The FITMAN “SEMed” is a modular oriented tool which provides a high flexible semantic data integration approach. The main objective was to develop a tool which, on the one hand, remove the need for a central data repository or federated schema for all product lifecycle and logistics data, and on the other hand, to introduce a layer of semantics on top of existing syntactic data structure descriptions to avoid semantic integration conflicts. It enables a seamless information exchange over a wide range of heterogeneous systems, standards and de-facto standards.
The SEMed is a mature middleware layer for semantic, virtual interoperability and integration specifically of item-level product lifecycle data. It facilitates a standard-based access to both PLM data, for example through its support for the QLM Messaging Interface and the standard product data and different kinds of common data sources like data bases and file based repositories. It consists of three kinds of modules, namely the “Integration Services”. “Semantic Mediator”and “Wrapper”.
High Level Architecture
Figure 1 presents the generic overview of the SEMed. To setup SEMed, the minimal requirements are that an instanced module for each kind of module is required.
Components of this type realize interfaces to request information by the SEMed. Up to now, there are a graphical user interface and a web service to interact with the Semantic Mediator. Both types of interfaces enable the formulating of SPARQL queries whereby an information request can be resolved.
This component is the core of the SEMed and implements the mediator functionality. In this role, the Semantic Mediator is connected to all data sources via wrapper and has the capabilities to calculate the logical view which results from all connected data sources. For this purpose, it does not need a federated schema or single data model to operate – new data sources can be integrated or removed quickly and flexibly.
Components of this type implement the functionality to link a data source to an instance of the Semantic Mediator. In so doing, the linkage between a data source and a wrapper is done via a configuration file. There is no need of implementing interfaces to bind a data source to a wrapper. In consequence, there are different kinds of wrapper to bind specific kinds of data sources. For example, there is a SQL Wrapper which can be connected to each kind of SQL data base.
All current implemented modules of SEMed are written in Java and are compatible with Java SE Development Kit 7 (32 Bit). The programming language Java was chosen on the one hand for achieving the interoperability between different operating systems and on the other hand for using a common and accepted programming language for web applications.
The modules Semantic Mediator and Wrapper represent its logical view in ontologies.
The Web Ontology Language (OWL-DL) was used for the specification of the ontology which describes the logical view beyond the individual data exchange formats. OWL-DL was chosen for three reasons: first of all, it is standardized by the W3C. Secondly, it was judged to be adequately expressive to cover the semantic description of both the standard exchange formats currently applied and the overarching concepts. Finally, a number of Java libraries and reasoners are readily available for OWL-DL.