面向互联网上网服务行业的用户画像系统设计  

User Portrait System Design for Industry of Internet Access Service

作　　者：杨欧亚 龚婕[2] 魏松杰 YANG Ouya;GONG Jie;WEI Songjie(School of Computer Science and Engineering,Nanjing University of Science and Technology,Nanjing 210094;Department of Automation,Tsinghua University,Beijing 100084)

机构地区：[1]南京理工大学计算机科学与工程学院,南京210094 [2]清华大学自动化系,北京100084

出　　处：《计算机与数字工程》2021年第9期1782-1787,共6页Computer & Digital Engineering

摘　　要：互联网上网服务行业用户画像系统通过对用户行为数据进行分析,利用网络爬虫、数据挖掘以及文本分类等技术,对用户进行画像,刻画了一个完整的用户全貌。系统对不同属性的数据采用不同的方法,挖掘出数据中的有用信息,提取用户的平台特征,为用户标记不同权重的标签,使得企业更加了解用户,为企业发展提供指导,同时也为用户提供个性化服务铺平道路。此外,针对K-means算法进行改进,从测试结果可以看出,经过改进的K-means准确率和稳定性都得到了极大的提升。The user portrait system analyzes user behavior data,and uses technologies such as web crawler,data mining and text classification to depict user portrait.The system adopts appropriate methods for data with different attributes,useful information in the data is mined,platform characteristics of users is extracted,and labels with different weights for users are marked,so that enterprises can better understand users,provide guidance for enterprise development,and pave the way of precise recommendation for users.In addition,the K-means algorithm is improved.It can be seen from the test results that the accuracy and stability of the improved K-means have been greatly improved.

关 键 词：用户画像系统 用户行为分析 用户价值模型 聚类算法 文本分类 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]