Design And Implementation Of Intelligent Pinyin Input Method Based On Android Platform

With the growing popularity of the Android system and widespread use of applications such as microblog, WeChat, people more and more dependent on communicating through mobile phone, tablet and other handheld devices. Input information into these handheld devices quickly and efficient ly become the people’s most basic requirements for the device. So intelligent pinyin input method based on the Android platform has become the essential configuration of Android devices in China market.An intelligent input method that can input sentence in one t ime was realized in this topic based on the Android platform. The input method provides users with both Chinese input mode and English input mode and under Chinese input method it has higher accuracy of syllable-to-character conversion.In the paper, first of all, the fundamental way to realize input method was recommended on the basis of input method framework in the Android platform so that the input method can interact correctly with user. Then the implementation of soft keyboard and candidate word display view were introduced to offer user the operational interface according to user ’s requirements of input method. Finally, the syllable-to-character conversion model for tablet was designed and realized according to the tablet’s characteristics of the small storage space to enable users to input Chinese. The syllable-to-character conversion model can be divided into pinyin stream segmentation and syllable characters conversion two modules and these two modules both use the prefix tree to store data to improve storage efficiency. Pinyin stream segmentation module brought in the complete input cost and incomplete input cost to handle the problems imported by segmentation ambiguity, in this way, the complete segmentation result can be obtained as the input of syllable characters conversion module. In the syllable characters conversion module, the binary grammar model was used to achieve transformation. Binary grammar model was built by sorting the unary and binary statistical probability of phrase, then the Laplace smoothing technique and N- Best pruning algorithm was used to optimize the model. The end result of transformation was the best path calculated by the dynamic programming algorithm.The input method implemented based on the model introduced above can smoothly running in the Android devices with limited storage space. In addition to statement level input, the input method also support uncomplete spell input and phrases guess input function, it can provide users with convenient, fluent, accurate and humanized input experience. The input method has been tested strictly by the test department of company and it can stablely running in the Tint8 tablet project.