摘要
游戏平台为优化游戏体验常提供决策AI托管功能,这导致平台在对用户进行画像时面临数据质量低的问题。日志中混合了托管AI和玩家自主作出的决策行为,有时还存在行为缺失的情况。提出一种面向决策可托管应用的行为提取与分析方法,通过基于启发式规则的行为特征提取方法从日志中提取玩家自主行为。最后使用多种画像方法在原始数据集和精简数据集上分别进行分析并比较其结果,得出该方法可在大幅度缩减数据量的同时较好地保留玩家的行为特征信息。
In order to optimize the game experience,the game platform often provides decision-making AI hosting functions,which causes the platform to face low data quality when presenting images to users.The log mixes the AI and the decision-making behavior made by the player,and sometimes there is a lack of behavior.This paper proposed a behavior extraction and analysis method for decision-oriented hosting applications.Through the behavior feature extraction method based on heuristic rules,this method extracted the player’s autonomic behavior from the log.Finally,it used a variety of portrait methods to analyze and compare the results on the original data set and the reduced data set respectively.It is concluded that the method can greatly reduce the amount of data while retaining the player’s behavior characteristics.
作者
高铭
李淑琴
孟坤
丁濛
郑蓝舟
Gao Ming;Li Shuqin;Meng Kun;Ding Meng;Zheng Lanzhou(School of Computer Science,Beijing Information Science&Technology University,Beijing 100101,China;Sensing&Computational Intelligence Joint Laboratory,Beijing 100101,China;MicroAI Technology Co.,Ltd.,Beijing 100101,China)
出处
《计算机应用研究》
CSCD
北大核心
2020年第11期3312-3315,3325,共5页
Application Research of Computers
基金
北京信息科技大学2020年度促进内涵发展科研水平提高重点研究培育项目(5212010937)
北京信息科技大学2019年度科技计划一般项目(KM201911232002)
促进高校内涵发展常规项目(5112011019)
促进高校内涵发展—研究生科技创新项目(5121911044)。
关键词
数据整理
用户画像
行为提取
决策可托管
data sorting
user profile
behavior extraction
decision-oriented hosting
