Design And Implementation Of Intelligent Advertisement Recognition System Based On Decision Tree

Recently the rapid development of Internet technology and wireless network technology has brought so much convenient to people's life. With the intelligent terminator which can be easily got, people can get Internet access in no time. As a new type of media, the role of web advertisement is becoming more important due to the advantages on itself and the help of Internet. However, with the cyber security problem being really serious, the negative part of web advertisement has appeared, as a result, there is such an important need to design and implement an intelligent advertisement recognition system for protecting Internet users'interests.To solve the problem just mentioned, this paper has content as follows:(1)We focus on the problem we presented and research the publishing means of internet advertisement. There is a relationship called associated between normal webs and advertisements. Due to the characteristics of this relationship appeared in the timestamp information and other information in user packets, we present a scheme which applied classification algorithm and used collected user packets as data source. Finally we convert the advertisement recognition problem to packet classification problem.(2) We present data attributes selected from initial data by analyzing the requirements on solving packet classification problem, then a detailed research on C4.5 algorithm were taken, as a result, we had a comprehensive of the calculating procedure. Based on the concept of graph theory we presented the method of processing data, meanwhile, by presenting new concept to introduce the new method of processing timestamp attributes. With the character of user data, we presented the new approach to classify user packages.(3)Due to the new classification approach we designed some related experiments, we get the training set by processing the user data, then input the training set into C4.5 algorithm to build the decision tree, at the end of the experiment, we explained the meaning of decision tree and evaluated the efficiency of the tree, the result shows that the accuracy is some kind of satisfying, so with this approach we can achieve the advertisement recognition easily.(4)We have presented the workflow of every module, then implemented Wi-Fi broadcasting module, data collection module, data process module, tree classification and user interact module step by step, after combining this five module into one,the problem mentioned at first is successfully solved.