Studies Of Online Advertising Push Service Based On Big Data

With the rapid development of internet and the arrival of big data era, the disadvantage of traditional online advertising push has become increasingly obvious in recent years for its extensiveness and aimlessness which makes the requirements for accurate online advertising become extremely urgent. Under this background, a new advertisement model named "online directional advertisement based on big data" was born. By taking specific technical measures, this kind of advertisement model can push ads to target user directionally. Meantime, the online advertising push service with directional advertisement at the core has sprung up. Notwithstanding the study has already made some achievements, the mismatch between advertisement and target user still exists. So how to improve the match accuracy between them has become an urgent problem to be resolved.One key to accurately matching advertisement with the target user is to analysis the interest point of the user so that the favorable advertisement could be delivered. While data management platform is not only a user data analysis platform but also the core of the whole online advertising service system. Its main function is to analysis user’s interest feature. In view of the above, this article concentrates on behavior orientation advertisement module to research the technologies used for user’s behavior characteristic analysis. It includes extractive technique for webpage feature word, aggregation method of similar webpage, analytic algorithm of user behaviors, and so on. Afterwards a webpage classification method based on redis and the above-mentioned aggregation method of similar webpage has been put forward to establish an optimizing strategy which aims to improve the accuracy for similar webpage aggregation. This optimizing strategy first collects vocabulary and categorical attributes of all hot industries to establish an unified modeling as well as Redis knowledge base. According to the knowledge base and relevant decision algorithm, the industry classification judgment for key words representing webpage characteristic then would be realized. Later, it would integrate all the key words in the webpage to calculate the industry classification through weight value. Finally the calculation result would be combined with the clustering algorithm used in former similar webpage aggregation and the vector space modal to refine the aggregation standard of similar webpage. Through this way, the goal to improve accuracy of similar webpage aggregation would be achieved. The higher the accuracy is, the user interest is correspondingly more accurately orientated, thus the match accuracy between user and ads would be elevated. Then refer to this optimizing strategy, the author integrated the webpage classification judgment based on Redis into the process of user interest and ads matching, thus improved the accuracy between user and ads matching to some extent.Ultimately, this article designed as well as realized the data management platform. Chosen part weblog of Beijing Jiaotong University’s information center as data source, this article took the similar webpage aggregation accuracy&recall rate and advertisement matching accuracy&recall rate as evaluation index. Through comparing with the similar webpage aggregation and advertisement matching results before and after optimizing, it finally turned out that the optimization method raised in this article could improve the similar webpage aggregation accuracy as well as the matching accuracy between advertisement and the target user.