Definition of words depends on morphological analyzer
ChaSen
Segmentation is still non-deterministic
(Ex.) ``とも + に'' ``ともに''
Align after concatenating compounds
Notation (use of Kanji/Kana) is also arbitrary
(Ex.) ``十二'' ``12''
(Ex.) ``一人'' ``ひとり''
But ``操作'' ``捜査''
Define accuracy in both Kanji-basis & Kana-basis