研究了时间窗口对基于10种用户相似性指标的个性化推荐算法的影响。在标准数据集MovieLens上的实验结果表明,只采用大约12.56%的用户近期历史记录,所得到的推荐结果准确性可以平均提高27.17%,而推荐列表多样性可以平均提高3.28%,极大地降低大规模数据所带来的计算复杂性问题。
In this paper, we investigate the effect of the time window on the personalized recommendation algorithm based on ten similarity measures. The experimental results on the benchmark dataset MovieLens indicate that by only adapting approximately 12.56% recent rating records, the accuracy could be improved by an average of 27.17%, and the diversity could be improved by 3.28%. Our work is valuable in both theory and practice, and it could largely reduce the calculation complexity triggered by massive data.
[1] Lü L Y, Medo M, Yeung C H, et al. Recommender systems[J]. Physics Reports, 2012, 519(1): 1-49.
[2] 刘建国,周涛,郭强, 等. 个性化推荐系统评价方法综述[J]. 复杂系统与复杂性科学,2009,6(3):1-10.
Liu Jianguo, Zhou Tao, Guo Qiang, et al. Overview of the evaluated algorithms for the personal recommendation systems[J]. Complex Systems and Complexity Science, 2009,6(3): 1-10.
[3] Koren Y. Collaborative filtering with temporal dynamics[J]. Communications of the ACM, 2010, 53(4): 89-97.
[4] Liu J, Deng G. Link prediction in a user–object network based on time-weighted resource allocation[J]. Physica A, 2009, 388(17): 3643-3650.
[5] Zhou T, Kuscsik Z, Liu J G, et al. Solving the apparent diversity-accuracy dilemma of recommender systems[J]. Proceedings of the National Academy of Sciences, 2010, 107(10): 4511-4515.
[6] Zhang Q M, Zeng A, Shang M S. Extracting the information backbone in online system[J]. PloS one, 2013, 8(5): e62624.
[7] Guo Q, Song W J, Hou L, et al. Effect of the time window on the heat-conduction information filtering model[J]. Physica A, 2014, 401(5): 15-21.
[8] Guo Q, Li Y, Liu J G. Information filtering based on users′ negative opinions[J]. International Journal of Modern Physics C, 2013, 24(5): 1350032.
[9] 刘兆兴,张宁,李季明. 基于协同过滤和网络结构的个性化推荐算法[J]. 复杂系统与复杂性科学,2011,8(2): 45-50.
Liu Zhaoxing, Zhang Ning, Li Jiming. One personal recommendation algorithm based on collaborative filtering and network structure[J]. Complex Systems and Complexity Science, 2011,8(2): 45-50.
[10] Jaccard, P. étude comparative de la distribution florale dans une portion des Alpes et des Jura[J]. Bull Soc Vandoise Sci Nat, 1901, 37: 547-579.
[11] Sørensen T. A method of establishing groups of equal amplitude in plant sociology based on similarity of species and its application to analyses of the vegetation on Danish commons[J]. Biol Skr, 1948, 5: 1-34.
[12] Pan X, Deng G S, Liu J G. Information filtering via improved similarity definition[J]. Chinese Physics Letters, 2010, 27(6): 068903.
[13] Zhou T, Lü L, Zhang Y C. Predicting missing links via local information[J]. The European Physical Journal B, 2009, 71(4): 623-630.
[14] Guo Q, Leng R, Shi K, et al. Heat conduction information filtering via local information of bipartite networks[J]. The European Physical Journal B, 2012, 85(8): 1-8.
[15] Adamic L A, Adar E. Friends and neighbors on the web[J]. Social networks, 2003, 25(3): 211-230.
[16] Ravasz E, Somera A L, Mongru D A, et al. Hierarchical organization of modularity in metabolic networks[J]. Science, 2002, 297(5586): 1551-1555.
