| With the development of Internet rapidly, advertising based on Internet has become the hottest form of advertising now. The mostly revenue for majority web sites is dependent on the advertisings based on Internet. For web site administrators, the optimally scheduling advertising resource becomes a primary problem. Therefore, it has an essential theoretical and practical meaning for researching on advertising control and decision problem based on Internet.In this paper, revenue on web site and space scheduling of advertisings on web page are studied for Internet advertising decision problem. The research contents include mainly several aspects as follows:1. This paper summarizes the background and the investigative statues at home and abroad of Internet advertising decision problem. At the same time, the paper points out the shortages of the problem research at the present time.2. The decision problem about revenue maximization on web site is studied in thepaper. Under the circumstances that the benefits of both web sites and advertisers are balanced, the linear programming model based on hybrid pricing for maximizing revenue on web site is proposed in this paper. GA(Genetic algorithm) for sloving the above model is presented. In comparison with Simplex method, simulation result shows that the weakness of over-targeted solutions aroused by Simplex method is overcome by GA.3. Considering heterogeneity in consumer click proneness can't be explained rightly by average click-through rate in previous research, the revenue maximization on web site decision model based on consumer click proneness is proposed in the paper. In the process of sloving the model, first the non-linear relation between impressions and click-through rate about advertisng is fitted by BP neural network for forecasting click-through rate, then sloving the model by GA. Simulation result shows that the effectiveness of this model and algorithm.4. For making better use of limited advertising space, minimizing space scheduling problem about advertising based on Internet is extended to one cluster web pages. The heuristic algorithm of LHD(Least Height Decreasing) and FFHD(First Fit Height Decreasing) are proposed for sloving the proposed model, then GA-FFHD is brought... |