本文针对推荐系统广泛存在的系统拥塞问题,首次提出了一种量化推荐拥塞程度的指标并比较了几种经典推荐算法的抗拥塞能力。结果显示推荐精度高的算法抗拥塞能力往往很差,而那些能较好防止推荐用塞的算法推荐精度又很低。为了解决这一两难问题,作者基于有向含权网络上的热传导过程提出一种新的推荐算法 DWC(Directed Weighted Conduction) 。在多个数据集上的实验表明该算法与以往经典算法相比,能够在保持推荐的准确性和多样性的同时,有效避免推荐系统陷入拥塞。本文提出的算法可应用于有限资源的产品或服务的推荐中,在电子商务领域具有广泛的应用前景。 论文下载地址: http://iopscience.iop.org/1367-2630/16/6/063057 作者:Xiaolong Ren, Linyuan Lu*, Runran Liu and Jianlin Zhang 摘要:Recommender systems use the historical activities and personal profiles of users to uncover their preferences and recommend objects. Most of the previous methods are based on objects' (and/or users') similarity rather than on their difference. Such approaches are subject to a high risk of increasingly exposing users to a narrowing band of popular objects. As a result, a few objects may be recommended to an enormous number of users, resulting in the problem of recommendation congestion, which is to be avoided, especially when the recommended objects are limited resources. In order to quantitatively measure a recommendation algorithm's ability to avoid congestion, we proposed a new metric inspired by the Gini index, which is used to measure the inequality of the individual wealth distribution in an economy. Besides this, a new recommendation method called directed weighted conduction (DWC) was developed by considering the heat conduction process on a user-object bipartite network with different thermal conductivities. Experimental results obtained for three benchmark data sets showed that the DWC algorithm can effectively avoid system congestion, and greatly improve the novelty and diversity, while retaining relatively high accuracy, in comparison with the state-of-the-art methods.
2013 年上映的电影《地心引力》浓墨重彩地描述了人类 失去 赖以生存的母体作用,生命也将随时面临危机,其关联关系显而易见。然而,万有引力与网络科学,与互联网有什么关系呢? 最近,虎哥的处女作 ” Gravity Effects on Information Filtering and Network Evolving ” 刚刚在 PLoSONE 上发表,详细地阐述了这方面的研究进展。 利用万有引力模型,进行社会经济信息方面的研究由来已久。如人口迁移 ,国际贸易 ,交通分析 ,人类空间行为预测 等。然而,在人类在线行为分析、建模和预测的工作中,万有引力模型的应用还很少见。本文利用人们的在线标注行为中的蕴藏的丰富信息,将用户和物品的标签数视为“质量”,将二者之间的共同兴趣大小 ( 相同标签个数 ) 视为“距离”,这样很自然地将引力模型引入到推荐系统中,刻画和预测未知二元关系的似然程度。进一步地,为了理解“万有引力”和网络增长的关系,我们将所改进的引力模型和 ER 及 BA 模型进行比较,发现基于引力模型演化的网络,其拓扑性质比 ER 、 BA 等随机网络更加贴近真实网络结构。总结而言,本文的贡献有以下三点: 1. 将万有引力模型引入到推荐系统中,方法简洁而新颖,结果也更优; 2. 基于兴趣的万有引力模型,比随机网络模型更能刻画真实网络; 3. 在推荐算法设计中,没有像传统推荐模型一样,显式地利用网络关系,而是从人的兴趣行为这一根本驱动力出发,匹配“人 - 物”这一对二元关系。更优算法的结果预示着人的兴趣行为可以用来有效地预测网络结构。这种从网络底层到网络表层的预测方法,越来越显示出一系列有趣、有效也更有解释性的优越性,我们在后面其他的工作中还会持续介绍。 另外,本文用到的数据,可在 论文官方网页 上提供免费下载。 编后注:利用引力模型做信息推荐,最初的灵感来自于博友 章成志 对博主一篇博文的评论,并介绍了一篇相关的计算机会议论文。当时我看完后,第一感觉是,问题很有趣,方法太繁琐,有如隔靴搔痒般的不爽利。经过大家几番试验后,终于采用最接近引力模型原始方式来处理。可以说,本工作完全是来自于科学网互动交流的启发。本文在最后也特别致谢了 章成志 博友。 参考文献: Karemera D, Oguledo VI, Davis B (2000) Agravity model analysis of international migration to north america. Appl Econ 32: 1745 – 1755. Rose AK (2004) Do we really know that thewto increases trade. Am Econ Rev 94: 98 – 114. Jung, WS, Wang F, Stanley, HE (2008)Gravity model in the Korean highway. EPL. 81: 48005 Simini, F, Gonz á lez MC, Maritan A, Barab á si, AL (2012). A universal model for mobility and migration patterns. Nature 484, 96-100 论文信息 :Jin-Hu Liu, Zi-Ke Zhang, Chengcheng Yang, Lingjiao Chen, Chuang Liu, XueqiWang. Gravity Effects on Information Filtering and Network Evolving. PLoS ONE 9(2014) e91070. 论文在线: http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0091070 本地下载: 2014PO-Gravity Effects on Information Filtering and Network Evolving.pdf