As the academic and educational use of ICT tools advances, there are increasing opportunities in which the entry of numerical expressions as digital data is required, in e-learning classes as well as writing papers. Previously, the input of expressions generally involved entry by selection from a panel or using specific commands. Problems encountered such as low character recognition rates and difficulty correcting misrecognitions meant that handwritten input by tablet is not at a commercially viable stage.

The symbols in expressions can be recognized using existing character-recognition techniques, but the recognition of expression structures is difficult, which caused a drop in the accuracy of expression recognition. Accordingly, the pattern of the structure is predicted and a space (=MathBox) is displayed in advance for the user to input the expression, through which the structure is easily recognized and accuracy increased. In addition, we have succeeded in simplifying the correction of misrecognitions by providing example symbols that could possibly be input in the specified space.

Due to the fact that it is possible to use this system in more than just writing papers, that it can be used even under low-spec environments, and that misrecognition correction and input times can be reduced, it also applicable to CBT, e-learning and other usage conditions where real-time input is necessary.

Compared with the former recognition rate of around 70% for expression structures, we have achieved rates of around 90% by using the simple correction method. Input times compared with a conventional numerical expression editor were also decreased by 33%.

Sponsord research, Collaboration research, Technical consultation