Based on Logistic Regression the online Reviews False Identification Model Study

Abstract

Abstract. In this paper, a questionnaire survey was made on the key factors to identify the false comments from the perspective of consumer perception, including words, number, comment tone, detailed degree, three days before the comments number, first week comments, total comments number, 2 weeks before comments number, additional comments number, photo show number, good review, medium comments or bad comments, sellers reputation, from 0 to 9 o'clock and 21 measurement items, in the case of taobao commodity comments, artificial traversal 60 kinds of commodities in 6885 comments to collect data, it is concluded that additional comments number, slide show number, and medium review number of significant differences when consumers perceive false comments. By constructing the Logistic regression model, detailed degree, three days before the comments number and slide show number are the key influencing factors, overall percentage prediction rate was 85%, the discriminant function effect is better. The results show that the three indexes can be used as a consumer in judging comment whether false goods, shopping is the basis of risk.