[目的/意义]针对阻碍高校智慧图书馆对新读者进行图书个性化推荐的用户冷启动问题,提出一种面向新读者的高校图书馆个性化推荐方法,为智慧型高校图书馆开展图书个性化推荐服务、提高新读者借阅率提供切实可行的方案。[方法/过程]以北华大学图书馆借阅流通大数据进行数据挖掘,得出属性相似的新读者和已有读者具有相似借阅偏好的结论。然后,通过奇异值分解解决数据稀疏问题,采用基于欧氏距离的蚁群算法对新读者与已有读者聚类,搭建了新读者和已有读者之间关系的桥梁。最后将已有读者借阅的图书采取Top-N算法对新读者推荐。[结果/结论]以2017级读者为实验对象,选取了3个学院的44名读者,用所提出的算法进行了实验检验。实验结果表明新算法推荐效果显著,操作简单可行,为后续个性化推荐工作奠定了基础。
Abstract
[Purpose/Significance]Aimed at the user cold start problem that prevent the university smart library from accurately recommending books to new readers, a personalized recommendation method for new college readers is proposed to provide practical solutions for carrying out personalized recommendation service and increasing new readers' borrowing rate.[Method/Process]Through data mining on borrowing circulating big data in Beihua university library, the conclusion is reached that new readers and existing readers with similar attributes have similar borrowing and reading preferences;Then, the singular value decomposition is used to solve the data sparse problem and the Euclidean distance and ant colony algorithm are used to cluster the new readers and existing readers, which building a bridge between new readers and existing readers. Finally, the Top-N algorithm is adopted to recommend the books borrowed by existing readers to new readers. [Result/Conclusion]Take the readers from grade 2017 as experiment subject, the proposed algorithm was tested on 44 readers from three academies. The results show that the proposed algorithm is effective and easy to operate, which lays a foundation for the subsequent personalized recommendation work.
关键词
新读者 /
个性化推荐 /
用户冷启动 /
数据稀疏 /
聚类
{{custom_keyword}} /
Key words
new readers /
personalized recommendation /
user cold start /
data sparsity /
clustering
{{custom_keyword}} /
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 中华人民共和国中央人民政府国务院.2014年政府工作报告[EB/OL].[2019-03-20]. http://www.gov.cn/guowuyuan/2014-03/05/content_2629550.htm.
[2] 中华人民共和国中央人民政府国务院.2015年政府工作报告[EB/OL].[2019-03-20]. http://www.gov.cn/guowuyuan/2015-03/16/content_2835101.htm.
[3] 中华人民共和国中央人民政府国务院.2016年政府工作报告[EB/OL].[2019-03-20]. http://www.gov.cn/guowuyuan/2016-03/05/content_5049372.htm.
[4] 中华人民共和国中央人民政府国务院.2017年政府工作报告[EB/OL].[2019-03-20]. http://www.gov.cn/premier/2017-03/16/content_5177940.htm.
[5] 中华人民共和国中央人民政府国务院.2018年政府工作报告[EB/OL].[2019-03-20]. http://www.gov.cn/zhuanti/2018lh/2018zfgzbg/zfgzbg.htm.
[6] 于良芝,于斌斌.图书馆阅读推广——循证图书馆学(EBL)的典型领域[J].国家图书馆学刊,2014,96(6):9-16.
[7] 王波. 阅读推广、图书馆阅读推广的定义——兼论如何认识和学习图书馆时尚阅读推广案例[J].图书馆论坛,2015,10:1-7.
[8] 许天才,杨新涯,王宁,魏群义.图书馆阅读推广的多元趋势研究——以首届高校图书馆阅读推广大赛为案例[J].图书情报工作,2016,60(2):82-86.
[9] 黄运红,于静.论高校图书馆阅读推广服务的创新——以第二届全国高校图书馆服务创新案例大赛为例[J].图书馆论坛,2018,10:1-7.
[10] JIA R F, JIN M Z, LIU C.A New Clustering Method For Collaborative Filtering[C], 2010 International Conference on Networking and Information Technology,2010,11.
[11] YANG Z H, XU L, CAI Z M.Re-scale AdaBoost for Attack Detection in Collaborative Filtering Recommender Systems[J]. Knowledge-Based Systems,2016,100:74-88.
[12] WANG F H, SHAO H M.Effective personalized recommendation based on time-framed navigation clustering and association mining[J]. Expert Systems with Applications,2004,27(3):365-377.
[13] 鲍玉斌,王大玲,于戈.关联规则和聚类分析在个性化推荐中的应用[J].东北大学学报(自然科学版),2003,24(12):1149-1152.
[14] ZHOU T, REN J, Medo M, etc. Bipartite network projection and personal recommendation[J]. Physical review. E, Statistical, nonlinear, and soft matter physic,2007,76(2):046115.
[15] 张新猛,蒋盛益.基于加权二部图的个性化推荐算法[J].计算机应用,2012,32(3):654-657,678.
[16] CHEN L J, ZHANG Z K, LIU J H, etc.A vertex similarity index for better personalized recommendation[J]. Physica A: Statistical Mechanics and its Applications,2017,466:607-615.
[17] 李克潮,黎晓.个性化图书推荐研究[J].图书馆学研究,2011,20(18):65-69.
[18] 凌霄娥,周兵,李克潮.面向新读者和新图书的数字图书馆个性推荐冷启动问题研究[J].情报理论与实践,2014,38(8):99-104.
[19] 刘庆麟. 基于小数据的图书馆精准服务研究[J].图书馆工作与研究,2017,5,45-50.
[20] 王祥德,雷玉霞,闫昱姝.基于矩阵填充的SVD协同过滤算法研究[J].微型机与应用,2017,36(19):55-61.
[21] 赵烨,黄泽君.蚁群K-medoids融合的聚类算法[J].电子测量与仪器学报,2012,26(9):800-804.
{{custom_fnGroup.title_cn}}
脚注
{{custom_fn.content}}
PDF(1985 KB)
