Takatoshi Yoshikawa

The new method of license key management for content distribution system

The importance of copyright protection has increased with the rapid spread of content distribution services. In one means of copyright protection, contents are protected by encryption. In this method, license keys are essential and distributed to decrypt contents. The paper presents a new method of license key management that can manage contents as groups. It uses some group keys in addition to license keys for decrypting content. Each of the group keys can decrypt all contents belonging to each group. As an example of applying the proposed method, a cooking recipe distribution system is introduced.

key management for grouping content

We propose a new method of post-processing for character recognition using pattern features and linguistic information. This method corrects errors in the recognition of handwritten Japanese sentences containing Kanji characters. This post-process method is characterized by having two types of character recognition. Improving the accuracy of the character recognition rate of Japanese characters is made difficult by the large number of characters, and the existence of characters with similar patterns. Therefore, it is not practical for a character recognition system to recognize all characters in detail. First, this post-processing method generates a candidate character table by recognizing the simplest features of characters. Then, it selects words corresponding to the character from the candidate character table by referring to a word and grammar dictionary before selecting suitable words. If the correct character is included in the candidate character table, this process can correct an error, however, if the character is not included, it cannot correct an error. Therefore, if this method can presume a character does not exist in a candidate character table by using linguistic information (word and grammar dictionary). It then can verify a presumed character by character recognition using complex features. When this method is applied to an online character recognition system, the accuracy of character recognition improves 93.5% to 94.7%. This proved to be the case when it was used for the editorials of a Japanese newspaper (Asahi Shinbun).