search

unobliged / EDICT2_Parser

This is a parser and search webapp for Japanese dictionary file EDICT2.
1 stars 0 forks source link

Morphological analysis? #3

Open unobliged opened 11 years ago

unobliged commented 11 years ago

If a person enters something like a conjugated verb, the simple search will return nothing. A beginner would not search by the root kanji or may just be copy-pasting something they have seen. Deriving the morpheme might be difficult even with integration of Mecab, but it might be worth experimenting with. Try implementing Mecab or coming up with some alternative to fuzzy search.

unobliged commented 11 years ago

MeCab/Natto doesn't seem to handle conjugation/deconjugation in the way I expected; a different (or custom) tool is needed. Perhaps the following?: http://kanjidict.stc.cx/verb=oishii | http://kanjidict.stc.cx/japverb3.php http://www.geom.uiuc.edu/~burchard/nihongo/ | http://www.geom.uiuc.edu/~burchard/nihongo/endings.html http://www.sljfaq.org/afaq/verb-conjugation.html http://www.jlpt-practice.com/lesson/verbs_conjugation.aspx http://www.as.ua.edu/nihongo/verb_table_roomaji.htm

It shouldn't be too hard to make a deconjugator...I hope.

unobliged commented 11 years ago

Will be making a separate deconjugator app, should make it easier to test and improve. It can be integrated later (perhaps a new app?).