Title: Daily user-online load forecasting for social network site based on the DFT resampling interpolation with periodic extension
Authors: Zong-chang Yang
Addresses: School of Information and Electronical Engineering, Hunan University of Science and Technology, Xiangtan, 411201, China
Abstract: Along with rapid development of the internet, a new type of information networks called 'online social networks' are emerging in popularity, which have been one global phenomenon. Then analysis and forecasting of user-online load movement for one social network site is increasingly important because of its significant effect on web traffic, information propagation, resource allocation, maintenance management and commercial operation. In this study, it is found that the daily online load movement roughly exhibits daily cyclical variations. Then with periodic extension for the daily user-online load movement, a forecast approach based on the called discrete Fourier transform (DFT) resampling interpolation is proposed, in which two-factor up-sampling is performed on the periodically adjusted user-online load observed sequence to predict (estimate) its future movement. Experiments and result analysis indicate potentiality of the proposed method that it yields satisfying results in forecasting daily user-online load movements at the Chinese social network websites.
Keywords: social networking sites; SNS; user-online load movement; periodic extension; discrete Fourier transform; DFT resampling interpolation; forecasting; online social networks; China.
International Journal of Information and Communication Technology, 2017 Vol.10 No.2, pp.162 - 184
Accepted: 11 Oct 2014
Published online: 11 Jan 2017 *