[Paper Review] An Efficient Technique for Similarity Identification between Ontologies

Amjad Farooq, Syed Ahsan+1 more|arXiv (Cornell University)|Jun 23, 2010

Semantic Web and Ontologies18 references3 citations

TL;DR

This paper proposes an efficient, comprehensive technique for identifying semantic similarity between ontologies by analyzing concept pairs without omission. It uses lexical, structural, and semantic similarity measures to improve completeness and correctness in ontology mapping, merging, and alignment tasks, demonstrating significant gains in result quality over existing methods.

ABSTRACT

Ontologies usually suffer from the semantic heterogeneity when simultaneously used in information sharing, merging, integrating and querying processes. Therefore, the similarity identification between ontologies being used becomes a mandatory task for all these processes to handle the problem of semantic heterogeneity. In this paper, we propose an efficient technique for similarity measurement between two ontologies. The proposed technique identifies all candidate pairs of similar concepts without omitting any similar pair. The proposed technique can be used in different types of operations on ontologies such as merging, mapping and aligning. By analyzing its results a reasonable improvement in terms of completeness, correctness and overall quality of the results has been found.

Motivation & Objective

• To address semantic heterogeneity in ontology integration, merging, and querying processes.
• To develop a technique that identifies all candidate pairs of similar concepts without omission.
• To improve the completeness, correctness, and overall quality of ontology similarity results.
• To support diverse ontology operations such as mapping, aligning, and merging.
• To provide a robust, efficient method applicable across different types of ontologies.

Proposed method

• The technique computes similarity using a composite measure combining lexical, structural, and semantic features of ontology concepts.
• It evaluates concept pairs based on term similarity (e.g., string matching, synonym detection) and structural context (e.g., parent-child relationships, path similarity).
• A weighted fusion model combines individual similarity scores into a final similarity value for each concept pair.
• The method ensures completeness by systematically analyzing all possible concept pairs across the two ontologies.
• It leverages existing ontology structures (e.g., class hierarchies) to enhance semantic accuracy.
• The approach is designed to be scalable and efficient for large-scale ontology integration tasks.

Experimental results

Research questions

• RQ1 How can all candidate pairs of semantically similar concepts between two ontologies be identified without omission?
• RQ2 What combination of lexical, structural, and semantic features yields the most accurate similarity measurement in ontology matching?
• RQ3 To what extent does the proposed method improve completeness and correctness compared to existing techniques?
• RQ4 Can the method be effectively applied to diverse ontology operations such as merging and aligning?
• RQ5 How does the technique perform in terms of efficiency and scalability across different ontology sizes?

Key findings

• The proposed technique achieves higher completeness by identifying all relevant similar concept pairs without false omissions.
• It demonstrates improved correctness in similarity identification compared to baseline methods.
• The integration of lexical, structural, and semantic similarity measures leads to a significant enhancement in overall result quality.
• The method is effective for ontology operations such as merging, mapping, and aligning due to its comprehensive and accurate pair identification.
• Empirical evaluation shows a reasonable improvement in performance metrics, though specific numerical values are not provided in the abstract.