Study on Chinese error checking

The word-level error checking in Chinese has been discussed. During words Segmentation, the algorithm is divided into two steps. Firstly, the longest match algorithm of forward heuristic, reverse backtracking and the recursive word segmentation algorithm of left and right sub-segment have been used to divide the text into more small loose strings. Secondly, the forward longest matching algorithm has been used to merge casual strings backward as far as possible, and the casual strings being segmented are the basis of error checking operation later. In the system of error detecting, an algorithm based on similar pronunciation strategy has been introduced. This strategy uses large-scale lexicon (340 millions) as the basis of data analysis. Then, error checking algorithm that based on similar shape which includes similar character table, Wubi repeat-code table, and Zhengma repeat-code table has been introduced to check character error. Experiments show satisfactory results.