Research On Potential Trust Relationship Based On Social Commerce Network

The development of Internet technology, the growing popularity and penetration of social applications, provides an opportunity for new business models of e-commerce, which is social commerce. Social commerce involves using Web 2.0 social media technology to support online interactions and user contributions to help purchase products and services. The development of socialized business can reduce the risk of online transactions and tap the potential of consumer demand, which provides a great value for the development of e-commerce. But it also brings a new problem - how to effectively obtain and identify reliable information sources in the network composed of asymmetric information,redundant information and non-acquaintances.The prediction of the potential trust relationship in the social commerce network can solve this problem effectively. The problem of trust relationship prediction is actually a special link prediction problem,which aims to predict the possibility of links between nodes in complex networks through various known network information (network node information, network topology information, etc.).In this paper, we first define and analyze the network topology of the complex network. And then combine the characteristics of the network structure, learn from the relevant trust research, start from the sociology and psychology theory, the mechanism of trust between people and influence. The factors that affect the trust relationship among users in the social commerce network are extracted, and the potential trust relationship model based on the social commerce network is established.The model of the trust relationship between the users is established,which includes the familiarity degree, the preference similarity degree,the reputation degree and the professional ability. Finally, the classical algorithm is used to provide the concrete algorithm for the quantitative measurement of each factor and trust degree in the model. In the process of proposing the algorithm, the calculation of the untrusted factor is introduced into the measure of the reputation degree of the trustee creatively.This paper obtains the user data of the real socialized commerce website, Epinions website. Based on the dataset, we carry out a series of experiments to train and evaluate the models and algorithms proposed in the study. The experimental results show that the method proposed in this study has a better performance in terms of the validity and accuracy of trust relationship prediction. The results show that: (1) the familiarity is sparse in the social electricity supplier network, but the change of the familiarity has the greatest influence on the change of the trust value; (2)the improvement of the professional ability of the trusted users And the similarity between the two trust subjects has a greater impact on the improvement of trust value;(3) the degree of reputation has a certain impact on the promotion of trust value, but the effect is limited.