[Snowball-discuss] different results from java vs. C for finnish

Martin Porter martin at porterloo.wanadoo.co.uk
Wed Jan 13 18:06:35 GMT 2010


Robert,

Ah, thank you for pointing this out. The C code is correct here, and the
java code in error. The java used to be correct ('arteeseen' being in our
test vocabulary for Finnish), but I think Richard Boulton made a change to
the java codegenerator late last year in a region of the code which would be
sensitive to this problem.

Richard, can you comment on this?

Martin



At 09:58 AM 1/12/2010 -0500, Robert Muir wrote:
>
>Hello Snowball developers,
>
>I was running some tests with the snowball vocabulary data, and
>noticed different results across the generated C code versus the
>generated java code for Finnish and Lovins
>
>A simple example is 'aarteeseen' with Finnish, the c code stems it to
>'aart' but the java code stems it to 'aartees'
>
>-- 
>Robert Muir






More information about the Snowball-discuss mailing list