A fast approximate Chinese word-matching algorithm is presented. The algorithm can be used to implement the Chinese fuzzy-matching conception. Based on the algorithm, an automatic Chinese text error correction approach using confusing-word substitution and language model evaluation is designed. Compared with Zhang's (1994) confusing-character substitution method, this new approach can deal with not only character substitution errors but also insertion, deletion and string substitution errors. Besides, the algorithm can handle Chinese "non-word" error, making it possible and easy to establish a two-level structure in Chinese spelling correction.
展开▼
