lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Shalin Shekhar Mangar <shalinman...@gmail.com>
Subject Re: Phonetic analysis with the spell-check component?
Date Mon, 27 Apr 2009 23:10:50 GMT
On Sun, Apr 26, 2009 at 11:55 PM, David Smiley @MITRE.org <DSMILEY@mitre.org
> wrote:

>
> It appears to me that the spell-check component can't build a dictionary
> based on phonetic similarity (i.e. using a Phonetic analysis filter).
>  Sure,
> you can go ahead and configure the spell check component to use a field
> type
> that uses a phonetic filter but the suggestions presented to the user are
> based on the indexed values (i.e. phonemes), not the original words.  Thus
> the user will be presented with a suggested phoneme which is a poor user
> experience.  It's not clear how this shortcoming could be rectified because
> for a given phoneme, there are potentially multiple words to choose from
> that could be encoded to a given phoneme.
>

Hmm. I think the problem here is that spell checker creates its own index
with the indexed tokens of a Solr field. So it does not have the original
words anymore. But if we could have an option to store the original words as
well into the spell check index, we could return them as suggestions.

Do you mind creating an Jira issue so that we don't forget about this?

-- 
Regards,
Shalin Shekhar Mangar.

Mime
  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message