Research On Network Targeted Advertising Based On Relevancy

The rapid development of the internet has led to the rapid expansion of the market,at the same time,the advertising market is rapidly expanding.The internet age of advertising should not only have the basic characteristics of traditional advertising,such as strong communication effect and dissemination,but also reflect the intelligent characteristics of big data age showed by data mining and knowledge discovery,such as precise push and accurate marketing.Advertising has a direct impact on the effectiveness of the product marketing,so how to be more accurate advertising is currently in urgent need of study.Present network targeted advertising can be divided into two categories,one is based on user behavior,and the other is based on user content,both enhance the accuracy delivery of network targeted advertising in a certain extent.But there are some problems:(1)network targeted advertising based on user behavior takes user behavior as a rule,but does not the impact of the user content and ad text on advertising into account,and also lacks of user multi-dimensional characterization,the accuracy is poor.(2)the dimensions of network targeted advertising based on user content are limited to the user content,and lacks of the reveal and preference analysis of the characteristics of user dynamic behavior,and the accuracy is even lower than the user behavior.Therefore,using the existing network targeted advertising,whether from the delivery object,or from the delivery time and other aspects are difficult to achieve the advertisers' psychological expectations,more difficult to capture the user heart.In this paper,we take the above problems as a research breakthrough.First,for the multi-dimensional portrait of the user,we build a Personas model based on K-means clustering algorithm.Personas model as an effective tool that outline target user and lock user needs,it links up user's various properties and behaviors to describe the user,so it can multi-dimensional describe the user characteristics accurately.User feature also divided into static user feature and dynamic user feature,it improves user characteristic dimension and provides the basis for the user feature vector.At the same time,the analytic hierarchy process is introduced to calculate the user's feature weight,which makes the weight of the feature vector more accurate.Secondly,for the text advertising problem,we also introduce the ad description dimensions,and characterization of advertising theme from multiple dimensions to ensure the accuracy of ad description.Besides,we improve the traditional TF-IDF algorithm for semantic feature,more accurately calculate ads feature weights.Finally,to ensure accuracy of user and advertise eigenvectors after weight accuracy,we use the vector space model's relevancy algorithm to match ads and users,and take the relevancy as arule of advertising accurate delivery.In the empirical study,this paper take Weibo users as empirical objects.We propose a Personas model to portrait Weibo users,while also use the improved TF-IDF algorithm to calculate the feature vector weight.After getting the feature vector of users and advertising,the vector space model is used to calculate the relevancy between users and advertising.The research shows that the method proposed in this paper can improve the relevancy of users and advertising,and improve marketing efficiency.