USER CLUSTERING METHOD USING HSOM FOR RECOMMENDING WEB INFORMATION ON INTERNET AND RECORDING MEDIUM

摘要

PURPOSE: A user clustering method using the HSOM(Hybrid Self-Organizing feature Maps) for recommending web information on the Internet and a recording medium are provided to overcome a modeling limit not solved in the SOM by suggesting the HSOM guaranteeing convergence for a clustering result while keeping a fast learning capability. CONSTITUTION: Input vector is initialized(S300). Weights used for a node of each feature map are generated from an initial Gaussian distribution(S310). A winning node of the HSOM is determined, a maximum range of given iterative learning is fixed, and the final weight determines the node of the feature map to be updated(S320).