Research On Demand Estimation Of E-Commerce Platform Considering Social Network

The mobile internet traffic dividend period has passed,and the cost of new users is high.For e-commerce platforms,accurate estimation of user demand to provide personalized products is significant in improving user stickiness and transaction conversion rate.At the same time,with the development of new e-commerce models such as social e-commerce,social networks are exerting more and more significant influence on the purchasing behavior of users.However,existing demand estimation methods often ignore the value of social information.Therefore,this study focuses on the demand estimation method of e-commerce platforms,considering user heterogeneity and social networks.Specifically,this article uses structural equation modeling to identify the key factors influencing user choice behavior in e-commerce platforms and the potential relationships among the key factors,especially the role of social networks.It provides the reference and control basis for user stratification,feature selection,and parameter estimation results.Then,this article takes the balance and flexible adjustment between accuracy and diversity as the goals of the design of the demand estimation method.Firstly,describes the influence of social networks by fusing the random walk algorithm and the heat conduction walk algorithm;Secondly,reconstructs the MMNL model using the neural network structure and embeds it into the Wide & Deep model.It improves the interpretability of the model under the premise of guaranteeing the memory and generalization ability.Finally,based on ensemble learning and user layering,this article establishes a weighted aggregation model to design a joint demand estimation method and build the theoretical framework of the user demand estimation method for e-commerce platforms by flexibly adjusting the proportion of diversity factors and the weight of social networks for different active user groups.After the experiment,firstly,the study finds that considering the social network can improve the effectiveness of demand estimation to some extent.Secondly,the study shows that social networks are more effective and can play a more critical role in estimation for high-active users with rich social information than low-active users.Thirdly,the study shows that introducing the MMNL model can improve the interpretability of the model and enhance the estimation effect.Finally,the result confirms that the proposed estimation method significantly improves the effect compared with the current mainstream algorithm.It can analyze the influence of related attribute characteristics on user behavior and assist platform operation decision-making with good theoretical significance and application value.