--------------------------------
Attachment: LUCENE-2933.patch
Implementation of node depth expansion into an array. This patch also piggybacks package-scope getBytes() so that other classes can access the internal representation (specifically, Util.toDot)
> Two-stage state expansion for the FST: distance-from-root and child-count criteria.
> -----------------------------------------------------------------------------------
>
> Key: LUCENE-2933
> URL: https://issues.apache.org/jira/browse/LUCENE-2933
> Project: Lucene - Java
> Issue Type: Improvement
> Components: Index
> Reporter: Dawid Weiss
> Assignee: Dawid Weiss
> Priority: Trivial
> Fix For: 4.0
>
> Attachments: LUCENE-2933.patch
>
>
> In the current implementation FST states are expanded into a binary search on labels (from a vector of transitions) when the child count of a state exceeds a given predefined threshold (NUM_ARCS_FIXED_ARRAY). This threshold affects automaton size and traversal speed (as it turns out when benchmarked). For some degenerate (?) data sets, close-to-the-root nodes could have a small number of children (below the threshold) and yet be traversed on every single seek.
> A fix of this is to introduce two control thresholds:
> EXPAND state if (distance-from-root <= MIN_DISTANCE || children-count >= NUM_ARCS_FIXED_ARRAY)
> My plan is to create a data set that will prove this first and then to implement the workaround above.
--
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