融合多源异构网络信息的标签推荐方法

doi:10.6040/j.issn.1671-9352.1.2018.100

摘要/Abstract

摘要：

标签通常被广泛地应用于标注各种在线资源，例如文章、图像、电影等，其主要目的是便于用户理解以及高效地管理和检索海量网络资源。因为人工对这些海量资源进行标注十分繁琐且耗时，所以自动化标签推荐技术被广泛关注。目前大部分标签推荐方法主要通过挖掘资源的内容信息进行推荐。然而，现实世界中很多数据信息并非独立存在，如文献数据通过相互引用关系而形成复杂的网络结构。研究表明，资源的拓扑结构信息和文本内容信息可分别从2个不同角度对同一资源的语义特征进行概括，并且从2个方面观察到的信息可以互为补充和解释。基于此，提出一种同时对资源内容信息和资源网络拓扑结构信息进行统一建模的概率主题模型和标签推荐方法。该方法通过结合标签和资源内容之间的标注关系以及资源之间的链接关系等多源异构信息，去挖掘资源潜在的语义信息为新的资源推荐若干功能语义相近的标签。

关键词: 标签, 标签推荐, 主题模型, 异构网络

Abstract:

Tags have been utilized extensively to associate various online resources, such as articles, images and movies, aiming at helping users understand and facilitate the process of managing and indexing huge web resources. Since it is time-consuming and prone for errors to create manual tags for these resources, automatic tag recommendation techniques have attracted widespread attention. At present, most tag recommendation methods mainly recommend tags by mining content information of resources. However, Most data information in the real world do not exist independently. For example, science articles have a complex network structure by referencing each other. The research show that the topology information and text content information of resources describe the similar semantic features of re-sources from two different perspectives, and the information from two aspects can complement and explain for each other. Based on this, we propose a probabilistic topic model and a tag recommendation method for simultaneously modeling content information and topology structure information of resource. This method uses multi-source heterogeneous information, such as tagging relationship between tag and resource content and link relationship between resources to mine potential semantic information of the resources to recommend several tags with similar functional semantics for the new resources. The experimental results on two real data sets prove the effectiveness of our proposed method.

Key words: tag, tag recommendation, topic model, heterogeneous network

中图分类号:

TP391

包恒泽,周栋,吴谈. 融合多源异构网络信息的标签推荐方法[J]. 《山东大学学报(理学版)》, 2019, 54(3): 56-66.

Heng-ze BAO,Dong ZHOU,Tan WU. Tag recommendation with multi-source heterogeneous networked information[J]. JOURNAL OF SHANDONG UNIVERSITY(NATURAL SCIENCE), 2019, 54(3): 56-66.

图/表 12

图1

图2

表1

图3

图4

表2

图5

图6

图7

图8

图9

表3

参考文献 22

1	王梦恬, 魏晶晶, 廖祥文, 等. 融合评论标签的个性化推荐算法[J]. 计算机科学与探索, 2016, 10 (10): 1429- 1438.
	WANG Mengtian , WEI Jingjing , LIAO Xiangwen , et al. Personalized recommendation algorithm fusing comment tag[J]. Journal of Frontiers of Computer Science & Technology, 2016, 10 (10): 1429- 1438.
2	BELÉM F M , ALMEIDA J M , GONALVES M A . A survey on tag recommendation methods[J]. Journal of the Association for Information Science and Technology, 2017, 68 (4): 830- 844. doi: 10.1002/asi.23736
3	ZHAO W , GUAN Z Y , LIU Z . Ranking on heterogeneous manifolds for tag recommendation in social taggingservices[J]. Neurocomputing, 2015, 148: 521- 534. doi: 10.1016/j.neucom.2014.07.011
4	BENZARTI M A, CHIDLOVSKⅡ B, VIJAYAKUMAR N. Local metric learning for tag recommendation in social networks using indexing.: US, 9600826[P], 2017.
5	GUO Z , ZHANG Z M , ZHU S H , et al. A two-level topic model towards knowledge discovery from citation networks[J]. IEEE Transactions on Knowledge and Data Engineering, 2014, 26 (4): 780- 794. doi: 10.1109/TKDE.2013.56
6	WEST J D , WESLEY-SMITH I , BERGSTROM C T . A recommendation system based on hierarchical clustering of an article-level citation network[J]. IEEE Transactions on Big Data, 2016, 2 (2): 113- 123. doi: 10.1109/TBDATA.2016.2541167
7	WANG M , NI B B , HUA X S , et al. Assistive tagging: a survey of multimedia tagging with human-computer joint exploration[J]. ACM Computing Surveys, 2012, 44 (4): 1- 24.
8	褚晓敏, 王中卿, 朱巧明, 等. 基于简介和评论的标签推荐方法研究[J]. 中文信息学报, 2015, 29 (6): 179- 184. doi: 10.3969/j.issn.1003-0077.2015.06.024
	CHU Xiaomin , WANG Zhongqing , ZHU Qiaoming , et al. Tag recommendation with summary and comment information[J]. Journal of Chinese Information Processing, 2015, 29 (6): 179- 184. doi: 10.3969/j.issn.1003-0077.2015.06.024
9	LIU Z, CHEN X, SUN M. A simple word trigger method for social tag suggestion[C]// Proceedings of the Conference on Empirical Methods in Natural Language Processing. Edinburgh: Association for Computational Linguistics, 2011: 1577-1588.
10	BELÉM F M , MARTINS E F , ALMEIDA J M , et al. Personalized and object-centered tag recommendation methods for Web 2.0 applications[J]. Information Processing & Management, 2014, 50 (4): 524- 553.
11	LI W, YEUNG D Y. Social relations model for collaborative filtering[C]// Proceedings of the Twenty-fifth AAAI Conference on Artificial Intelligence. San Francisco: AAAI Press, 2011: 803-808.
12	LOPS P , DE GEMMIS M , SEMERARO G , et al. Content-based and collaborative techniques for tag recommendation: an empirical evaluation[J]. Journal of Intelligent Information Systems, 2013, 40 (1): 41- 61. doi: 10.1007/s10844-012-0215-6
13	BAO Y, FANG H, ZHANG J. TopicMF: simultaneously exploiting ratings and reviews for recommendation[C]// Proceedings of the Twenty-eighth AAAI Conference on Artificial Intelligence. Québeca: AAAI Press, 2014: 2-8.
14	BURKE R, VAHEDIAN F, MOBASHER B. Hybrid recommendation in heterogeneous networks[C]// Proceedings of the International Conference on User Modeling, Adaptation, and Personalization. Aalborg: Springer, 2014: 49-60.
15	AN S, ZHAO Z, ZHOU H. Research on an agent-based intelligent social tagging recommendation system[C]// Proceedings of International Conference on Intelligent Human-Machine Systems and Cybernetics. Hangzhou: IEEE, 2017: 43-46.
16	闫俊, 刘文飞, 林鸿飞. 基于标签混合语义空间的音乐推荐方法研究[J]. 中文信息学报, 2014, 28 (4): 117- 122. doi: 10.3969/j.issn.1003-0077.2014.04.016
	YAN Jun , LIU Wenfei , LIN Hongfei . Music recommendation study based on tags multi-space[J]. Journal of Chinese Information Processing, 2014, 28 (4): 117- 122. doi: 10.3969/j.issn.1003-0077.2014.04.016
17	于洪, 邓明瑶, 胡峰. 考虑用户标注状态的标签推荐方法[J]. 模式识别与人工智能, 2014, 27 (8): 673- 682. doi: 10.3969/j.issn.1003-6059.2014.08.001
	YU Hong , DENG Mingyao , HU Feng . Tag recommendation method considering users tagging status[J]. Pattern Recognition and Artificial Intelligence, 2014, 27 (8): 673- 682. doi: 10.3969/j.issn.1003-6059.2014.08.001
18	WANG H, SHI X, YEUNG D Y. Relational stacked denoising autoencoder for tag recommendation[C]// Proceedings of the Twenty-ninth AAAI Conference on Artificial Intelligence. Austin: AAAI Press, 2015: 3052-3058.
19	KRESTEL R, FANKHAUSER P, NEJDL W. Latent dirichlet allocation for tag recommendation[C]// Proceedings of the third ACM conference on Recommender systems. New York: ACM, 2009: 61-68.
20	SI X , SUN M . Tag-LDA for scalable real-time tag recommendation[J]. Journal of Computational Information Systems, 2008, 6 (1): 1- 8.
21	CHANG J, BLEI D M. Relational topic models for document networks[C]// Proceedings of the 12th International Conference on Artificial Intelligence and Statistics. Florida: JMLR, 2009: 81-88.
22	WANG H, CHEN B, LI W J. Collaborative topic regression with social regularization for tag recommendation[C]// Proceedings of the Twenty-third International Joint Conference on Artificial Intelligence. Beijing: AAAI Press, 2013: 2719-2725.

多维度评价

Viewed

Full text

Abstract

Cited

Shared

Discussed

符号	含义说明
C	表示文章C
C′	表示文章C′
W^(c)	表示文章C的文本内容
W^(c′)	表示文章C′的文本内容
θ^(c)	表示W^(c)的主题分布向量
θ^(c′)	表示W^(c′)的主题分布向量
T	表示模型中主题的数量
W	语料库中词汇数量
W_t	语料库中标签数量
ϕ	维度为W的向量表示主题下词的分布
ϕ_t	维度为W_t的向量表示主题下标签的分布
N	根据文章间相似性选择与候选文章相似度最高的N篇文章
M	通过标签过滤算法选出得分最高的M个标签
S	未排序的待推荐标签集
S′	排序后的待推荐标签集

数据集	数据	数量
citeulike-a	文章	16980
	标签	19107
	移除使用次数少于5次后剩余标签	7450
	引用关系	294072
citeulike-t	文章	25975
	标签	52946
	移除使用次数少于5次后剩余标签	8311
	引用关系	180103

模型	主题最相近文章	推荐标签	正确标签
RTM	“an efficient algorithm to rank web resources” “rank algorithm web graph link” “searching social networks” “thermal barrier coatings for gas-turbine engine applications”	pager，resource，scale，web，rank，search，citation，engine，hyperlink，application	web, rank, engine
TRTM	“the anatomy of a large scale hyper textual web search engine” “an overview of audio information retrieval” “pager citation ranking bringing order to web” “searching social networks”	engine，retrieval，link，hyperlink，search，citation，rank，web，relevance，citation	engine，link，search，rank，web

[1]	王雪梅,陈兴蜀,王海舟,王文贤. 基于标签和分块特征的新闻网页关键信息自动抽取[J]. 《山东大学学报(理学版)》, 2019, 54(3): 67-74.
[2]	杜漫,徐学可,杜慧,伍大勇,刘悦,程学旗. 面向情绪分类的情绪词向量学习[J]. 山东大学学报（理学版）, 2017, 52(7): 52-58.
[3]	张新猛, 蒋盛益, 张倩生, 谢柏林, 李霞. 基于用户偏好加权的混合网络推荐算法[J]. 山东大学学报（理学版）, 2015, 50(09): 29-35.
[4]	马宇峰, 阮彤. 基于LDA及标签传播的实体集合扩展[J]. 山东大学学报（理学版）, 2015, 50(03): 20-27.
[5]	郑妍, 庞琳, 毕慧, 刘玮, 程工. 基于情感主题模型的特征选择方法[J]. 山东大学学报（理学版）, 2014, 49(11): 74-81.
[6]	王少鹏, 彭岩, 王洁. 基于LDA的文本聚类在网络舆情分析中的应用研究[J]. 山东大学学报（理学版）, 2014, 49(09): 129-134.
[7]	焦潞林, 彭岩, 林云. 面向网络舆情的文本知识发现算法对比研究[J]. 山东大学学报（理学版）, 2014, 49(09): 62-68.
[8]	刘璇1,许洁萍1*,陈捷2. 以Web标签为基础的相似歌曲研究[J]. J4, 2012, 47(5): 53-58.