利用ε-贪婪学习和用户行为反馈的搜索引擎网页排序算法  

Page ranking algorithm based on ε-greedy and user behavior in search engine

作　　者：张春玲[1] 姜成晶 Zhang Chunling;Kang Sun-kyoung(School of Computer & Software,Weifang University of Science & Technology,Shouguang Shandong 262700,China;College of Computer Software Engineering,Wonkwang University,Iksan 54538,South Korea)

机构地区：[1]潍坊科技学院计算机软件学院,山东寿光262700 [2]韩国圆光大学计算机软件工程学院,韩国益山54538

出　　处：《计算机应用研究》2019年第8期2300-2304,共5页Application Research of Computers

基　　金：国家自然科学基金资助项目(61402067)

摘　　要：为了提高网页排序的准确性,提出一种基于ε-贪婪学习和用户点击行为的网页排序算法。首先,根据用户查询,通过轮盘赌策略向用户推荐相关网页列表;然后,根据用户点击网页的行为进行ε-贪婪学习,计算得到排序系统中的强化信号,通过奖励和惩罚机制为每个网页计算相关性程度值;最后,根据相关性程度对网页进行重新排序。随着用户反馈的信息越来越多,相关网页会排列在列表的最高等级上。实验结果表明,提出的算法能够准确地推荐出相关网页,在P@n、NDCG和MAP性能指标上都获得了较优的性能。In order to improve the accuracy of Web page ranking,this paper proposed a Web page ranking method based on ε-greedy learning and user click behavior.Firstly,it recommended to the user a list of related Web pages by the roulette strategy according to the user query.Then, it performed ε-greedy learning based on the behavior of the user clicking on the Web page, and calculated the fortified signal in the ranking system.After that,it calculated the relevancy degree value for each Web page through reward and punishment mechanisms.Finally,it reordered the network according to the degree of relevance. As more and more information were fed back by users, related Web pages would be ranked at the highest level of the list. The experimental results show that the proposed method can accurately recommend the relevant Web pages and obtain better performance on P@n , NDCG and MAP performance indexes.

关 键 词：搜索引擎 网页排序 ε-贪婪学习 用户行为 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]