Abstract: An on-line handwriting mathematics editing system in pursuance of ubiquitous computing theory and improving the recognition rates and interaction rates is designed and implemented. In this system, we applied context-aware technique to help segmentation and resolve the ambiguity meaning of characters, applied best-choice and direct-assignment method to correct the recognition results, applied local-magnification to segment the adjoining strokes, and eventually provided an algorithm for supplementing strokes. As we can see from experiments, the recognition rate and interaction rate are improved effectively by applying these techniques. At the same time, the system shows natural and implicit interaction features.