[jira] [Comment Edited] (CODEC-248) language.DaitchMokotoffSoundex gives overly broad results for tokens containing RS

["Melloware (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/CODEC-248?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16852963#comment-16852963 ] 

Melloware edited comment on CODEC-248 at 5/31/19 1:04 PM:
----------------------------------------------------------

[~bkazez] I am not sure this is correct.  If I remove "rs" from the table we get unit tests that fail.  Namely.

{code:java}
        Assert.assertEquals("734000|739400", soundex("Peters"));
        Assert.assertEquals("734600|739460", soundex("Peterson"));
{code}

Which according to this documentation are correct:

https://en.wikipedia.org/wiki/Daitch%E2%80%93Mokotoff_Soundex

Also this test fails..

{code:java}
        // GERSCHFELD
        // G-E-RS-CH-F-E-L-D
        // 5--4/94-5/4-7-8-3 -> wrong
        // 5--4/94-5/--7-8-3 -> correct
        Assert.assertEquals("547830|545783|594783|594578", soundex("GERSCHFELD"));
{code}

GERSCHFELD would become just 594783




was (Author: melloware):
[~bkazez] I am not sure this is correct.  If I remove "rs" from the table we get unit tests that fail.  Namely.

        Assert.assertEquals("734000|739400", soundex("Peters"));
        Assert.assertEquals("734600|739460", soundex("Peterson"));

Which according to this documentation are correct:

https://en.wikipedia.org/wiki/Daitch%E2%80%93Mokotoff_Soundex

Also this test fails..

        // GERSCHFELD
        // G-E-RS-CH-F-E-L-D
        // 5--4/94-5/4-7-8-3 -> wrong
        // 5--4/94-5/--7-8-3 -> correct
        Assert.assertEquals("547830|545783|594783|594578", soundex("GERSCHFELD"));

GERSCHFELD would become just 594783



> language.DaitchMokotoffSoundex gives overly broad results for tokens containing RS
> ----------------------------------------------------------------------------------
>
>                 Key: CODEC-248
>                 URL: https://issues.apache.org/jira/browse/CODEC-248
>             Project: Commons Codec
>          Issue Type: Bug
>            Reporter: Ben Kazez
>            Priority: Minor
>
> I am using Apache commons codec in Elasticsearch (via Lucene).
> # GIERSZLIK codes to 548500 or 594850
> # GOTSALK codes to 548500
> # These names don't sound alike, but the matching codes means a search for one returns the other.
> Solution: I exchanged emails with Gary Mokotoff, co-creator of the algorithm, who said:
> {quote}I would drop RS from the table. ... I cannot think of any language where RS is pronounced "S" (4).{quote}
>  
>   



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)