PCA-based dimensionality reduction method for user information in Universal Network

Abstract

Universal Network (UN) is one kind of future Internet architecture. The collection and analysis of user information is a core in the management system of UN. However, users' high-dimensional data affects the performance greatly because it brings in a long response delay when matching user information with strategy rules. An efficient dimensionality reduction method is important to improve the matching efficiency on high-dimensional data. This paper introduces a statistic computational method based on Principal Component Analysis (PCA) for the reduction of user information. The method converts multiple indicators into fewer overall indicators by taking the advantage of the relations among attributes. Then, we apply this algorithm in the user information management system of UN and make several experiments to evaluate and analyze its performance. Experimental results show that the time of querying and matching is reduced by the proposed method on the condition of not losing much information of original attributes. It proves that this method reduces the dimension effectively and can be applied in the high-dimensionality user information management system.