Introduction Section Writing

Objective

Build a reviewer-facing funnel: meaningful task -> unmet target or failure case -> root technical bottleneck -> proposed solution -> why it works -> evidence and contributions.

Backward Planning Questions

Answer these before writing:

1. What technical problem is solved, and why is there no established solution?
2. What exactly is new: task, metric, pipeline, module, design choice, finding, or insight?
3. Why should this contribution solve the bottleneck?
4. Which experiments or analyses later support the main promises?

Forward Section Roles

Role	Required Content	Failure Signal
Task/application opening	Define task or application value and target requirement	Opens with generic importance but no task boundary
Prior-work limitation	Summarize representative methods by mechanism and limitation	Paper-by-paper list with no technical reason
Technical bottleneck	State the unresolved challenge and root cause	Gap is a marketing claim, not a mechanism
Proposed solution	Introduce pipeline or key insight with a figure anchor if present	Method appears before the challenge is clear
Why it works	Explain technical advantage in bounded terms	Claims novelty without mechanism
Evidence/contributions	Preview experiments and enumerate contributions	Contributions do not map to later evidence

Introduction Patterns

Use one pattern that matches the paper:

• Task first: for niche tasks; define input/output before applications.
• Application first: for familiar tasks; open with use cases and target requirements.
• General-to-specific: for a new setting inside a familiar area.
• Open with challenge: when the unresolved failure case is immediately understandable.

Anti-Pattern

Do not write the story as "a naive baseline exists, then we patch it." That makes the contribution look obvious. Instead, explain the real technical bottleneck that prior families of methods still cannot resolve.

Claim-Evidence Closure

Every contribution sentence should map forward:

Intro promise -> Method mechanism -> Experiment evidence -> Conclusion answer

If a contribution has no corresponding experiment or analysis, either mark missing evidence or weaken the claim.