[目的/意义] 为了提高传统协同过滤算法的计算速度，解决目标用户随着时间推移发生兴趣偏移而导致推荐系统质量下降的问题，以期进一步提升推荐系统运行效率和推荐质量。[方法/过程] 提出预先计算用户相似度算法和引入时间评分权重计算相似度矩阵的两种算法的改进，并利用Hadoop平台实证分析改进后的算法。[结果/结论] 实验结果证明：预先计算用户相似度算法缩短了对读者推送相关信息的时间，从而有效地提升了计算速度；引入时间评分权重计算相似度矩阵大大降低了MAE值，从而提高了推荐质量，两种算法同时应用后推荐系统在计算速度、准确率和新颖性方面都有显著提升。
[Purpose/significance]This paper aims to improve the calculating speed of the traditional collaborative filtering algorithm, to solve the problems that the quality of the recommendation system declines because the target users may have the interest shift as time goes on, and to promote the operating efficiency and recommendation quality furtherly.[Method/process]An improved algorithm which calculates the user similarity algorithm in advance and introduces the time score weight to calculate the similarity matrix is proposed, and this improved algorithm is analysed empirically with Hadhoop.[Result/conclusion]The pre-calculation of user similarity algorithm can improve the calculation speed effectively, which would shorten the time when system sends the readers related informations. The introduction of the time score weight to calculate similarity matrix can reduce MAE value which would improve recommendation quality. And the recommendation system has a significant improvement in the calculating speed, accuracy and novelty when the two algorithms are applied simultaneously.
[1] 严旗令.一种改进的协同过滤推荐算法[J].软件导刊,2016,15(4):63-65.
[2] 刘剑涛.个性化推荐系统中用户多态聚类研究[J].现代图书情报技术,2012,28(2):18-22.
[3] 唐秋鸿,曹红兵,唐小新,等.高校图书馆个性化专题推荐研究[J].图书馆学研究,2012(13):53-58.
[4] 李微娜,马小琪,冯艳光.基于MADM方法的个性化推荐研究[J].现代情报,2011,31(4):20-22,25.
[5] 何金晶.个性化信息推荐服务中用户潜在兴趣挖掘研究[J].现代情报,2013,33(4):8-11,16.
[6] 黄晓斌.基于协同过滤的数字图书馆推荐系统研究[J].大学图书馆学报,2006(1):53-57.
[7] 张红燕.高校图书馆新书推荐系统的研究与实现[J].大学图书馆学报,2013(5):106-113.
[8] 王国霞,刘贺平.个性化推荐系统综述[J].计算机工程与应用,2012,48(7):66-69.
[9] 王敏,嵇绍春.基于模糊聚类和模糊模式识别的数字图书馆个性化推荐研究[J]. 现代情报,2016,36(4):52-56.
[10] 王海艳,张大印.一种可信的基于协同过滤的服务选择模型[J].电子与信息学报,2013,35(2):349-354.
[11] NILASHI M,JANNACH D,IBRACHIM O,et al.Clustering-and regression-based multi-criteria collaborative filtering with incremental Updates[J].Information sciences,2015,293(3):235-250.
[12] 张莉,秦桃,滕丕强.一种改进的基于用户聚类的协同过滤算法[J].情报科学,2014,32(10):24-27,32.
[13] 李聪,梁昌勇.基于n序访问解析逻辑的协同过滤冷启动消除方法[J].系统工程理论与实践,2012,32(7):1537-1545.
[14] GOLDBERG D,NICHOLS D,OKI B M,et al. Using collaborative filtering to weave an information tapestry[J].Communications of the ACM,1992,35(12):61-70.
[15] 徐琳,杨姝. 基于改进的协同过滤算法的练习测试推荐系统[J].沈阳师范大学学报(自然科学版),2015,33(2):270-273.
[16] SARWAR B, KARYPIS G, JKONSTAN J,et al. Item-based collaborative filtering recommendation algorithms[C]]//Proceedings of the 10th international WWW conference. New York:ACM, 2001:285-295.
[17] ZANKER M, FELFERNIG A, FRIEDRICH G. Recommender systems:an introduction[M]. Cambridge:Cambridge University Press, 2011:124-142.
[18] 朱郁筱, 吕琳媛.推荐系统评价指标综述[J].电子科技大学学报, 2012, 41(2):163-175.
[19] WENG L T, XU Y, LI Y, et al. Improving recommendation novelty based on topic taxonomy[C]//IEEE/WIC/ACM international conferences on Web intelligence and intelligent agent technology. Washington:ACM, 2007:115-118.
[20] ZIEGLER C, MCNEE S M, KONSTAN J A,et al. Improving recommendation lists through topic diversification[C]//Proceedings of the 14th international conference on World Wide Web. Chiba:ACM, 2005:22-32.
[21] GE M, DELGADO-BATTENFEL D,JANNACH D. Beyond accuracy:evaluating recommender systems by coverage and serendipity[C]//RecSys (2010):the 2010 ACM conference on recommender systems.Barcelona:ACM,2010:257-260.
[22] 李鹏飞,吴为民. 基于混合模型推荐算法的优化[J].计算机科学,2014,41(2):68-73.
[23] 黄文彬,徐山川,吴家辉,等. 移动用户画像构建研究[J].现代情报,2016,36(10):54-62.
[24] 张新猛,蒋盛益,李霞,等. 基于网络和标签的混合推荐算法[J].计算机工程与应用,2015,51(1):119-124.
