Evaluation research for Internet users/services attributes extraction and classification

Abstract

Internet has been changing communications and social network of human beings during the past years due to its convenience. The number of Internet users and services are increasing incredibly with the development of various networking technologies, but at the same time these cause the difficulty of managing the network and providing better user experience. User/service classification can alleviate this problem. A thorough evaluation system can not only assess the performance of the classification models, but also improve them significantly. In this paper, we build a new kind of user/service classification model and propose an evaluation system. Taking diverse viewpoints into account and ensuring the evaluating results to be as complete and unbiased as possible, we design a multi-layer and multi-dimensionality assessment architecture. This architecture includes reliability, validity, flexibility, efficiency and cost. We believe our evaluation system could behave better and come up with a more overall and scientific consequence when it comes to the specific classification model.