Unsupervised Matrix Factorization Based Trigger Action Programming Rules Recommendation

doi:10.3969/j.issn.1671-1122.2023.02.011

Abstract

Abstract:

TAP has been widely used in customized IoT device linkage. In addition to the conditional trigger relationship between items, the TAP data also contains the text description information about the relevant rules. How to use the multi-source heterogeneous attributes of TAP data is one of the important researches in the application of the IoT. In this paper, TAP data was modeled as a heterogeneous graph containing multiple types of nodes and edges, which realized the fusion of multiple types of relationships between multi-source heterogeneous data, and then generated a relationship matrix according to the connection between different types of nodes. Non-negative matrix factorization (NMF) was used to unsupervised learn the feature of each node in the TAP heterogeneous graph for TAP rule recommendation. This paper proposed three weighted relational matrix generation methods, which were called co-occurrence frequency weight (CFW), concept similarity weight (CSW) and TF-IDF weight (TIW). The experimental results show that feature vectors obtained from NMF, which decomposes the matrix generated by CFW have better performance in TAP rule recommendation.

Key words: trigger-action programming, non-negative matrix, weighted matrix

CLC Number:

TP309

WANG Ming, XING Yongheng, WANG Feng. Unsupervised Matrix Factorization Based Trigger Action Programming Rules Recommendation[J]. Netinfo Security, 2023, 23(2): 96-103.

Figures/Tables 5

References 22

[1]	ASHTON K. That ‘Internet of Things’ Thing[J]. RFID Journal, 2009, 22(7): 97-114.
[2]	UR B, MCMANUS E, PAK Y H M, et al. Practical Trigger-Action Programming in the Smart Home[C]// ACM. SIGCHI Conference on Human Factors in Computing Systems. New York: ACM, 2014: 803-812.
[3]	SONG Yu. Representation Learning for Heterogeneous Edge Networks Based on Multi-Stage Non-Negative Matrix Factorization[D]. Wuhan: Huazhong University of Science and Technology, 2021.
	宋宇. 基于多阶段非负矩阵分解的异质边网络表示学习研究[D]. 武汉: 华中科技大学, 2021.
[4]	LEENDERS J. Modeling Social Influence Through Network Autocorrelation: Constructing the Weight Matrix[J]. Social Networks, 2002, 24(1): 21-47. doi: 10.1016/S0378-8733(01)00049-1 URL
[5]	SPEER R, CHIN J, HAVASI C. Conceptnet 5.5: An Open Multilingual Graph of General Knowledge[EB/OL]. (2016-12-12)[2022-09-20]. https://arxiv.org/pdf/1612.03975.pdf.
[6]	MILLER G A. WordNet: An Electronic Lexial Database[M]. Cambridge: MIT Press, 1998.
[7]	ROBERTSON S. Understanding Inverse Document Frequency: on Theoretical Arguments for IDF[J]. Journal of Documentation, 2004, 60(5): 503-520. doi: 10.1108/00220410410560582 URL
[8]	BIAN Han, CHEN Xiaohong, JIN Zhi, et al. A Method of Generating TAP Rules in IoT System Based on Environment Modeling[J]. Journal of Software, 2021, 32(4): 934-952.
	边寒, 陈小红, 金芝, 等. 基于环境建模的物联网系统 TAP 规则生成方法[J]. 软件学报, 2021, 32(4): 934-952.
[9]	WANG Bo, ZHANG Yu, GENG Jianing, et al. SSRules: Make Smart Home Automation Rules Easier to Write and Check[J]. Journal of Software, 2021, 32(12): 3728-3750.
	王博, 张昱, 耿佳宁, 等. SSRules: 让智能家居自动化规则更易于编写和检查[J]. 软件学报, 2021, 32(12): 3728-3750.
[10]	WANG Chen. Conflict Detection Framework for End-User Rules of Smart IoT Devices Based on SMT[D]. Shanghai: East China Normal University, 2022.
	汪晨. 基于SMT的智能IoT设备终端用户规则的冲突检测框架[D]. 上海: 华东师范大学, 2022.
[11]	CORNO F, RUSSIS L, ROFFARELLO A M. A Semantic Web Approach to Simplifying Trigger-Action Programming in the IoT[J]. Computer, 2017, 50(11): 18-24.
[12]	CORNO F, RUSSIS L, ROFFARELLO A M. A High-Level Semantic Approach to End-User Development in the Internet of Things[J]. International Journal of Human-Computer Studies, 2019, 125: 41-54. doi: 10.1016/j.ijhcs.2018.12.008 URL
[13]	CORNO F, RUSSIS L, MONGE R A. RecRules: Recommending IF-THEN Rules for End-User Development[J]. ACM Transactions on Intelligent Systems and Technology (TIST), 2019, 10(5): 1-27.
[14]	CORNO F, RUSSIS L, MONGE R A. Empowering End Users in Debugging Trigger-Action Rules[C]// ACM. 2019 CHI Conference on Human Factors in Computing Systems. New York: ACM, 2019: 1-13.
[15]	HU Liang, WU Gang, XING Yongheng, et al. Semantic Modeling in the Internet of Things with Graph Representation Learning[J]. IEEE Internet of Things Journal, 2019, 7(3): 1939-1948. doi: 10.1109/JIOT.2019.2962630 URL
[16]	KLEMA V, LAUB A. The Singular Value Decomposition: Its Computation and some Applications[J]. IEEE Transactions on Automatic Control, 1980, 25(2): 164-176. doi: 10.1109/TAC.1980.1102314 URL
[17]	LEE D, SEUNG H S. Algorithms for Non-Negative Matrix Factorization[EB/OL]. (2001-05-11)[2022-09-22]. http://www.cs.cmu.edu/-11755/lectures/Lee_Seung_NMF.pdf.
[18]	SEDGHI H, GUPTA V, LONG P M. The Singular Values of Convolutional Layers[EB/OL]. (2018-05-26)[2022-09-22]. https://arxiv.org/pdf/1805.10408.pdf.
[19]	KOREN Y, BELL R, VOLINSKY C. Matrix Factorization Techniques for Recommender Systems[J]. Computer, 2009, 42(8): 30-37.
[20]	CHEN Chong, ZHANG Min, ZHANG Yongfeng, et al. Efficient Neural Matrix Factorization without Sampling for Recommendation[J]. ACM Transactions on Information Systems (TOIS), 2020, 38(2): 1-28.
[21]	XING Yongheng, HU Liang, ZHANG Xiaolu, et al. Nonnegative Matrix Factorization Based Heterogeneous Graph Embedding Method for Trigger-Action Programming in IoT[J]. IEEE Transactions on Industrial Informatics, 2021, 18(2): 1231-1239. doi: 10.1109/TII.2021.3092774 URL
[22]	HU Liang, GONG Yanlei, XING Yongheng, et al. Semantic Representation with Heterogeneous Information Network Using Matrix Factorization for Clustering in the Internet of Things[J]. IEEE Access, 2019, 7: 31233-31242. doi: 10.1109/ACCESS.2019.2903310

权重矩阵	Precision@1	Precision@3	Precision@5
SR_128	0.6474	0.5507	0.4955
CFW_128	0.6783	0.5697	0.5100
CSW_128	0.6463	0.5417	0.4851
TIW_128	0.3066	0.2630	0.2399
SR_256	0.6673	0.5451	0.4809
CFW_256	0.6853	0.5837	0.5134
CSW_256	0.6763	0.5607	0.5042
TIW_256	0.3256	0.2700	0.2417
SR_512	0.6643	0.5514	0.4899
CFW_512	0.6913	0.5907	0.5248
CSW_512	0.6713	0.5594	0.4931
TIW_512	0.3896	0.3276	0.2923

权重矩阵	MRR
SR_128	0.8380
CFW_128	0.8559
CSW_128	0.8502
TIW_128	0.6217
SR_256	0.8557
CFW_256	0.8603
CSW_256	0.8465
TIW_256	0.6454
SR_512	0.8491
CFW_512	0.8708
CSW_512	0.8539
TIW_512	0.6714