The software system running on the server cluster needs large-scale data sets of Web log to meet the performance test requirement, but the existing simulation generation algorithm cannot meet the requirements due to the single model. Aiming at this problem, a new multi-fractal Web log simulation generation algorithm based on alpha stable process was proposed. Firstly, the self-similarity of Web log was described by alpha stable process in Long Range Dependence (LRD). Secondly, the multi-fractal of Web log was described by binomial- b model in Short Range Dependence (SRD). Finally, the model of long range dependence and the model of short range dependence were integrated into the improved ON/OFF framework. Compared with the single model, the parameters of the proposed algorithm has clear physical meaning equipped with good performance of self-similarity and multi-fractal. The experimental results show that the proposed algorithm can accurately simulate the real Web log and be effectively applied in Web log simulation generation with large-scale data sets.

When the existing simulation generation algorithm uses the distribution of the static model to generate a Web log, there is a big difference with real data. In order to solve the problem, a new algorithm of Web Log Simulation Generation based on user interest migration (WLSG) was proposed. Firstly, the relationship between Web log and time was modeled. Secondly, the migration of user interest was simulated when the user accessed to the file in different time. Finally, it was also simulated that the user adaptively access to the file which he was most interested in at the current moment. Compared with the distribution of the existing static model, the proposed algorithm had significantly improved the self-similarity by about 2.86% on average. The experimental results show that, the proposed algorithm can well simulate Web log by user interest in migration to change user access sequence, which is capable of being effectively applied in the Web log simulation generation.

Concerning the poor generalization ability (the recommended performance for new users and items) and low predictive accuracy of Probabilistic Matrix Factorization (PMF) in recommendation system, a new algorithm of Probabilistic Matrix Factorization algorithm based on AdaBoost (AdaBoostPMF) was proposed. Firstly, the initial weight for each sample was assigned. Secondly, the feature vectors of users and items were learned by each round of PMF stochastic gradient descent method and the global mean and standard deviation of the prediction error were calculated. The sample weights were adaptively adjusted by using AdaBoost from the a global perspective, which made the proposed algorithm pay more attention to training those samples with the larger prediction error than others. Finally, the sample weights were assigned to predictive error, which found the more appropriate optimum direction for feature vectors of users and items. Compared with traditional PMF algorithm, the proposed AdaBoostPMF algorithm could significantly improve the prediction precision by about 2.5% on average. The experimental results show that, the proposed algorithm can better fit the user feature vector and the item feature vector and improve the prediction accuracy by weighting the samples with larger prediction error.The proposed algorithm can be effectively applied to the personalized recommendation.