[目的/意义]以现有图书馆借阅记录为基础，结合图书阅读相关性进行深入挖掘，探讨识别借阅场景下图书专业性质量和实现相应个性化图书推荐服务的有效方法。[方法/过程]利用图书的阅读相关性提出图书相关性链接关系，结合图书质量的迭代识别算法来识别专业图书资源。同时利用图书类别相关性链接关系，提出读者用户个性化模式的表达方法，并从长期兴趣推荐和短期兴趣的即时推荐两个方面给出个性化图书推荐策略设计原理和实现方法。[结果/结论]在图书质量识别方面，该方法更易于识别出专业性较强的优质图书资源，适用面比较灵活，也可以在限定图书范围内进行专业图书识别。在个性化图书推荐方面，发现不论长期兴趣推荐方法还是短期兴趣推荐方法，第二类用户的平均推荐命中度要高于第一类用户，在第一类用户中，最高相似度区间（75%以上）和较低相似度区间（15%-50%）的短期兴趣推荐方法的平均推荐命中度要高于长期兴趣推荐方法。本研究通过读者借阅序列分析方法识别专业图书并实现相应的个性化推荐图书方法，有利于改善现有图书馆借阅服务水平和提高读者的满意度。
[Purpose/significance] With the analysis of the existing library records and mining of the reading relevance of books, this paper discusses the effective methods to identify high-quality professional books and implement a personalized recommendation service. [Method/process] This paper introduces the iterative algorithm of recognizing high-quality professional books from links of books relevance based on reading relevance. Then the construction of reader personalized profile is discussed based on the definition of links of book categories. The design and implementation of long-term and short-term personalized recommendation methods are also proposed. [Result/conclusion] In the aspect of book quality identification, it is easier to identify the professional books resources with higher quality. This application is more flexible and also can identify the high-quality professional books within the collection of specific books. It is found that whether long-term or short-term interest interest recommendation method, the average hit degree of users with higher lending is higher than users with lower lending. In the group of users with higher lending, the average hit degree of short-term interest recommendation method in the highest similarity range (more than 75%) and lower similarity range(15% to 50%) is higher than the long-term interest recommendation method.
[1] WU F, HU Y, WANG P. Developing a novel recommender network-based ranking mechanism for library book acquisition[J]. Electronic library, 2017, 35(1):50-68.
[2] OUNI A, KULA R, KESSENTINI M, et al. Search-based software library recommendation using multi-objective optimization[J]. Information &software technology, 2017, 83(3):55-75.
[3] MATHEW P, KURIAKOSE B, HEGDE V. Book recommendation system through content based and collaborative filtering method[C]//Proceedings of international conference on data mining and advanced computing.Ernakulam, India:IEEE, 2016:47-52.
[4] 徐宾. 图书馆图书h指数的研究[J]. 情报学报,2014(8):892-896.
[5] 李克潮,梁正友. 基于多特征的个性化图书推荐算法[J]. 计算机工程,2012,38(11):34-37.
[6] PING H. The research on personalized recommendation algorithm of library based on big data and association rules[J]. Open cybernetics &systemics journal, 2015, 9(1):2554-2558.
[7] 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, 37(8):100-104.
[8] 刘丹. 利用Apache Mahout部署个性化图书推荐服务[J]. 现代图书情报技术,2015,31(10):102-108.
[9] HE B, ZHANG H. Library personalized information recommendation of big data[C]//. Proceedings of online analysis and computing science.Chongqing, China:IEEE, 2016:289-292.
[10] 郑祥云,陈志刚, 黄瑞,等. 基于主题模型的个性化图书推荐算法[J]. 计算机应用,2015,35(9):2569-2573.
[11] 朱文奇. 推荐系统用户相似度计算方法研究[D]. 重庆:重庆大学, 2014.
[12] 马健,杜泽宇,李树青. 基于多兴趣特征分析的图书馆个性化图书推荐方法[J]. 现代图书情报技术,2012,28(6):1-8.
[13] 李克潮, 蓝冬梅, 凌霄娥. 云模型和多特征的高校读者借阅偏好不确定性图书推荐研究[J]. 现代图书情报技术,2013,29(5):54-58.
[14] 景民昌,于迎辉. 基于借阅时间评分的协同图书推荐模型与应用[J]. 图书情报工作,2012,56(3):117-120.
[15] 江周峰,鄂海红,杨俊. 基于时间上下文信息的借阅次数评分模型与应用[J]. 图书情报工作,2014,58(s2):220-223.
[16] 王进良,张鹏,狄增如,等. 北京师范大学图书借阅系统的网络分析[J]. 情报学报,2009,28(1):137-141.
[17] 傅林华,郭建峰,朱建阳. 图书馆图书借阅系统与单标度二元网络模型[J]. 情报学报,2004,23(5):571-575.
[18] 燕飞,张铭,孙韬,等. 基于网络特征的用户图书借阅行为分析——以北京大学图书馆为例[J]. 情报学报,2011,30(8):875-882.
[19] ZHAO S, ZHAO Y, SUN F, et al. Study on single mode weighted network of library lending network[J]. Journal of residuals science & technology, 2016, 13(6):241-246.
[20] 李树青,徐侠,许敏佳. 基于读者借阅二分网络的图书可推荐质量测度方法及个性化图书推荐服务[J]. 中国图书馆学报,2013,39(3):83-95.
[21] 袁虎声,赵洗尘. 基于加权借阅网络的个性化推荐算法与实现[J]. 图书情报工作,2016,60(10):130-134.
[22] 蓝冬梅. 大数据量图书下多数据集的二部图多样化推荐[J]. 情报理论与实践,2016,39(2):69-72.
[23] SUN Y, YIN H, REN X. Recommendation in context-rich environment:an information network analysis approach[C]//Proceedings of the 26th international conference on World Wide Web companion. Perth, Australia:International WWWconferences steering committee, 2017:941-945.
[24] 邱均平,张聪. 高校图书馆馆藏资源协同推荐系统研究[J]. 图书情报工作,2013,57(22):132-137.
[25] 李树青,孙颖. 基于加权关键词共现时间元的个性化学术研究时序路径发现及其可视化呈现方法[J]. 情报学报,2014,33(1):55-67.
[26] PAGE L. The PageRank citation ranking:bringing order to the web[J]. Stanford Digital Libraries Working Paper, 1998, 9(1):1-14. 作者贡献说明:李树青:提出研究思路,进行论文撰写、修改及最终版本修订; 庄光光:进行实验算法设计和相关数据整理分析; 秦嘉杭:负责原始借阅记录信息整理和相关数据处理; 徐侠:负责推荐实验结果评价分析。
