Research On The Construction Of Fine-grained User Portraits For Online Games Based On User Behavior

In today's increasingly rapid development of the game industry,more and more Internet users choose online games as an important part of Internet entertainment life.Not only that,people's original entertainment environment is gradually shifting from the real space to the network space.Compared with the openness and uncertainty of the real space,when people use specific software or applications for online game activities,their own needs,behavioral goals and behavioral patterns are all restricted and defined,which provides a basic environment for the research on the construction of user portraits based on user behavior.The concept of user portrait has been widely used in e-commerce,statistics,search,social networking and other fields,but there are few related research contents at home and abroad in the game industry.Although the data analysis work in the game industry has made some progress in the research of user portraits in recent years,with the further enrichment of game content in the network,the scale of information resources generated is also increasing.Only a few large companies can be equipped with specialized data analysts to organize and analyze large amounts of data.This paper aims to achieve dynamic adjustment of user portraits from multiple perspectives based on user behavior pattern analysis to meet the needs of data analysis in many small companies,and to adapt to evolvement current of the game industry.The problem is researched and summarized as follows:(1)For a large number of original behavior logs in the game,establish a complete data engineering to accomplish the preliminary cleaning and analysis of the data,and make preparation for the subsequent construction of user portraits.(2)A set of design methods for user portraits are proposed,especially in the industry's research on user category division,label selection and evaluation mechanism establishment.(3)Three methods based on original log information,rule-based and K-Means clustering algorithm are used to generate labels and construct user portraits.(4)Build a user portrait management service system based on Spring Boot and Vue.js to accomplish data visualization analysis of user portraits and related composition data for relevant personnel.