Write Related Work

Write Related Work

Purpose

Use Related Work as the evidence layer of the paper's motivation. Organize prior work by mechanisms that matter to the target problem, establish what each research line already supports, and end each line at the exact assumption or capability boundary that motivates the paper.

Required resources and routing

1. Always read the core contract and quality gates in full.
2. Read the MEC domain pack only for mobile edge computing, UAV-MEC, HAP-UAV, vehicular edge computing, dispersed edge computing, or directly adjacent MEC settings.
3. If no domain pack matches, derive the classification axes and terminology from the target manuscript and verified literature. Do not import MEC taxonomies into another field.

Governing distinction

• Introduction: a category-level map that establishes importance, focus, and motivation.
• Related Work: paper-level evidence that validates the map and delimits the unresolved boundary.

Begin from the focus already established in the Introduction. Do not replay the complete motivation chain or pre-write the proposed method.

Workflow

1. Recover the paper's claim chain

Read the Introduction, contribution list, model, method, and relevant bibliography. Extract:

1. the target operating condition;
2. the modeled mechanism that closest work omits or simplifies;
3. the algorithmic difficulty created by that mechanism;
4. the proposed module that addresses each residual difficulty.

Mark unknown items as evidence gaps. Do not infer them from titles.

2. Build a coverage matrix

Every subsection needs one explicit comparison axis. Every boundary must follow from the papers discussed immediately before it. Every major motivation in the paper must have a defensible home in the matrix.

3. Choose motivation-aligned axes

Use the smallest set of axes that mirrors the paper's argument. Typical axes include system architecture, the closest physical or service mechanism, problem formulation, and solver family. Add an axis only when it supports a distinct boundary. A shared keyword or publication chronology is not enough.

Within one subsection, keep the criterion stable and make the categories mutually distinguishable and collectively adequate for the evidence actually covered. Do not claim a universal taxonomy unless the search supports it.

4. Draft category paragraphs

Use:

1. category center sentence defining the shared mechanism;
2. representative evidence grouped by that mechanism;
3. class-level capability;
4. exact retained assumption or missing capability;
5. transition to the next class or paper-level need.

A representative paper normally needs only its setting, distinguishing mechanism or decision, and the result necessary for comparison. Expand only the closest work whose exact assumption establishes the target boundary. The paragraph must remain intelligible without author names.

5. Control claims and citations

Place each citation after the smallest supported proposition. Do not move one paper across subsections merely to fill a taxonomy. When one source genuinely supports two axes, name the distinct analytical role each time.

End each subsection with:

shared capability -> retained assumption -> consequence in the target setting

End the section by combining those residual boundaries into the smallest defensible gap. Prefer bounded formulations such as “these mechanisms have not been jointly evaluated in the reviewed settings.” Avoid “no work,” “the first,” and universal absence claims without a search record that can support them.

6. Run the audit

Apply the core classification, evidence, gap, citation, paragraph, and resolution gates, followed by the selected domain gates. Confirm that:

• headings alone reproduce the literature-side logic;
• opening sentences form a coherent classification story;
• every paper belongs under the declared criterion;
• every limitation follows from described evidence;
• the final gap maps to actual model or method needs;
• the section neither duplicates the Introduction nor leaks method details.

Output constraints

• Preserve citation keys, terminology, and target-venue conventions.
• When drafting, provide the finished section rather than a literature-search diary.
• When auditing, report failures by subsection and paragraph role, then provide a corrected evidence map or revision.
• Do not alter bibliography entries or manuscript sections outside the authorized scope.