Research On Ott Business Data Feature Extraction Method

With the rapid development of mobile Internet in recent years, applications on mobile devices make people’s lives more convenient.Taobao, prepaid recharge on mbile terminal and internet financial services affect people’s habits gradually. Application of OTT is especially pronounced influence on people’s communication habits. OTT means video and data services in the mobile Internet, typical OTT applications include Wechat, Miliao and Yixin. It’s more and more common for people to communicate with each other through Wechat. As the Wechat voice function becomes more popular, the voice service of the communication service provider face a huge threat.Therefore, it is particularly important to monitor the use of these OTT applicatons. Mobile internet network monitoring technology which is one of the important methods to obtain the operating conditions of mobile communications network has been a hot research, and OTT business data feature extraction plays an important role on it.This thesis studies OTT business data feature extraction method and proposes a new method based on data mining ideas. Firstly, we introduce the key technologies of OTT business identification, describe several important feature extraction methods and illustrate the hierarchical design and framework of OTT business data feature extraction system. Secondly we design data preparation module, analyze the operating principle, and detail the extraction, trans format ion, loading of the data payload. Then thesis analyzes two key algorithms-PDA algorithm and SDA algorithm, constructs data analysis module based on two algorithms,gets the features of typical OTT application,sets up the environment of the OTT business data feature extraction system,carries out a test experiment for version4.0of Wechat, gets the features of typical OTT business and analyzes the test result. Finally, we present deficiencies of the feature extraction method and recommendations for follow-up development work.