Scholarlink

<p>Sensor ontologies play a crucial role in the Digital Economy by enabling seamless integration and utilization of diverse sensor data, but their practical application is hindered by heterogeneity arising from semantic discrepancies and structural variations. Ontology Matching (OM) addresses this challenge by aligning semantically equivalent entities, yet existing methods face limitations in balancing multiple objectives and ensuring interpretability. To tackle these issues, this paper proposes an Interpretable Semi-Supervised Entity Matching (ISEM), which leverages a novel Multi-Objective Evolutionary Algorithm (MOEA) that emphasizes interpretability, and dynamic user guidance. In particular, the designed ISEM integrates user preferences to enhance alignment accuracy, employs a multi-objective optimization model to balance the completeness and correctness of matching results, and utilizes an improved MOEA for high-quality alignments. Experimental results on OAEI’s Conference dataset and ten pairs of real sensor OM tasks demonstrate that the proposed approach achieves superior matching performance, advancing sensor knowledge integration for the Digital Economy.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Interpretable semi-supervised sensor knowledge integration for advancing digital economy

  • 期刊论文

  • 发表日期：

    2025年9月16日

查看全文
  • Chang-Wei Yang

摘要

Sensor ontologies play a crucial role in the Digital Economy by enabling seamless integration and utilization of diverse sensor data, but their practical application is hindered by heterogeneity arising from semantic discrepancies and structural variations. Ontology Matching (OM) addresses this challenge by aligning semantically equivalent entities, yet existing methods face limitations in balancing multiple objectives and ensuring interpretability. To tackle these issues, this paper proposes an Interpretable Semi-Supervised Entity Matching (ISEM), which leverages a novel Multi-Objective Evolutionary Algorithm (MOEA) that emphasizes interpretability, and dynamic user guidance. In particular, the designed ISEM integrates user preferences to enhance alignment accuracy, employs a multi-objective optimization model to balance the completeness and correctness of matching results, and utilizes an improved MOEA for high-quality alignments. Experimental results on OAEI’s Conference dataset and ten pairs of real sensor OM tasks demonstrate that the proposed approach achieves superior matching performance, advancing sensor knowledge integration for the Digital Economy.