摘要
传统PageRank算法用于文献排序时主要关注引证关系,新文献被阅读的概率非常低。本文引入CiteRank算法,结合出版时间、下载次数等指标,提出一种多维度混合排序的方法,应用于万方数据搜索引擎,对3亿篇文献进行多指标混合排序。实证分析结果表明,该排序方法提高了新文献和热门文献被访问的概率,排序结果更加契合用户搜索文献的需求。
PageRank algorithm only pays attention to the citation relationship when it is used in literature ranking,and the probability of new literatures being read is very low.This paper conducts research from the perspective of multi-dimensional sorting,introduces the CiteRank algorithm,uses data such as publication year,download counts,etc.,to perform multi-index mixed sorting on 300 million literatures.The results show that the new sorting method improves the probability of new literatures and popular literatures being visited,and the search result is more in line with users’needs for finding literatures.
作者
张勇
杨赛军
黄华
ZHANG Yong;YANG Saijun;HUANG Hua(Institute of Scientific and Technical Information of China,Beijing 100038;Wanfang Data Company limited,Beijing 100038)
出处
《中国科技资源导刊》
2021年第4期32-37,共6页
China Science & Technology Resources Review
