Research On Optimization Of Cloud Input Method For Set - Top Box

In video on demand, the key-word search function is a highly effective means for viewers to search for the target program from massive resources, and the Internet business need the users to submit information, all above requires set-top boxes have Chinese input capacity. Set-top boxes mainly have remote control as its input terminal, its input efficiency is low. Because of its weak processing capacity and lack of storage, it cannot provide intelligent input. Combining cloud computing and set-top boxes, this paper proposed a cloud input method oriented set-top boxes, improving the efficiency of input method through phrase learning, words’order adjustment and so on:key information of remote control is sent to cloud servers over the network, servers do input method computing, and send back the candidate words to the user, and the servers analysis the user’s choice, do words’order adjustment. With the powerful computing and storage capacity, the servers provide users with intelligent input service, while reduce the computing cost of set-top boxes.Traditional Chinese thesaurus is great, but in video-on-demand, the viewers find a program mainly by the program names, actors’names and story content, and the input information will be effective only if is included in the meta-data. Therefore, we can improve the efficiency of the input method by full-text search technology:processing the meta-data using Chinese word segmentation technology, then building program-name thesaurus, actor-name thesaurus, virtual characters names thesaurus and story content thesaurus. When the viewers request input method services, the input method servers only search from the corresponding thesaurus. Compared with traditional, the scale of the thesaurus was reduced so that the speed of retrieval was increased; on the other hand, the number of the candidate words was decreased, which cut or even avoided page-turns.According to the high demand of names library of set-top box input method, this paper improves the traditional name recognition method based on surnames. Most virtual role names in the story are generally not included in the traditional dictionary, and because its structure is complex and diverse, it is difficult for the Chinese name recognition technology to extract them. This paper proposes a new Chinese names recognition algorithm:remain the fragments after the Chinese word segmentation processing and extract the unknown names by further processing. The experiments turn out effect of the new Chinese names recognition algorithm is good.At the same time, this paper does lots of optimization in building the cloud input method service, which improved the service performance and stability of the input method.