MeCab the Graduate School of Informatics, Kyoto University Unit - Joint
Research Project Communication Science Laboratories, Nippon Telegraph and
Telephone Corporation morphological analysis engine that was developed
through open source. The basic policy is to design a general-purpose
language-independent, dictionary and corpus.  (Conditional Random Fields
for the estimation of the parameters CRF has been used), ChaSen performance
has improved compared to the hidden Markov model is employed. In addition,
on average ChaSen , Juman , KAKASI run faster than.  Turnip Undaria pinnatifida
(wear eye) is the way the author's favorite food.

* Generic design that does not depend dictionary, the corpus
* Conditional Random Fields ( CRF on the basis of high analysis accuracy)
* ChaSen and KAKASI faster than
* Structure algorithm / data dictionary lookup is a fast TRIE structure Double-Array adopted.
* Libraries that can be re-entrant
* Various scripting language bindings (perl / ruby / python / java / C #)

WWW:   http://mecab.sourceforge.net/