Research And Application Of Broadband Family Portrait Based On Big Data Analysis

As the network environment continuously improve and the network functions become more powerful,more and more smart devices enter the Internet family.How to better understand users and improve user experience has become the most important topic in the field of Internet services.Massive home network user data has provided abundant materials for operators and e-commerce platforms.How to make good use of these data has become a hot topic for researchers.Researching portraits of the family users is one of them.With the continuously expansion of the scale of Internet users and the e-commerce market,it is far from enough to study only the personal behavior of users.This article adopts DPI deep packet inspection technology to collect,clean,extract and analyze all network traffic data of home users under the broadband account of the operator,and uses the DBSCAN algorithm to identify the real user of the family account,URL accessed by users under and terminal information quickly and accurately.Through the collection and research of the actual shopping data of the users on the e-commerce website side,the research proposes a study strategy of family portrait construction.By calculating the weight of user tags and improving user information,the strategy can build the user's portraits under home broadband.Based on user portraits,considering the interaction between users in the home,setting different thresholds to obtain home users' real-time interest sets,the research uses the Bayesian dynamic model to predict the family's potential interest sets based on the actual home network consumption data.Finally,the multivariate linear regression equation is used to unify the two data to construct a more comprehensive and accurate family portrait.On the basis of family portraits,a combination of weighted algorithms is used to obtain similar behavioral preferences for families and similar products and to regularly recommend products and services that are potentially of interest to relevant families.The effectiveness and reliability of this personalized home recommendation system were verified through the establishment of an experimental environment,combined with the user click-through rate and ad conversion rate as measured by third-party platforms.Under the family fixed-line broadband,a complete family portrait is created by combining the data of the complete user's Internet access data owned by the operator and the actual consumption data provided by the e-commerce platform.Personalized recommendations for home broadband users not only provide operators with a reliable way to realize cash flow,but also provide tremendous support to the e-commerce platform user mining,ad placement,product recommendation.And then the result can provide more convenient services for home users under fixed-line broadband and greatly improve the quality of user experience.