Research on Handwriting Recognizer Technology based on Comprehensive Data Mining and Collections

In recent years, the emergence of comprehensive data mining and collection technology has promoted the development of handwritten Chinese character recognition. This paper uses the Softmax confidence interval to statistically analyze a large number of handwritten recognition samples, and analyzes the confidence interval and handwritten Chinese character sample recognition from the recognition rate and the confidence distribution relationship. The multi-level presentation of recognition results and dual recognition engines are designed to improve the user’s writing rate and recognition accuracy. In the cloud, a sticky load balancing node is built to keep the data synchronization between the client and the cloud, and to achieve concurrent user access processing. Three similar word discovery methods based on confidence interval sorting and information entropy are applied to the realization of the construction of handwriting recognition data service platform.