The Research And Implementation On Socialized Marketing Via Social Network Analysis

With the rapid development of the Internet and social networks,people have more access to information and the traditional offline social interaction has gradually transformed into an online social interaction.More frequent social events generate a wealth of social information,which diffuse into different groups of users explosively with social networking activities,which change the way users behave.At the same time,these changes have greatly facilitated the development of social network analytics.At present,social network analysis technology has been successfully applied to many real-life scenarios,such as online friends-recommendation,products recommendation of online shopping,content push services of news application based on the user's social relations,and other fields.An important issue to apply to social networking analytics in these production and life styles is how to efficiently integrate the network topology structures into these applications.In order to better retain and use the rich information contained in the network topology structures,this dissertation focuses on the research of a network embedding learning algorithm based on deep learning technology.In this dissertation,we study the problem of embedding representation of network structure information first,then give a formalized definition of the problem.Afterwards,in order to demonstrate different types of network structures in network embedding representation learning task,we research and propose an algorithm ANE that uses Adversarial Autoencoders to learn the network embedding representations in this paper.It uses the objective functions to adapt to the different network structure to learn the latent code of the network topology Structure,and using the adversarial training to alleviate manifold fracturing problems caused by the deep Auto-encoder.In order to validate the effictiveness of ANE algorithm,this paper has carried on the sufficient experiment on three actual datasets,the experiment indicated that ANE algorithm has good performance in the network embedding representation learning task and the different network analysis application.At present,most of the social network analysis is based on the social network formed by online social applications.This dissertation studies the bank transaction data under the financial scenario,selects the transaction data of bank customer to construct the simulated social network,and designs the socialized accurate marketing system based on social network analysis,and realized the user profile visual prototype subsystem.The use of the prototype subsystem can more intuitively show attributes and social relations of users and community,effectively improve the pertinence of bank marketing activities.