Matching methods for semantic interoperability in Product LifecycleManagement.

Semantic Interoperability;Product Lifecycle Management (PLM);Data Matching;Data Translation;Support Vector Machine (SVM);Ontology Matching;Mechanical Engineering;Engineering;Mechanical Engineering

Yeo, IlSrinivas, Lakshmi Y. ;

University of Michigan

关键词: Semantic Interoperability;    Product Lifecycle Management (PLM);    Data Matching;    Data Translation;    Support Vector Machine (SVM);    Ontology Matching;    Mechanical Engineering;    Engineering;    Mechanical Engineering;   

Others  :  https://deepblue.lib.umich.edu/bitstream/handle/2027.42/64796/yeoil_1.pdf?sequence=1&isAllowed=y

瑞士|英语

来源: The Illinois Digital Environment for Access to Learning and Scholarship

【 摘 要 】

Product lifecycle management (PLM) is a business strategy that enables seamless information flow in today;;s collaborative, but distributed product development environment. In such environment, geographically and functionally distributed teams are involved in the development process, and the teams use different software systems with different ways of representing product data.As the product development process gets bigger and complicated, product semantics also needs to be translated in addition to the syntactic information, but ISO 10303, the current industry standard, cannot successfully translate the semantics; this has led to a new approach toward semantics-based product data integration. Semantics-based integration first requires participating domains to use semantic representation of product data. Given the semantic representations, it further requires techniques to determine semantic maps across product representations that will enable semantically correct interoperability of product data, and we propose the enabling techniques in this research. In order to determine semantic maps, we propose a method - Instance-Based Concept Matching (IBCM) that detects 1-to-n maps by exploiting implicit semantics captured in the instances of product models. The use of implicit semantics adds a new dimension in the area of product development, where most of the previous research has focused on using schema or data definition that are explicitly defined. Any single matching method is not enough to determine the semantic maps across the different systems, since each method presents only one view. We propose a method - FEedback Matching Framework with Implicit Training (FEMFIT) to combine the different matching approaches using ranking Support Vector Machine. The method overcomes the need to explicitly train the algorithm before it is used, and minimizes the decision-making load on the domain expert. Finally, we propose a framework to automatically determine the translation rules to enable translation of concepts from one system to another. Even after the semantic maps are obtained, the syntax in the sending system should properly transform to the syntax in the receiving system. We use a graph search method that obtains the overall translation rule as a combination of multiple basic functions. Using such rules, data from one system can be easily translated to another system.