Re: request a new feature in fuzzystrmatch

On Fri, May 17, 2013 at 8:56 PM, Liming Hu <dawninghu(at)gmail(dot)com> wrote:
> On 5/17/2013 8:23 AM, Atri Sharma wrote:
>>
>> On Fri, May 17, 2013 at 8:51 PM, Liming Hu <dawninghu(at)gmail(dot)com> wrote:
>>>
>>> Hi Atri,
>>>
>>> Thanks for the quick response.
>>>
>>> levenshtein edit distance defines operations of: insertion, deletion,
>>> modification.
>>> Levenshtein-Damerau edit distance defines operations of: insertion,
>>> deletion, modification
>>> and transposition (
>>> it will be two operations in levenshtein edit distance:
>>> one deletion, and one insertion,
>>> ).
>>>
>>> In spelling checker/corrector, i.e.
>>> levenshtein('cta', 'cat') will return 2.
>>> but Levenshtein-Damerau('cta', 'cat') will return 1.
>>> if the maximum error is 1, we can not get 'cat'.
>>>
>>> In practice, Levenshtein-Damerau is more widely used than Levenshtein.
>>> I believe you notice "Google automated search suggestions", they use
>>> Levenshtein-Damerau.
>>>
>> Sounds interesting. How can we build this over our current
>> implementation, or do we need to build it from scratch?
>>
>> Regards,
>>
>> Atri
>
> I know how to write the code, but just need approval of accepting into the
> new version.

Well, acceptance depends largely on the implementation and actual
benefit statistics. I would suggest implementing a basic version and
then demonstrating its potential benefits here. It will lead to
clearer ideas for us and lead to improvements in the implementation.

Regards,

Atri
--
Regards,

Atri
l'apprenant