计算机工程

• 人工智能及识别技术 • 上一篇    下一篇

基于兴趣衰减的个性化排序算法

王林,刘继源,马安进   

  1. (西安理工大学 自动化与信息工程学院,西安 710048)
  • 收稿日期:2016-08-03 出版日期:2017-09-15 发布日期:2017-09-15
  • 作者简介:王林(1963—),男,教授、博士,主研方向为社交网络、数据挖掘;刘继源,硕士研究生;马安进,硕士。

Personalization Sorting Algorithm Based on Interest Attenuation

WANG Lin,LIU Jiyuan,MA Anjin   

  1. (School of Automation and Information Engineering,Xi’an University of Technology,Xi’an 710048,China)
  • Received:2016-08-03 Online:2017-09-15 Published:2017-09-15

摘要: 目前多数个性化排序算法未考虑用户兴趣随时间产生的漂移变化,从而影响排序质量。为此,提出一种融合用户兴趣衰减的个性化排序算法。利用传统个性化排序算法的用户兴趣模型,及用户搜索兴趣的变化规律,分析搜索兴趣程度的时间衰减性,以人类遗忘曲线为基础给出适合搜索兴趣变化的指数遗忘函数,并将其运用到传统个性化排序算法中。实验结果表明,与基于兴趣模型的个性化排序算法相比,该算法能提高个性化搜索引擎的查准率。

关键词: 搜索引擎, 个性化排序, 搜索兴趣, 兴趣漂移, 遗忘曲线

Abstract: The most currentranking algorithm don’t take users’ interest drifting over time into consideration,which affects the sorting quality.In order to solve this problem,a method ranking algorithm merging user interest attenuation is proposed.In this method,users’ interest model of traditional personalization ranking algorithm and changing law of users’ searching interests are used to analyze the time attenuation of search interest.The exponential forgetting function fitting searching interest based on human forgetting curve is proposed and applied to the personalized ranking algorithm.Experimental results show that the new algorithm can improve the precision of personalized search engine compared with the ranking algorithm based on users’ interest model.

Key words: search engine, personalization sorting, search interest, interest drifting, forgetting curve

中图分类号: