To run the Semantic Mapper, simply navigate to the root folder of TOS-DI and run one of the executables, TOS_DI-win32-x86.exe or TOS_DI-win-x86_64.exe, depending on the architecture of the computer where it is executed. If all the previous steps have been successfully performed, the ICE Data Gateway splash screen should appear, followed by the User License Agreement, and the selection of the first project (see Figure 1).
Figure 1: ICE Data Gateway Selection of First Project
At the selection of the first project step, using the “Manage Connections” button (see Figure 1), it is possible to insert the vf-OS settings of the current session such as the vf-OS username and password and different URLs of services used by the Data Mapper. Clicking “Finish” (Figure 1) will create an empty project and load the Data Mapper.
To create a manufacturing map, a workflow must be created to host it. This is performed by opening the contextual menu on the “Workflows” item in the Data Harmonisation panel (see Figure 2).
Figure 2: Creating a Workflow in ICE Data Gateway
Also input and output schemas need to be imported under the “Schemas” element. The sequence of Figure 3 to Figure 6, show an example of how to import a schema.
Figure 3: Importing a Schema in ICE Data Gateway
Figure 4: Selecting Source File for a new Schema in ICE Data Gateway
Figure 5: Selecting Fields from Imported Schema in ICE Data Gateway
Figure 6: Fine-Tuning Fields for an Imported Schema in ICE Data Gateway
The recently created Schema can be then used on an input component such as tFileInputJSON. To do so, open a workflow and drag and drop a tFileInput component into the frame (see Figure 7).
Figure 7: Adding an Input Component to Canvas in ICE Data Gateway
Once all previous steps have been carried out, ie adding source and target components (eg, tFileInputJSON, tFileOutputXML, etc.) to the canvas, a tSemanticMap component can be placed in between, which will be responsible for mapping the input schema to the output. By opening the advanced UI of the tSemanticMap with a double click, the interface will show a 5 stages transformation, from input schema fields, to input concepts, variables (which are optional), output concepts and output schema fields. On the bottom, once the Reasoner services are developed, suggested concepts will appear for non-assigned schema fields (see Figure 8).
Figure 8: Input to Output Concept Suggestions in ICE Data Gateway
On the main panel, a semantic map is presented, where Input Concepts will be rendered on the left, as green boxes, and Output Concepts will be rendered on the right, as red boxes. Mapping suggestions will be presented as lines of different thickness and tones from light grey to black, visually connecting the Input with the Output Concepts, where a strong similarity has been found by the Reasoner service. Users will be able to draw their own connections between concepts by clicking on the connector of an Input Concept and dragging to the right until reaching the desired Output Concept. Also, Users will be able to accept the suggested mappings by pressing on Apply button. In both cases, affected lines from Input to Output Concepts will be rendered in green. Users will be able also to temporarily ignore any Concept if it is not needed for the data mapping or to revert this action.
To make an Annotated Map available to build a manufacturing process, maps need to be published to the vf-Store. This can be performed by opening the context menu of a Workflow with right click and then selecting Publish Map. This action will open a dialog (see Figure 9).
Figure 9: Publish Map Dialog Before Building the Dockerised Map
Before being able to publish a map, it is necessary to build it from the source files, making use of the button Build (the Publish button will not operate until a build has been performed). This action will trigger TOS-DI Build Job routine, which has been adapted and simplified when called from a vf-OS Asset to show a minimal set of required information to generate the dockerised map (see Figure 10).
Figure 10: Build vf-OS Map Interface with Minimal Set of Fields
The publishing process is then handled by the DHS service, which will receive the generated file and process it to convert the map files into a Docker image, suitable to be published to the vf-Store (see Figure 11). When the mapping has been successfully published, the location in the vf-Store will be displayed (see Figure 12).