基于真⁃伪三连体神经网络的传感器本体匹配方法

doi:10.13229/j.cnki.jdxbgxb.20250233

摘要/Abstract

摘要：

针对物联网系统中传感器本体交互所面临的异质性问题，提出了一种基于真-伪三连体神经网络（GP-TNN）的传感器本体匹配方法。首先，根据实体的出度入度提取核心实体，进而提升训练样本的质量；然后，通过设计确定性的混合相似度度量方法，用于生成高置信度的锚点匹配对，以此避免训练过程中对参考对齐的依赖；最后，结合孪生网络和伪孪生网络对同类及异类标注属性的表征能力，构建GP-TNN增强深层次语义特征挖掘能力。评估实验分别在本体对齐评测计划（OAEI）中Benchmark数据集以及3个真实的传感器本体上进行，实验结果表明，本文方法较对比方法能够更加有效地提高匹配结果的质量，为物联网中传感器本体匹配任务提供了可靠的理论支持。

关键词: 计算机应用, 深度学习, 真?伪三连体网络, 传感器本体匹配, 本体对齐评测计划

Abstract:

To address the heterogeneity problems in sensor ontology interaction within Internet of Things （IoT） systems， this paper proposes a sensor ontology matching method based on a Genuine-Pseudo Triplet Neural Network（GP-TNN）. Firstly， core entities are extracted by analyzing the in-degree and out-degree of entities to enhance the quality of training samples. Secondly， a deterministic hybrid similarity measurement method is designed to generate high-confidence anchor matching pairs， thereby reducing the dependence on reference alignments during training. Finally， by leveraging the representation capabilities of Siamese and pseudo-Siamese networks for homogeneous and heterogeneous annotation attributes， the GP-TNN is constructed to enhance the extraction of deep semantic features. Evaluation experiments are conducted on the Benchmark dataset from the Ontology Alignment Evaluation Initiative（OAEI） and three real-world sensor ontologies. Experimental results show that the proposed method can more effectively improve the quality of matching results than the comparative methods， and provides reliable theoretical support for the sensor ontology matching task in the Internet of Things.

Key words: computer application, deep learning, genuine-pseudo triplet neural network, sensor ontology matching, ontology alignment evaluation initiative

中图分类号:

TP391

卢家伟,剡昌锋. 基于真⁃伪三连体神经网络的传感器本体匹配方法[J]. 吉林大学学报(工学版), 2026, 56(9): 2456-2466.

Jia-wei LU,Chang-feng YAN. Sensor ontology matching method based on genuine⁃pseudo triplet neural network[J]. Journal of Jilin University(Engineering and Technology Edition), 2026, 56(9): 2456-2466.

图/表 11

图1

图2

表1

图3

图4

表2

表3

表4

图5

图6

图7

参考文献 31

[1]	Abdulazeez D H, Askar S K. A novel offloading mechanism leveraging fuzzy logic and deep reinforcement learning to improve IoT application performance in a three-layer architecture within the fog-cloud environment[J]. IEEE Access, 2024, 12: 39936-39952.
[2]	Chen X, Cao B, Ahmad I. Lightweight neural network‐based viewport prediction for live VR streaming in wireless video sensor network[J]. Mobile Information Systems, 2021, 2021(1): 1-12.
[3]	Nivetha A, Preetha K S. A review on cyber-twin in sixth generation wireless networks: architecture, research challenges & issues[J]. Wireless Personal Communications, 2024, 138(3): 1815-1865.
[4]	Geng A, Lü Q. A multi-objective particle swarm optimization with density and distribution-based competitive mechanism for sensor ontology meta-matching[J]. Complex & Intelligent Systems, 2023, 9(1): 435-462.
[5]	Hamane Z, Samih A, Fennan A. Ontology matching using deep learning[J]. Journal of Theoretical and Applied Information Technology, 2023, 101(10): 4041-4058.
[6]	Bento A, Zouaq A, Gagnon M. Ontology matching using convolutional neural networks[C]∥Proceedings of the 12th Language Resources and Evaluation Conference, Marseille, France, 2020: 5648-5653.
[7]	Jiang C, Xue X. Matching biomedical ontologies with long short-term memory networks[C]∥IEEE International Conference on Bioinformatics and Biomedicine, Seoul, Korea, 2020: 2484-2489.
[8]	Iyer V, Agarwal A, Kumar H. VeeAlign: a supervised deep learning approach to ontology alignment[C]∥Proceedings of the 15th International Workshop on Ontology Matching, Athens, Greece, 2020: 216-224.
[9]	Xue X, Jiang C, Zhang J, et al. Matching sensor ontologies through Siamese neural networks without using reference alignment[J]. PeerJ Computer Science, 2021, 7: No.e602.
[10]	Wang P, Zou S, Liu J, et al. Matching biomedical ontologies with GCN-based feature propagation[J]. Mathematical Biosciences and Engineering, 2022, 19(8): 8479-8504.
[11]	段阳, 侯力, 冷松. 金属切削加工知识图谱构建及应用[J]. 吉林大学学报: 工学版, 2021, 51(1): 122-133.
	Duan Yang, Hou Li, Leng Song. Building and application of metal cutting knowledge graph[J]. Journal of Jilin University(Engineering and Technology Edition), 2021, 51(1): 122-133.
[12]	Xue X, Yang C, Jiang C, et al. Optimizing ontology alignment through linkage learning on entity correspondences[J]. Complexity, 2021, 2021(1): 1-12.
[13]	Xue X, Wang Y, Improving the efficiency of NSGA⁃II based ontology aligning technology[J], Data & Knowledge Engineering, 2017, 108: 1-14.
[14]	Chu S C, Xue X, Pan J S, et al. Optimizing ontology alignment in vector space[J]. Journal of Internet Technology, 2020, 21 (1):15-22.
[15]	Mascardi V, Locoro A, Rosso P. Automatic ontology matching via upper ontologies: a systematic evaluation[J]. IEEE Transactions on Knowledge and Data Engineering, 2009, 22(5): 609-623.
[16]	Lu J, Xue X, Lin G, et al. A new ontology meta-matching technique with a hybrid semantic similarity measure[C]∥Advances in Intelligent Information Hiding and Multimedia Signal Processing, Changchun, China, 2020: 37-45.
[17]	Phan T T M, Duong A T, Tran P T M. WordNet-based computation of semantic similarity between two Vietnamese nouns[J]. Technium Social Sciences Journal, 2023, 43: 532-545.
[18]	Yu Y, Zhou B, Chen L, et al. Identifying important nodes in complex networks based on node propagation entropy[J]. Entropy, 2022, 24(2): 275-291.
[19]	Chicco D. Siamese neural networks: an overview[J]. Artificial Neural Networks, 2021, 2190: 73-94.
[20]	Ni W, Xu Q, Jiang Y, et al. PSNEA: Pseudo-siamese network for entity alignment between multi-modal knowledge graphs[C]∥Proceedings of the 31st ACM International Conference on Multimedia, New York, USA, 2023: 3489-3497.
[21]	Li Y, Chen C L P, Zhang T. A survey on siamese network: methodologies, applications, and opportunities[J]. IEEE Transactions on Artificial Intelligence, 2022, 3(6): 994-1014.
[22]	王神龙, 雍宇, 吴晨睿. 基于伪孪生神经网络的低纹理工业零件6D位姿估计[J]. 电子学报, 2023, 51(1): 192-201.
	Wang Shen-long, Yong Yu, Wu Chen-rui. 6D pose estimation of low texture industrial parts based on Pseudo-Siamese neural network[J]. Acta Electronica Sinica, 2023, 51(1): 192-201.
[23]	Gong Y, Yue Y, Ji W, et al. Cross-domain few-shot learning based on Pseudo-Siamese neural network[J]. Scientific Reports, 2023, 13(1): 1-14.
[24]	Xue X, Jiang C, Zhu H. Matching Ontologies Through Siamese Neural Network[C]∥Mobile Multimedia Communications: 14th EAI International Conference, Online, 2021: 715-724.
[25]	Faria D, Pesquita C, Balasubramani B S, et al. OAEI 2016 results of AML[C]∥Proceedings of the 11th International Workshop on Ontology Matching co-located with the 15th International Semantic Web Conference, Kobe, Japan, 2016: 138-145.
[26]	Jiménez-Ruiz E, Grau B C, Cross V. LogMap family participation in the OAEI 2017[C]∥Proceedings of the 12th International Workshop on Ontology Matching Co-located with the 16th International Semantic Web Conference, Vienna, Austria, 2017: 153-157.
[27]	Djeddi W E, Khadir M T, Yahia S B. XMap results for OAEI 2016[C]∥Proceedings of the 11th International Workshop on Ontology Matching co-located with the 15th International Semantic Web Conference, Kobe, Japan, 2016: 222-226.
[28]	Jean-Mary Y R, Shironoshita E P, Kabuka M R. Ontology matching with semantic verification[J]. Journal of Web Semantics, 2009, 7(3): 235-251.
[29]	Noessner J, Niepert M, Meilicke C, et al. Leveraging terminological structure for object reconciliation[C]∥Extended Semantic Web Conference, Heraklion, Greece, 2010: 334-348.
[30]	Xu P, Wang Y, Cheng L, et al. Alignment results of SOBOM for OAEI 2010[C]∥Proceedings of the 5th International Workshop on Ontology Matching, Shanghai, China, 2010: 203-211.
[31]	Fernandez S, Marsa-Maestre I, Velasco J R, et al. Ontology alignment architecture for semantic sensor web integration[J]. Sensors, 2013, 13(9): 12581-12604.

相关文章 15

[1]	王兴旺,罗锦宽,张紫洋,迟玉璞,蒋金佑,于美铭. 基于改进型DQN的无人机覆盖路径规划算法[J]. 吉林大学学报(工学版), 2026, 56(9): 2444-2455.
[2]	郭瑞军,范超冉,付明迪. 基于PSO-GRU模型的车辆换道意图识别[J]. 吉林大学学报(工学版), 2026, 56(8): 2084-2094.
[3]	王金鹏,窦顺瑶,王珏,赵昕. MIMO-NOMA系统中基于深度学习CCDNN框架的传输速率提高和能量效率优化算法[J]. 吉林大学学报(工学版), 2026, 56(3): 830-842.
[4]	蔡晓东,龙顺宏,梁焜峻. 基于增强对象学习和注意力网络的视频描述方法[J]. 吉林大学学报(工学版), 2026, 56(2): 516-522.
[5]	岳丹,王崇帅,杨亚廷,聂海涛,揣雅惠. 一种简单自适应光学系统高分辨计算成像技术[J]. 吉林大学学报(工学版), 2026, 56(2): 543-554.
[6]	姚宗伟,陈辰,高振云,靳鸿鹏,荣浩,李学飞,黄虹溥,毕秋实. 基于合成图像数据集的挖掘机关键点识别[J]. 吉林大学学报(工学版), 2026, 56(1): 76-85.
[7]	王琳虹,刘宇阳,刘子昱,鹿应佳,张宇恒,黄桂树. 基于YOLOv5的轻量化桥梁缺陷识别[J]. 吉林大学学报(工学版), 2025, 55(9): 2958-2968.
[8]	艾青林,刘元宵,杨佳豪. 基于MFF-STDC网络的室外复杂环境小目标语义分割方法[J]. 吉林大学学报(工学版), 2025, 55(8): 2681-2692.
[9]	廉敬,张继保,刘冀钊,张家骏,董子龙. 基于文本引导的人脸图像修复[J]. 吉林大学学报(工学版), 2025, 55(8): 2732-2740.
[10]	刘元宁,王星喆,黄子彧,张家晨,刘震. 基于多模态数据融合的胃癌患者生存预测模型[J]. 吉林大学学报(工学版), 2025, 55(8): 2693-2702.
[11]	张宇飞,王丽敏,赵建平,贾智尧,李明洋. 基于中心选择大逃杀优化算法的机器人逆运动学求解[J]. 吉林大学学报(工学版), 2025, 55(8): 2703-2710.
[12]	朴燕,康继元. RAUGAN：基于循环生成对抗网络的红外图像彩色化方法[J]. 吉林大学学报(工学版), 2025, 55(8): 2722-2731.
[13]	车翔玖,李良. 融合全局与局部细粒度特征的图相似度度量算法[J]. 吉林大学学报(工学版), 2025, 55(7): 2365-2371.
[14]	袁靖舒,李武,赵兴雨,袁满. 基于BERTGAT-Contrastive的语义匹配模型[J]. 吉林大学学报(工学版), 2025, 55(7): 2383-2392.
[15]	李文辉,杨晨. 基于对比学习文本感知的小样本遥感图像分类[J]. 吉林大学学报(工学版), 2025, 55(7): 2393-2401.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

案例	异质类型	实体数量	对齐数量
总计		8 347	6 586
101	源本体	109	97
201-202	不同的词汇和语言特征	1 090	970
221-247	不同的结构特征	1 538	1 194
248-262	不同的词汇、语言和结构特征	5 610	4 325

案例	AML	LogMap	LogMapLt	LogMapBio	XMap	本文方法	案例	AML	LogMap	LogMapLt	LogMapBio	XMap	本文方法
101	1.00	0.94	0.56	0.50	0.94	1.00	241	1.00	0.85	0.67	0.66	1.00	1.00
201	1.00	0.96	0.51	0.49	0.93	1.00	246	1.00	0.85	0.67	0.66	1.00	0.93
202	1.00	0.95	0.51	0.50	0.93	1.00	247	1.00	0.85	0.67	0.66	1.00	0.91
221	1.00	0.94	0.56	0.50	0.94	1.00	248	1.00	0.94	0.51	0.47	0.94	1.00
222	1.00	0.00	0.57	0.00	0.94	1.00	249	1.00	0.94	0.81	0.50	0.93	1.00
223	1.00	0.94	0.56	0.50	0.94	0.98	250	1.00	0.90	0.27	0.66	1.00	1.00
224	1.00	0.94	0.83	0.50	0.94	1.00	251	1.00	0.00	0.51	0.00	0.93	1.00
225	1.00	0.94	0.56	0.50	0.94	0.99	252	1.00	0.96	0.51	0.51	0.95	1.00
228	1.00	0.85	0.32	0.66	1.00	1.00	253	1.00	0.97	0.80	0.49	0.93	1.00
232	1.00	0.94	0.83	0.50	0.94	1.00	254	1.00	0.86	0.27	0.64	1.00	1.00
233	1.00	0.85	0.32	0.66	1.00	1.00	257	1.00	0.88	0.64	0.64	1.00	1.00
236	1.00	0.85	0.67	0.66	1.00	1.00	258	1.00	0.00	0.82	0.00	0.93	1.00
237	1.00	0.00	0.84	0.00	0.94	1.00	259	1.00	0.94	0.81	0.49	0.94	1.00
238	1.00	0.94	0.83	0.50	0.94	0.98	261	1.00	0.86	0.27	0.64	1.00	0.96
239	1.00	0.85	0.32	0.66	1.00	0.93	262	1.00	0.87	0.63	0.65	1.00	1.00
240	1.00	0.85	0.32	0.66	1.00	0.91	平均	1.00	0.79	0.58	0.50	0.96	0.99

案例	AML	LogMap	LogMapLt	LogMapBio	XMap	本文方法	案例	AML	LogMap	LogMapLt	LogMapBio	XMap	本文方法
101	0.00	0.96	0.99	0.56	1.00	1.00	241	1.00	1.00	1.00	1.00	1.00	1.00
201	0.28	0.76	0.78	0.44	0.64	0.99	246	1.00	1.00	1.00	1.00	1.00	0.97
202	0.28	0.76	0.78	0.44	0.65	0.70	247	1.00	1.00	1.00	1.00	1.00	0.91
221	0.34	0.96	0.99	0.56	1.00	1.00	248	0.29	0.74	0.78	0.42	0.68	0.70
222	0.33	0.00	0.99	0.00	0.67	1.00	249	0.27	0.75	0.79	0.44	0.72	0.70
223	0.34	0.95	0.99	0.56	1.00	0.97	250	0.83	0.79	0.79	0.79	0.79	0.70
224	0.34	0.95	0.99	0.56	1.00	1.00	251	0.27	0.00	0.78	0.00	0.45	0.70
225	0.34	0.96	0.99	0.56	1.00	0.99	252	0.28	0.76	0.79	0.45	0.65	0.70
228	1.00	1.00	1.00	1.00	1.00	1.00	253	0.27	0.76	0.78	0.44	0.62	0.70
232	0.34	0.95	0.99	0.56	1.00	1.00	254	0.83	0.79	0.79	0.79	0.79	0.70
233	1.00	1.00	1.00	1.00	1.00	1.00	257	0.82	0.79	0.79	0.79	0.79	0.70
236	1.00	1.00	1.00	1.00	1.00	1.00	258	0.28	0.00	0.79	0.00	0.46	0.70
237	0.33	0.00	0.99	0.00	0.67	1.00	259	0.28	0.75	0.78	0.44	0.71	0.70
238	0.34	0.96	0.99	0.56	1.00	0.97	261	0.83	0.79	0.79	0.79	0.79	0.70
239	1.00	1.00	1.00	1.00	1.00	0.97	262	0.83	0.79	0.79	0.79	0.79	0.70
240	1.00	1.00	1.00	1.00	1.00	0.91	平均	0.56	0.77	0.90	0.61	0.83	0.86

案例	AML	LogMap	LogMapLt	LogMapBio	XMap	本文方法	案例	AML	LogMap	LogMapLt	LogMapBio	XMap	本文方法
101	0.00	0.95	0.71	0.52	0.97	1.00	241	1.00	0.92	0.80	0.80	1.00	1.00
201	0.44	0.85	0.62	0.47	0.76	0.99	246	1.00	0.92	0.80	0.80	1.00	0.95
202	0.43	0.85	0.62	0.47	0.77	0.82	247	1.00	0.92	0.80	0.80	1.00	0.91
221	0.51	0.95	0.72	0.52	0.97	1.00	248	0.44	0.82	0.62	0.44	0.79	0.82
222	0.50	0.00	0.72	0.00	0.78	1.00	249	0.43	0.84	0.80	0.47	0.81	0.82
223	0.51	0.94	0.72	0.53	0.97	0.97	250	0.91	0.84	0.40	0.71	0.88	0.82
224	0.51	0.94	0.90	0.53	0.97	1.00	251	0.43	0.00	0.62	0.00	0.61	0.82
225	0.51	0.95	0.72	0.52	0.97	0.99	252	0.43	0.85	0.62	0.48	0.77	0.82
228	1.00	0.92	0.48	0.80	1.00	1.00	253	0.43	0.85	0.79	0.47	0.74	0.82
232	0.51	0.94	0.90	0.53	0.97	1.00	254	0.90	0.82	0.40	0.70	0.88	0.82
233	1.00	0.92	0.48	0.80	1.00	1.00	257	0.90	0.83	0.70	0.71	0.88	0.82
236	1.00	0.92	0.80	0.80	1.00	1.00	258	0.44	0.00	0.80	0.00	0.62	0.82
237	0.50	0.00	0.91	0.00	0.78	1.00	259	0.44	0.84	0.80	0.47	0.81	0.82
238	0.51	0.95	0.90	0.52	0.97	0.97	261	0.90	0.82	0.40	0.70	0.88	0.81
239	1.00	0.92	0.48	0.80	1.00	0.95	262	0.90	0.83	0.70	0.71	0.88	0.82
240	1.00	0.92	0.48	0.80	1.00	0.91	平均	0.66	0.77	0.68	0.54	0.88	0.91