Recommended System Research And Application For Mobile Internet Marketing

Internet advertising is a through online advertising platform, it use the website advertising banners, text links, etc. to deliver advertising information on the web. With the rapid increase of mobile phone, mobile marketing is become popular. The low cost of cover a very wide audience on mobile Internet advertising, make it becomes the new tool for enterprises to increase users. There are large-scale of mobile network user groups, which push message is not limited to time, place, and information delivery quickly, inexpensively on it. Mobile Internet advertising has a huge domestic market potential and room for growth, is a major field of advertising new opportunities and challenges.With the development of information technology and the Internet, people gradually from a lack of information to the era of information overload. It became more difficult for information consumers find the information they are interested in from the mass of information; For information producers, it is also difficult to make the information produced by themselves be extracted from the mass of information. The search engine allows users to find the information they need by key word, but when a user cannot express what he or she wanted by keyword or his(her) demand is not clear, the search engine appears stretched. Recommendation system is an effective tool to solve this situation, it is different from search engines that it does not require users to provide a clear demand, it take the user’s past behavior to train a model for user’s interest, and then recommend some information to the user that one may interest. Now, Recommendation System is widely used in a lot of fields. In academia, Recommendation Systems are also very popular.In this paper, we based on a large scale mobile network user behavior data; try to combines it with Recommendation System for mobile advertising. Compared with the electricity supplier data or television program data, mobile network data is more accurately, it can be precisely targeted to a specific person in the real world. In this paper, we combined the data that mobile phone users access App with some ideas of Recommendation System, tried analysis and experiments as follows:1) Find some algorithm and parameters which is more suitable for mobile network advertising and recommendation;2) looking for a data partitioning Approach which not only make lower computational complexity but ensuring the effect of advertising delivery; When training the model parameters, it is feasible for a small amount of data, but when the amount of data is large, the total amount of data will lead to large computation load, but also the whole data may not necessarily enhance the performance of the algorithm. After the data which participating the training model reaches a certain amount, the same algorithm cannot dig deeper data on user’s interest.3) find some relationship between recommendation algorithm and the amount of training data, to find out how much data make the algorithm effect "saturation", so that we can adjusting the amount of data and algorithm parameters in engineering Applications, reduce the computational complexity.