Abstract: 【Purpose/significance】This article constructs a multi group user portrait and resource portrait of a video media platform, re?
vealing the significant characteristics of user behavior in different groups and the correlation between the portraits, providing reference
for optimizing personalized services for video media.【Method/process】Through in-depth analysis of the connotation, tag system, and
algorithm of user portraits, obtain user review data on Douban Film website. Apply k-means algorithm to cluster Douban Film user
groups to obtain user portraits. Construct resource portraits based on data tags. Based on Pearson correlation analysis algorithm, con?
duct correlation analysis on user portraits and resource portraits, and deeply explore the potential relationship between user portraits
and resource portraits, The ultimate goal is to accurately depict user portraits.【Result/conclusion】In the era of big data, user groups
are clustered based on the theory and method of user portraits, exploring the correlation between user portraits and resource portraits, thereby helping video media comprehensively understand user information.【Innovation/limitation】Through research on the correla? tion between user portraits and resource portraits, accurate recommendation of video media resources can be achieved. Subsequent re? search can continue to explore the correlation between user portraits and resource portraits in videos with different content themes.