Case context: You are building a system for a legal office that reads a contract draft and produces a full summary memo describing risks, missing clauses, and suggested edits. Your team is unsure whether this should be treated as ordinary classification.

Question: Using the idea of rich-output learning, what kind of learning problem is this, and why?

Sample answer: This is rich-output learning. The system must generate a detailed memo, not just a class label or a single score. If you have many labeled examples of (contract draft, summary memo) pairs, you can train the model end to end to map each input draft to a structured textual output.

Key points:

• Classifies the task as rich-output learning
• Notes that the output is a detailed text memo rather than a single category or number
• Relies on training pairs that match each input draft with the correct memo

Rubric: The student must identify the task as rich-output learning and justify the choice by contrasting the detailed memo output with a simple classification target, using the general translation-style idea of input-output pairs.