amazon-textract questions and possible solutions

[davelina]

Please use the form below, leaving the prefilled data to help us. Thank you.

Page link: amazon-textract

Document link: aws-textract.md

My Issue/Suggestion

Thank you for developing something as complex as document analysis. The full integration of AWS Textract in an adapter form is also very nice. However, I do have a few findings that I am not particularly happy with and a possible extension:

1. The description of how to call everything and the example for the AnalyzeDocument part is nice, but since the module does not have security set up, I either have to add security to the module (which is negated when I re-import the module) OR I have to make a copy of all entities to my own module so that I can manage the permissions. If the second option is the right way to go, I would like to know which entities I need to transfer that are relevant to display on the front end, because the module contains quite a lot of entities.
2. The example given earlier for the AnalyzeDocument section is well thought out, but there is little to no information about the AnalyzeExpense section. Since the module makes extensive use of inheritance, I am again faced with the challenge of determining which entities are relevant to display on the front end, which ones I need, etc.
3. I understand that Java actions are used to process everything. But this is often a black box for Mendix developers. In addition, you have to display everything on the front end and therefore create pages to see a result. What might be nicer and lead to faster results is if the Java action returns a JSON that you could then run through a standard import mapping. Initially, during development, you can then see the results directly in the JSON and you don't have to work everything out in Mendix first. In my opinion, this saves a lot of work for developers and increases the likelihood that people will embrace the adapter.
4. There is now also an ID API for Textract, and it would be nice if this functionality were also adopted in the module.

[katarzyna-koltun-mx]

@davelina Hi Dave,

I've passed the feedback to the dev team, to add to the conversation happening on Slack. I'll keep the feedback open for now while the team decides on the technical aspects.