[BE] AI provides possible solutions to the tickets

[AnnaH3003]

Description

The AI should to be trained to already suggests initial approaches to solve the user's issue described in the ticket. Moreover, it should generate a score indicating the certainty of the response's success.

The generated solution should be noted in the form of a comment on the corresponding ticket; the author of the ticket should be called "Talktix". The comment should include following disclaimer in the beginning: "This response was created by artificial intelligence."

Users should be notified when changes, such as a comment, have been made on their ticket via email. The email's content should state:

"Hi there!

an initial comment has been made on your ticket [ID] by our Talktix Artificial Intelligence. Please handle with care, as it is still learning and may not always find the right solution straight away:

[display comment - if ticketing system is in place replace with link to the ticket behind "ticket ID"]

Best regards, Talktix"

Nice to have

• Mini-ticketing system where ticket can be displayed and closed by the user.

Acceptance criteria

• [ ] the answer is to be created without manual effort
• [ ] the answer states a feasible solution to the reported problem
• [ ] the automatic answer contains the disclaimer
• [ ] the user is notified about the comment via mail following the structure stated above

Definition of done (DoD)

• Work products are uploaded to the Github repository and merged to the main branch
• If necessary, the bill of materials section of the planning documents is updated
• Code/feature passes suited tests
• Code is peer-reviewed
• Help documentation (README) is updated
• Fits into our architecture, where possible

DoD general criteria

• Feature has been fully implemented
• Feature has been merged into the mainline
• All acceptance criteria were met
• Product owner approved features
• All tests are passing
• Developers agreed to release

[AnnaH3003]

If training the AI and getting the certainty score is independent of each other, it should be done after another

[AnnaH3003]

Where could we find the data for the AI training? Which data is required?