Kanji Sentence Segmentation Method and Application to Chinese Character Error Checking (CEC) System

摘要

The present invention discloses a scheme for efficiently detecting errors in Chinese character text. By considering error strings, names and places, or organization names, numbers, combinations of numbers and metric units commonly implied with words in the regular dictionary as common segmentation units and assigning appropriate weights to them, the most likely A dynamic programming method integrated in the Chinese character error checking system for selecting the segmentation is used. In this way, the CEC system of the CRL is able to achieve the highest error detection rate while maintaining the false alarm rate as low as among similar-purpose systems.