Network User Profiling Based on Heterogeneous Attribute Propagation
[Purpose/significance]To address the issue of single-dimensional user profiling based on online reviews,a meth-od for network user profiling based on heterogeneous attribute propagation is proposed.[Method/process]A graph model is con-structed based on users,movies,and tags.User attributes are initialized from multiple dimensions,including basic attributes,movie preferences,emotional preferences,and rating behaviors,serving as user node attributes.These user attributes are then con-tinuously updated through iterative propagation.[Result/conclusion]Experimental results show that the proposed method can sig-nificantly enrich the dimensions of user profiling.Compared to the current best deep learning model,the mean squared error(MSE)is reduced from 0.113 to 0.083.Through attribute augmentation and propagation,this method can provide rich and accurate user profiling capabilities.[Limitations]The experimental data is sourced from movie reviews,and the user profiles are based on movie rating users.This scenario is relatively limited and lacks validation in other domains.
万方数据
维普
