Study on Social Tag Relevance Judgment Based on Relative Frequency

摘要：

　　Tag relevance judgment is the foundation of tag application research and practice.Currently, tag relevance judgment strategies meet some problems in experiment effect and universality.To ameliorate both the effect and generalization of social tag relevance judgment strategies, this paper proposes a tag relevance judgment algorithm based on relative frequency.We verify this strategy by experiment based on 675 351 Douban Movie users social tagging data.The results show that the effect of tag relevance judgment for tags whose frequencies are at least 5 performs well, for the recall is 84.16% and precision is 90.77%.We compare this strategy to the Top-N social tag relevance judgment strategy, and the result is obviously better than the Top-N strategy.Moreover, this strategy has good universality.It is suitable for various media such as video, audio, txt, image etc.Thus, it can be concluded that this strategy can solve tag relevance judgment problem well.

关键词： social tagging tag relevance relative frequency

作者: SHI Yu ZHA Mengjuan LIN Xin

作者单位: School of Information Management, Wuhan University, Wuhan 430072, China School of Information Management, Central China Normal University, Wuhan 430079, China

会议类型: 国际会议

会议名称: 第二届信息获取与知识服务国际会议

会议地点: 武汉

会议语种:英文

页码: 285-289

在线出版日期: 2016-10-21（万方平台首次上网日期，不代表论文的发表时间）