Reasons behind identified outcomes

[naregeff]

Description
As a user, I would like to understand why the system retains a given outcome

Scope of Work

• =

Story Points

• [1, 2, 3, 5, 8, 13, 20, 40]

Priority

• Medium

Risk

• [Low, Medium, High]

Demo requirement

1. The system displays likeliest outcomes
2. The user requests to see the reasoning behind the outcome
3. For each outcome, the system lists the inputs used to achieve it
4. For each outcome, the system lists the legal reasoning that was applied

Acceptance Criteria

• The user knows why each decision was taken
• The user understands the influence of his/her inputs on the outputs
• The user understands the legal reasoning that was reproduced by the AI

Documentation, files, diagrams, etc. relating to this feature:
N/A

Changes:

• [Optional]

[arekmano]

Is this a duplicate of #14? Both seem to capture the requirement that the system explain itself.

[naregeff]

This is broader than #14 (which focuses solely on displaying the precedents used), and better articulated as a user story.

[arekmano]

What are the outcomes in the context of this story? Are they predicted outcomes of a judgement (i.e. 500$ punitive damage, 200$ restitutionary damages) or are they avenues of resolution that they system will "recommend" to the user (i.e. you should go to small claims court, you should send a cease and desist letter, etc)?

[naregeff]

Predicted outcomes of the judgment

…

On Nov 2, 2017 5:57 PM, "Arek Manoukian" ***@***.***> wrote: What are the outcomes in the context of this story? Are they predicted outcomes of a judgement (i.e. 500$ punitive damage, 200$ restitutionary damages) or are they avenues of resolution that they system will "recommend" to the user (i.e. you should go to small claims court, you should send a cease and desist letter, etc)? — You are receiving this because you were assigned. Reply to this email directly, view it on GitHub <#123 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/Aexl3Yj6GV0D0GHfrcaDbFwYAv3Rsal_ks5syjrmgaJpZM4QO9xU> .

[arekmano]

Further discussions will be needed concerning this story

[arekmano]

This paper explains how to achieve trust in model predictions: https://arxiv.org/abs/1602.04938

Relevant to the story