This paper presents a constructive approach for solving problems associated with inputting Chinese characters for the recognition processing by digital computers. The idea presented here is "Construction Pattern Recognition (CPR)", which is essentially different from the existing Chinese character inputting methodologies. With the CPR, a Chinese character is treated as a graph rather than as a character of a language. Its construction pattern could be encoded by the "Construction Pattern Internal Coding System (CPICS)" and presented in a grid form with the "equivalence classes" as well as the matrix representation. The CPR method is developed based upon human learning and recognition of Chinese characters, applying the theory graph to form the body of the CPR algorithm. A proposed computer neural network will be presented as an upgradable learning and recognition algorithm for the future research on the CPR method.
展开▼