4686a9255e
pre-everything target - update pkg-descr PR: 175195 Submitted by: maintainer, yasu@utahime.org
18 lines
1 KiB
Text
18 lines
1 KiB
Text
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/
|