Application Of Equilibrium Model Based Analysis Of Conflict In The Optimization Of Commodity Recommendation Algorithm

When recommend commodities, because a single kind of recommendation algorithm owns shortcomings, so introducing other algorithms to work with it to overcome these defects. However, introducing additional algorithms will bring defects of new algorithms, and recommending commodities with multiple algorithms will bring conflicts then lead to bad results. Game theory is usually used as basis for multiple objects to make optimal decision in conflict state. Nash equilibrium is the balance point of a game in conflict state. The objects will get the optimal utility value in a game when they are in Nash equilibrium. Therefore, using Nash equilibrium related theory to optimize the commodity recommendation algorithm which was consisted of various algorithms can solve the problems of the recommendation algorithm.In order to solve the problems, use optimization of recommendation algorithm which was generated by combining collaborative filtering with clustering as an example and do the following work. First, according to the basic elements of a game, build related game model by analyzing the procedure of commodity recommendation which works with various algorithms. Second, analyze conflicts between different algorithms to identify the Nash equilibrium in the established game model then establish an equilibrium model. Third, according to the analysis results from the equilibrium model and the workflow of stochastically optimization algorithm, abstract a stochastic optimization algorithm. Forth, use the stochastic optimization algorithm from the equilibrium model to optimize the commodity recommendation algorithm. Fifth, do experiments on large data sets for verifying the actual effects of the optimization algorithm.The results of the experiments show that:First, the recommendation algorithm which was optimized can overcome the defects of recommending by one single algorithm. Second, the optimized recommendation algorithm can coordinate multiple algorithms which consist of it to work efficiently. Third, the optimized recommendation algorithm can get higher quality commodity recommendation results. Therefore, using equilibrium model based conflict analysis to optimize the commodity recommendation algorithm can get good results.