Step 2 – Create a MICO worfklow (~40 minutes) Step 2 – Realize that the extractors cannot work together (several minutes of pure terror)
NOTE: Deleted text means that we did it during the first integration attempt with the old platform versions, due to the limitations of the previous API. Don’t try this at home!
The second step for a successful integration consisted of setting up a MICO workflow. The only way of setting it up, was to implement and start the extractors, then to check the broker page and hope that all the connections were displaying correctly. Since the release of the MICO platform version 2, the workflow creation consists of 2 steps:
- Describing the required extractors in a predefined XML structure
- Setting up the workflow by accessing the MICO workflow management webapp
After reading the description of the registration XML in the MICO Volume 4 Combined Deliverable, writing down our XML files for the C++ was pretty straightforward — mainly thanks to the previous design phase. The secret for connecting the three C++ extractors to the Java annotation helper, was to carefully specify their output data:
<output>
<semanticType>
<name>Output XML Annotation</name>
<description>Annotation about the detected tampering in proprietary format</description>
</semanticType>
<dataType>
<mimeType>text/vnd.fhg-audiotamperingdetection-xml</mimeType>
<syntacticType>atp:TamperingDetectionXml</syntacticType>
</dataType>
<location><![CDATA[mmm:hasPart [ mmm:hasAsset / dc:format is text/vnd.fhg-audiotamperingdetection+xml ]]]></location>
</output>
Which had to correspond to the input data of the Java annotation helper:
<input>
<semanticType>
<name>Input XML Annotation</name>
<description>Annotation about a detected tampering in proprietary format</description>
</semanticType>
<dataType>
<mimeType>text/vnd.fhg-audiotamperingdetection-xml</mimeType>
<syntacticType>atp:TamperingDetectionXml</syntacticType>
</dataType>
</input>
After registering the extractors, we could already see them connected, and stored our route:
We could then move forward to the next step, being sure that our set of distributed extractors would have communicated properly… and still without writing a single line of source code!
