신경망을 이용한 SNS상에서의 정보확산 예측모형: Digg 사례를 중심으로

초록

In this paper, we suggest a model based on artificial neural network (ANN) to predict the amount of accumulated information diffusion according to time-flow on social network service (SNS). If the number of output variables is only one and the values of the last two input variables are similar to each other, then the suggested model predicts the value of output variable as the value of last input variable. Otherwise, it predicts the value(s) of output variable(s) by means of ANN. The suggested model is applied to real data in Digg, which is a social bookmarking site, and, as a result, the suggested model in this research outperforms, in terms of prediction accuracy, ARIMA model which is a typical time-series analysis model.

키워드