Testing and Refining Templates

Depending on your use case, testing and refining templates can be crucial to the success of your Semaphore project. Sometimes, the out-of-the-box templates and perhaps some tweaking of the Semaphore settings on a manageable number of concepts / labels in the model, will be enough but at other times, you really need to tweak the template(s), as the changes you wish to make are applicable to all applicable concepts / labels of the model, and not only to the handful of them.

Once you have worked out what you need to change in the template(s) or settings you need to iterate through the following:

1. Making the change(s).
2. Re-publish the model (using the adjustments you have made).
3. Re-classify your testing set of documents. You can test classification / extraction using the “Document Analysis” component of Studio for single documents and the “classify.exe”/“classify.py” tool (see Using Bulk Classification: classify.py and classify.exe) for sets of documents.
4. Repeat the above until it is working as required.

Ultimately, you are looking to make as much of the tweaking internal to the templates, leaving only an handful of settings, if any, to be tweaked using the Semaphore settings within the model itself.