Design And Implementation On User Labels In Mobile Internet

With the rapid development of the mobile internet and the spread of mobile devices,the users having great demands for data services,and the behavior gradually show the features of personalization and diversification.In the exploration of the "user-centred" operating mode,because of the simplification and low efficiency of traditional analysis method,operaters can not understand the characteristic of the behavior deeply.At the same time,with the explosive growth of the amount of data,the relational database architecture suffered great pressure and can not support the existing needs in market anymore.Therefore,understanding the characteristics of the behavior deeply through big-data related technology becomes more and more important for operaters to increase market share and promote business growth.Based on the project of "Ministry of Education-China Mobile" Research and Innovation Fund,a label system that covers the basic characteristics and personalized behavior characteristics of the user is designed in this paper.In the process of user segmentation,an adaptive RK-means algorithm based on Hadoop is proposed to realize the user segmentation.The main contents of this paper are as follows:Firstly,based on the deep research of signaling analysis technology,web crawler technology,data analysis technology,big data storage and processing technology,this article builds the mobile internet user label library covering user basic information,internet usage information,terminal usage information,telecom service relationship,marketing support information.Secondly,faced with the shortcomings of the traditional user behavior analysis system which based on relational database with a single node,the mobile internet user label system based on hadoop ecological environment is proposed to improve the data processing and data analysis ability.Thirdly,in the process of customer segmentation,this paper points out that the traditional K-means subdivision algorithm is sensitive to the initial clustering center and is easy to fall into the local optimum,the clustering centers is selected adaptively according to the number and data feature,and the RK-means algorithm is used to ensure the global optimization of the clustering center,and the paralleled adaptive RK-means algorithm is used to realize the customer segmentation.Finally,the paper fulfills the complete identification of the user characteristics and verifies the accuracy and reliability of the labels.The system is now running in the China Mobile(Hangzhou)R&D center and an operator in southwest.The scheme realized the identification of user characteristics,it provides effective data supports for data extraction and marketing.