AUTOMATIC RULE ACQUISITION BASED ON TRANSFORMATION FOR A CHINESE SYLLABLE TO CHARACTER CONVERSION SYSTEM

Here described is a new approach to automatically learning error correct rules from online text of unlimited size for a syllable to character conversion system.It is shown that this method not only can improve the precision of the result of syllable to character conversion, but also can enhance the flexibility and expansibility of the system.This automatically learning model together with the former syllable to character conversion system forms a new type of rule statistic hybrid syllable to character conversion system.