Abstract
Efficient information sharing and exchanging are becoming increasingly important, but model heterogeneity hinders communication and collaboration between different enterprises. To solve this problem, a semantic mapping method based on an ontology structure graph (OSG) is proposed. This paper analyzes the requirements for model semantic interoperability, defines the concept, and summarizes semantic clashes between enterprise models. Furthermore, it presents a formalization of semantic mapping rules and a detailed mapping process—including model element choice, 3-level semantic matching, and rule execution—supported by an OSG-based representation schema and semantic similarity functions.
Introduction: The Challenge of Model Heterogeneity
In modern enterprise environments, collaboration is often stifled by "model heterogeneity." Different organizations or departments frequently use disparate data structures and terminologies to describe identical concepts. For instance, one system might record a "Phone Number" as a string, while another uses a long integer; one might label a resource as "Material" while another calls it "Raw Resource." These discrepancies lead to "semantic clashes," which act as significant roadblocks to automated data exchange and interoperability.
Addressing Semantic Clashes
To systematically resolve these misunderstandings, the authors categorize semantic clashes into three distinct levels: Data, Schema, and Relation. This taxonomy allows for a more engineering-focused approach to resolving conflicts rather than relying on ad-hoc fixes.
The inclusion of the "Relation Level" is a key innovation, as many existing methods overlook how complex relationships (like the link between a "car" and a "motor-car") contribute to semantic confusion.
The Semantic Mapping Process (SMP)
The paper proposes a structured six-step workflow to bridge the gap between unique enterprise models and a shared domain ontology. This process utilizes an Ontology Structure Graph (OSG)—a directed graph representing concepts, attributes, and rules—as a universal "Rosetta Stone."
The workflow consists of:
- Model Element Choice: Identifying the specific data elements intended for sharing.
- Ontology Element Selection: Matching these elements to "gold standard" terms in the OSG.
- 3-Level Semantic Matching: Analyzing the data, schema, and relations to identify specific clashes.
- Semantic Mapping Rules (SMR) Generation: Creating formal logic to translate between models.
- Evaluation: Validating the rules to ensure no data or meaning is lost.
- Execution: Running the rules to facilitate actual data exchange.
Key Insights and Theoretical Foundation
A central component of this method is the Semantic Similarity Function. This mathematical framework allows the system to quantify the closeness of a model element to an ontology element, assigning scores (e.g., 1.0 for identity, 0.8 for synonyms) to automate the matching process. By mapping every enterprise model to a central OSG rather than directly to every other model, the system achieves a scalable hub-and-spoke architecture for interoperability.
Conclusion and Future Work
While this paper provides a robust theoretical framework and a formal taxonomy for conflict resolution, the authors note that further work is required to refine the underlying matching algorithms. Future research will focus on large-scale real-world validation and exploring machine learning techniques to further automate the generation of semantic mapping rules.
Summary updated