An Empirical Study Early-warning for Network Quality Safety of Consumer Based on BP Neural Network Information-质量院英文网
Position: Home > Journals > Content

An Empirical Study Early-warning for Network Quality Safety of Consumer Based on BP Neural Network Information

April 28, 2016
Yu Fan, Yu Hongwei, Xu Wei

Abstract: To Cope with  frequently occurred issues of network quality safety,this paper aims to study information earlywarning for network quality safety by using BP neutral network. Data source is all posts about Shengyuan milk powder downloaded from Tianya forum. BP neutral network includes three layers: input layer,hidden layer,and output layer.Eight indicators includes posts per unit,cumulative posts,posting frequency per unit,cumulative posting frequency,reply per unit,cumulative reply,reply frequency per unit,cumulative reply frequency are constructed,so as to five earlywarning levels including safety and mild warning,middle warning,serious warning,very serious warning. The empirical study finds out that taking advantage of BP neutral network has an efficient impact on earlywarning. However,the earlywarning results are lower than expected.And the earlywarning sensitivity is to be strengthened.


Key Words: BP Neutral Network; Network Quality Safety; Information Earlywarning