›› 2012,Vol. 27 ›› Issue (3): 527-540.doi: 10.1007/s11390-012-1241-0

• • 上一篇    下一篇

使用社会影响力进行个性化标签推荐

Jun Hu (胡军), Bing Wang (王兵), Yu Liu (刘禹), Member, CCF, and De-Yi Li (李德毅), Fellow, CCF   

  • 收稿日期:2011-08-31 修回日期:2012-01-19 出版日期:2012-05-05 发布日期:2012-05-05

Personalized Tag Recommendation Using Social Influence

Jun Hu (胡军), Bing Wang (王兵), Yu Liu (刘禹), Member, CCF, and De-Yi Li (李德毅), Fellow, CCF   

  1. State Key Laboratory of Software Development Environment, Beihang University, Beijing 100191, China
  • Received:2011-08-31 Revised:2012-01-19 Online:2012-05-05 Published:2012-05-05
  • About author:Jun Hu is now a Ph.D. candidate in computer science and technology in Beihang University. He received his Master's degree in computer scie-nce from Information Engineering University, China, in 2007. His re-search interests include social net-work analysis, recommendation sys-tem, data mining methodologies and machine learning algorithms.
  • Supported by:

    This work was supported by the National Basic Research 973 Program of China under Grant No. 2007CB310803, the National Natural Science Foundation of China under Grant Nos. 61035004, 60974086, and the Project of the State Key Laboratory of Software Development Environment of China under Grant Nos. SKLSDE-2010ZX-16, SKLSDE-2011ZX-08.

标签推荐可以鼓励用户为资源添加更多的标签,从而丰富资源的语义描述,便于资源的组织和分享.另外,通过标签推荐可以实现多媒体资源的自动标注,为基于内容的多媒体信息检索提供了一种可行的解决方案.本文选用Flickr这一社会标注系统为载体,利用其中的连接信息构建在线的社会网络.从网络拓扑的角度,提出基于拓扑势的用户社会影响力度量方法.通过使用该指标,可以实现对用户社会关系的区分,找到对用户具有真正影响力的邻居节点,并从中挖掘出用户的潜在偏好.标签的推荐在全局标签同现基础上,融合用户个人历史标注信息、社会网络上影响力用户所属偏好社区信息,并给出相应的推荐框架.最后使用采集到的Flickr真实数据对算法进行测试与验证,并且与目前主流的个性化标签推荐算法进行比较.实验结果不仅证明了基于用户影响力的潜在偏好挖掘方法的有效性,还表明个性化标签推荐算法在推荐准确度和个性化推荐能力上要优于其他算法.
本文的主要贡献有以下两个方面:一是提出了一种用于社会网络挖掘新的用户影响力度量方法,在物理学和信息论的指导下,从网络拓扑出给出了用户影响力的定义,用于挖掘用户的潜在偏好.二是提出了融合用户影响力、用户标注行为和全局信息的个性化推荐算法,与其他算法相比推荐的成功率s@3从68%提高到83%.
相关研究成果可以直接应用社会标注系统中,用于提高查询的准确性与改善用户体验.

Abstract: Tag recommendation encourages users to add more tags in bridging the semantic gap between human concept and the features of media object, which provides a feasible solution for content-based multimedia information retrieval. In this paper, we study personalized tag recommendation in a popular online photo sharing site —— Flickr. Social relationship information of users is collected to generate an online social network. From the perspective of network topology, we propose node topological potential to characterize user's social influence. With this metric, we distinguish different social relations between users and find out those who really have influence on the target users. Tag recommendations are based on tagging history and the latent personalized preference learned from those who have most influence in user's social network. We evaluate our method on large scale real-world data. The experimental results demonstrate that our method can outperform the non-personalized global co-occurrence method and other two state-of-the-art personalized approaches using social networks. We also analyze the further usage of our approach for the cold-start problem of tag recommendation.

[1] Sigurbjörnsson B, van Zwol R. Flickr tag recommendationbased on collective knowledge. In Proc. the 17th Inter-national Conference on World Wide Web, Beijing, China,Apr. 21-25, 2008, pp.327-336.

[2] Smeulders A, Worring M, Santini S, Gupta A, Jain R.Content-based image retrieval at the end of the early years.IEEE Transactions Pattern Analysis Machine Intelligence,2000, 22(12): 1349-1380.

[3] Ames M, Naaman M. Why we tag: Motivations for annota-tion in mobile and online media. In Proc. the 2007 SIGCHIConference on Human Factors in Computing Systems, SanJose, USA, Apr. 28-May 3, 2007, pp.971-980.

[4] Gemmell J, Ramezani M, Schimoler T, Christiansen L,Mobasher B. The impact of ambiguity and redundancy ontag recommendation in folksonomies. In Proc. the 3rd ACMConference on Recommender Systems, New York City, USA,Oct. 23-25, 2009, pp.45-52.

[5] Vojnovic M, Cruise J, Gunawardena D, Marbach P. Rankingand suggesting popular items. IEEE Transactions on Know-ledge and Data Engineering, 2009, 21(8): 1133-1146.

[6] Guan Z Y, Bu J J, Mei Q Z, Chen C, Wang C. Personalizedtag recommendation using graph-based ranking on multi-typeinterrelated objects. In Proc. the 32nd Annual InternationalACM SIGIR Conference on Research and Development in In-formation Retrieval, Boston, USA, July 19-23, 2009, pp.540-547.

[7] Yin D W, Xue Z Z, Hong L J, Davison B D. A probabilisticmodel for personalized tag prediction. In Proc. the 16th ACM SIGKDD International Conference on Knowledge Discoveryand Data Ming, Washington, USA, Jul. 25-28, 2010, pp.959-968.

[8] Lipczak M, Milios E. Learning in efficient tag recommenda-tion. In Proc. the 4th ACM Conference on RecommenderSystems, Barcelona, Spain, Sept. 26-30, 2010, pp.167-174.

[9] Koren Y. Factorization meets the neighborhood: A multi-faceted collaborative filtering model. In Proc. the 14th ACMSIGKDD International Conference on Knowledge Discoveryand Data Mining, Las Vegas, USA, Aug. 24-27, 2008, pp.426-434.

[10] J?aschke R, Marinho L, Hotho A, Schmidt-Thieme L, StummeG. Tag recommendations in folksonomies. In Proc. the 11thEuropean Conference on Principles and Practice of Know-ledge Discovery in Databases, Warsaw, Poland, September17-21, 2007, pp.506-514.

[11] Symeonidis P, Nanopoulos A, Manolopoulos Y. Tag recom-mendations based on tensor dimensionality reduction. InProc. the 2nd ACM Conference on Recommender Systems,Lausanne, Switzerland, October 23-25, 2008, pp.43-50.

[12] Rendle S, Marinho B L, Nanopoulos A, Schmidt-Thieme L.Learning optimal ranking with tensor factorization for tagrecommendation. In Proc. the 15th ACM SIGKDD Interna-tional Conference on Knowledge Discovery and Data Mining,Paris, France, Jun. 28-Jul. 1, 2009, pp.727-736.

[13] Tatu M, Srikanth M, D'Silva T. Tag recommendations us-ing bookmark content. In Proc. the 2008 ECML/PKDDDiscovery Challenge Workshop, Antwerp, Belgium, Sep. 15-19, 2008, pp.96-107.

[14] Koren Y. Collaborative filtering with temporal dynamics. InProc. the 15th ACM SIGKDD International Conferenceon Knowledge Discovery and Data Mining, Paris, France,Jun. 28-Jul. 1, 2009, pp.447-456.

[15] Garg N, Weber I. Personalized, interactive tag recommen-dation for flickr. In Proc. the 2nd ACM Conference onRecommender Systems, Lausanne, Switzerland, October 23-25, 2008, pp.67-74.

[16] Rae A, Sigurbjornsson B, van Zwol R. Improving tag reco-mmendation using social networks. In Proc. the 9th Inter-national Conference on Adaptivity, Personalization and Fu-sion of Heterogeneous Information, Paris, France, Apr. 28-30,2010, pp.92-99.

[17] Ma H, King I, Lyu M R. Learning to recommend with explicitand implicit social relations. ACM Transactions on Intelli-gent Systems and Technology, 2011, 2(3): Article No. 29.

[18] Lipczak M, Milios E. The impact of resource title on tagsin collaborative tagging systems. In Proc. the 21st ACMConference on Hypertext and Hypermedia, Toronto, Canada,Jun. 13-16, 2010, pp.179-188.

[19] Mislove A, Koppula H S, Gummadi K P, Drusche P, Bhat-tacharjee B. Growth of the flickr social network. In Proc.the 1st Workshop on Online Social Networks, Seattle, USA,Aug. 18, 2008, pp.25-30.

[20] Cha M Y, Mislove A, Gummadi K P. A measurement-drivenanalysis of information propagation in the flickr social net-work. In Proc. the 18th International Conference on WorldWide Web, Madrid, Spain, Apr. 20-24, 2009, pp.721-730.

[21] Li D Y, Du Y. Artifical Intelligence With Uncertainty. BocaRaton, Florida: Chapman and Hall/CRC Press, 2008.

[22] Wu L, Yang L J, Yu N H, Hua X S. Learning to tag. InProc. the 18th International Conference on World WideWeb, Madrid, Spain, Apr. 20-24, 2009, pp.361-370.

[23] Liu N N, Yang Q. Eigenrank: A ranking-oriented approachto collaborative filtering. In Proc. the 31st Annual ACMSIGIR International Conference on Research and Develop-ment in Information Retrieval, Singapore, Singapore, Jul. 20-24, 2008, pp.83-90.

[24] Shen D, Sun J T, Yang Q, Chen Z. A comparison of implicitand explicit links for web page classification. In Proc. the15th International Conference on World Wide Web, Edin-burgh, Scotland, May 23-26, 2006, pp.643-650.
No related articles found!
Viewed
Full text


Abstract

Cited

  Shared   
  Discussed   
[1] 许志明;. Discrete Interpolation Surface[J]. , 1990, 5(4): 329 -332 .
[2] 蔡士杰; 张福炎;. A Fast Algorithm for Polygon Operations[J]. , 1991, 6(1): 91 -96 .
[3] 廖先湜; 金兰;. A Mechanism Supporting the Client/Server Relationship in the Operating System of Distributed System “THUDS”[J]. , 1991, 6(3): 256 -262 .
[4] 沈一栋;. Form alizing Incomplete Knowledge in Incomplete Databases[J]. , 1992, 7(4): 295 -304 .
[5] 吴信东;. Inductive Learning[J]. , 1993, 8(2): 22 -36 .
[6] 张钹; 张铃;. On Memory Capacity of the Probabilistic Logic Neuron Network[J]. , 1993, 8(3): 62 -66 .
[7] 姜文彬;. A Method for Minimization Design of Two-Level Logic Networks Using Multiplexer Universal Logic Modules[J]. , 1994, 9(1): 92 -96 .
[8] 黄国勇; 李三立;. TSP: A Heterogeneous Multiprocessor Supercomputing System Based on i860XP[J]. , 1994, 9(3): 285 -288 .
[9] 余盛可;. Reasoning in H-Net: A Unified Approach to Intelligent Hypermedia Systems[J]. , 1996, 11(1): 83 -89 .
[10] 田增平; 王宇君; 曲云尧; 施伯乐;. On the Expressive Power of F-Logic Language[J]. , 1997, 12(6): 510 -519 .
版权所有 © 《计算机科学技术学报》编辑部
本系统由北京玛格泰克科技发展有限公司设计开发 技术支持:support@magtech.com.cn
总访问量: