Hi,
Is there any successfull spell checker implementation for an agglunative language (Turkish, Finnis, Hungarian etc.) based on current myspell engine? Do one need to modify myspell deeply or just use current affix definition mechanism and dictionary structure to make a spell checker for such a language.
Dear Mehmet,
I think, you need Hunspell. (I don't know capabilities of your Zemberek tool. We have a Java version of Hunspell called JSpell.) Hunspell uses MySpell's affix definition mechanism and dictionary structure with little, but important extensions. For linguist, we have also a great preprocessor tool called Hunlex.
I will extends this short descriptions with the basics (base MySpell format), and introduction of other extension too: http://nlp.mokk.bme.hu/resources/hunmorph/extensions-en/
Best regards:
Laci
Thanks for all the answers.
Dear Laci,
Thanks for the answer. Zemberek is a Turkish NLP library and spell checker for OOo implemented in Java. I asked this question because we had a long discussion with one of ye members of this mailing list on why not adopting myspell for a Turkish spell checker and introducing a completely new architecture . Also there has been some requests on implementing the turkish spell checker in c++, so I was curious whether it is possible - or easy to use myspell in an agglunative language. As I browsed through myspell code I thought there may be problems because Turkish (probably like Hungarian) has more complex affix rules than Indo-european languages. We will probably continue working on Java implementation, because Zemberek is planned to be a generalized NLP library not only an extension.
What I need is how can I find all affix rules and meanings in myspell in case someone wants to make a Turkish spell checker in C++. Can anyone points out a documentation on this?
Mehmet D. Akin http://zemberek.dev.java.net
--------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
