Research On Consumer Trust Modeling And Application Of Key Opinion Leaders In Social E-commerce

In the context of social e-commerce,key opinion leaders play a crucial role in marketing activities.They will guide and stimulate consumer’s behavior while consumers fully trust them.How to judge the existence of trust relationship and measure the degree of trust between consumer and KOL is the topic of this thesis.In previous studies,graph simplification model is often used to characterize trust in social networks.Due to the characteristics of trust network in social e-commerce,such as large scale,complex structure and miscellaneous information,the modeling of trust is full of challenges.In this thesis,how to simplify the trust network and what information should be retained are the focus and difficulty.After fully considering the comprehensive,implicit,subjective and other characteristics of trust in the context of social e-commerce,this thesis established a personalized comprehensive trust model of consumers in key opinion leaders.Different people have different views on trust because trust is subjective,and consumers’ personal preferences will affect the final trust evaluation.The personalized consumer trust preference was defined in this thesis and was used to measure consumer direct trust to key opinion leaders.Different from previous studies that require consumers to customize their trust preferences,this thesis extracts implicit trust preferences from consumers’ historical trust data,which can avoid the problem of consumers’ unclear understanding of their own preferences.Besides,consumers’ trust in their neighbors in the network will also have individualized changes in different fields.This thesis defines the personalized user product type preference and the trust propagation ability of users in the trust network,which are used to select strong trust neighbors when simplifying the trust network to calculate consumers’ indirect trust in key opinion leaders.Finally,the model integrates consumers’ direct and indirect trust in key opinion leaders to obtain comprehensive trust evaluation.The trust model proposed in this thesis has been verified by standard data sets and has good performance in MAE,accuracy,recall rate,F-Score and other aspects,indicating that the model has high validity.The trust model proposed in this thesis has been applied in the social data of Xiaohongshu APP,a social e-commerce platform.In different product classification situations and different user classification situations,the model still has good performance in MAE,accuracy,recall rate,F-score and other aspects,indicating that it has high application value.In addition,this thesis classifies consumers and explores the influence of key opinion leader fans and recommended product prices on consumer trust for consumers with different trust tendencies.The results show that consumers with strong trust tendencies more trust key opinion leaders with a large number of fans and low product prices;the number of fans of key opinion leaders has no significant effect on trust,while product prices has a negative and significant effect on trust for consumers with weak trust tendencies.This study improves the existing graph simplified trust model and complements the relevant research on consumer trust modeling of key opinion leaders in social e-commerce.In the empirical study part,this thesis puts forward management enlightenment based on model application in different situations,so that the platform can know different types of key opinion leaders of consumer trust according to it’s needs.