| With the development of the Internet era, as the most simple and widely-used method of man-computer interaction, handwriting recognition technique plays an important role in intelligent device. Handwriting character recognition is an import approach in human-computer interaction. However, the technical threshold of this technology is high, and it will take a long time to develop the application for industry. For the industry, it needs an integrated platform which can offer handwriting recognition and handwriting data resources, including samples scheduling. The development of cloud computing and big data technology makes it possible to build the platform for offering the data service about the handwriting samples. The implementation of this handwriting recognition data as Service offers a solution for those needs.In recent years the convolution neural network technology promotes the development of handwritten character recognition. We use Softmax confidence interval for a large number of handwriting recognition sample for statistical analysis and draw some conclusion from the relationship between the confidence interval and handwriting character samples, and apply the conclusion on some research, including relabeling the mis-labeled handwriting samples, and finding the pairs of similar characters based on candidate, confidence and entropy. Also, We build the handwriting character recognition data as a service, and the handwriting samples scheduling is based on the confidence interval analysis.The service platform we implied is specially presented for the field of handwriting recognition. We achieve it from data storage, data service and access layer. Based on handwriting data, we offer the following services: a) large scale data storage, providing foundation for data service. b) Online handwriting recognition service, sample scheduling service and admin service providing user privilege and handwriting data management. These services are the core business logic for the platform. c) Access layer implies the authentication services for inside and outside layer and implies SDK using 3 development language, which reduces the complexity for users to use the service of the platform. |