match-style

Compare original and translation side by side

🇺🇸

Original

English

🇨🇳

Translation

Chinese

Match Style

Aligns a draft to a target voice without touching its substance. Three modes, combinable:

author voice — match the researcher's own prior papers (terminology, rhythm, hedging, structure) so a new draft reads as theirs.
venue register — match the published style of the target venue (via
```
study-exemplars
```
).
co-author merge — make a multi-author draft read in one consistent voice instead of a patchwork.

This is part of the acceptance-maximizing toolkit, alongside

polish-prose

(de-AI-ify + tighten),

benchmark-paper

(venue-fit score),

simulate-reviewers

(content critique), and

preflight-check

(desk-reject defects). Be honest with the user: no rewrite guarantees acceptance. What style alignment buys is consistency, venue fit, and readability — real but bounded contributors to how a paper is received.

在不改变草稿实质内容的前提下，将其与目标风格对齐。支持三种可组合的模式：

作者风格——匹配研究者本人过往论文的风格（术语、句式节奏、措辞委婉度、结构布局），让新草稿读起来极具作者个人特色。
期刊风格——匹配目标期刊已发表文章的风格（通过
```
study-exemplars
```
工具）。
合著风格统一——让多作者合著的草稿风格统一，而非拼凑感十足。

该工具属于“最大化录用率工具集”的一部分，与

polish-prose

（去AI化+精简）、

benchmark-paper

（期刊适配度评分）、

simulate-reviewers

（内容评审）以及

preflight-check

（初审缺陷检查）工具配合使用。请如实告知用户：改写无法保证论文被录用。风格对齐带来的是风格一致性、期刊适配度和可读性——这些是影响论文接收结果的真实但有限的因素。

When to use vs. polish-prose

适用场景对比 polish-prose

```
polish-prose
```
fixes objective problems (AI tells, wordiness, inconsistent terms) against general/venue norms.
```
match-style
```
aligns to a specific target voice you provide (your corpus and/or venue exemplars). Run
```
polish-prose
```
first to clean, then
```
match-style
```
to align.

```
polish-prose
```
针对通用/期刊规范修复客观问题（AI写作痕迹、冗余表述、术语不一致）。
```
match-style
```
则是与你提供的特定目标风格（你的语料库和/或期刊范例）对齐。建议先运行
```
polish-prose
```
清理草稿，再运行
```
match-style
```
进行风格对齐。

Inputs

输入内容

The draft:
```
.tex
```
/
```
.md
```
/ text.
For author voice: paths to the author's own prior papers (
```
.tex
```
/
```
.pdf
```
/
```
.txt
```
). These must be the author's own work — see Guardrails.
For venue register: the target venue id →
```
venues/
```
profile →
```
study-exemplars
```
corpus.
Optional: a glossary of preferred terms/acronyms.

草稿文件：
```
.tex
```
/
```
.md
```
/ 纯文本格式。
若需作者风格：作者本人过往论文的文件路径（
```
.tex
```
/
```
.pdf
```
/
```
.txt
```
格式）。这些必须是作者本人的作品——详见防护规则。
若需期刊风格：目标期刊ID →
```
venues/
```
配置文件 →
```
study-exemplars
```
语料库。
可选内容：偏好术语/缩写词汇表。

Process

操作流程

Build the style signature of the target.
- Author voice: run
```
python3 scripts/style_signature.py --corpus <files...> --out author.json
```
  . It measures terminology frequency, mean/var sentence length, hedging density, connective usage (however/moreover/thus…), passive-voice rate, citation density, section-heading vocabulary.
- Venue register: run the same over the
```
study-exemplars
```
  corpus (open-access only) →
```
venue.json
```
  .

Profile the draft the same way:

python3 scripts/style_signature.py --corpus <draft> --out draft.json

Diff draft vs target(s):
```
python3 scripts/style_signature.py --compare draft.json --against author.json [venue.json] --out style-gap.md
```
. This reports where the draft deviates (e.g. "you use 'we propose' 0×; your prior papers average 3×", "hedging 2× higher than venue exemplars").
Surface conflicts. When author voice and venue register disagree (e.g. your habitual first-person vs a venue that prefers passive), present the trade-off — do not silently pick one.
Align the draft section by section, guided by
```
style-gap.md
```
and the rewrite norms in references/style-dimensions.md. Preserve every number, result, claim, and citation verbatim. Keep a change log.
Verify nothing substantive moved: re-run
```
verify-citations
```
if references were near edits; diff claims/numbers before/after.

构建目标风格特征
- 作者风格：运行
```
python3 scripts/style_signature.py --corpus <files...> --out author.json
```
  。该命令会测量术语使用频率、句子长度均值/方差、措辞委婉度密度、连接词使用情况（however/moreover/thus等）、被动语态使用率、引用密度、章节标题词汇特征。
- 期刊风格：针对
```
study-exemplars
```
  语料库（仅开放获取内容）运行相同命令 → 生成
```
venue.json
```
  。

分析草稿风格特征：运行

python3 scripts/style_signature.py --corpus <draft> --out draft.json

。

对比风格差异：运行
```
python3 scripts/style_signature.py --compare draft.json --against author.json [venue.json] --out style-gap.md
```
。该命令会报告草稿与目标风格的差异（例如：“你使用‘we propose’0次；你过往论文平均使用3次”、“措辞委婉度比期刊范例高2倍”）。
呈现冲突点：当作者风格与期刊风格存在冲突时（例如：作者习惯使用第一人称，而期刊偏好被动语态），需向用户说明权衡选项——不得擅自选择其中一种风格。
逐段对齐草稿风格：根据
```
style-gap.md
```
和 references/style-dimensions.md 中的改写规范进行调整。需完整保留所有数据、研究结果、结论和引用内容。记录变更日志。
验证实质内容未改动：若引用内容附近有编辑操作，需重新运行
```
verify-citations
```
；对比修改前后的结论/数据。

Output

输出内容

```
style-gap.md
```
— the measured deviations (draft vs author and/or venue), each with the numbers behind it.
The aligned draft (or a section-by-section diff for the author to accept).
A short note on any author-vs-venue conflicts and which way each was resolved.

```
style-gap.md
```
——草稿与作者风格和/或期刊风格的量化差异报告，每项差异均附具体数据。
已对齐风格的草稿（或逐段差异对比文件，供作者确认）。
简短说明：作者风格与期刊风格的冲突点及最终解决方案。

Guardrails

防护规则

Only the author's own papers are used for the author-voice corpus. If a user supplies someone else's paper to "write like them", decline the copyright/impersonation request and offer venue-register matching from open-access exemplars instead.
Venue exemplars are fetched on demand from open-access sources and processed transiently (see
```
study-exemplars
```
) — never bundled or stored.
Style only: never alter a technical claim, number, result, or citation. Flag, don't fix, anything that would.
No acceptance promises. Frame outcomes as consistency and venue fit, not a guaranteed decision.
Respect the venue's AI-use disclosure policy (in the venue profile); never use style-matching to evade an AI-detection or disclosure requirement.

仅使用作者本人的论文作为作者风格语料库。若用户提供他人论文并要求“模仿该作者风格”，需拒绝此涉及版权/ impersonation的请求，转而提供基于开放获取范例的期刊风格匹配服务。
期刊范例按需从开放获取源获取并临时处理（详见
```
study-exemplars
```
）——绝不打包或存储。
仅调整风格：绝不得改动技术结论、数据、研究结果或引用内容。若发现可能涉及实质内容修改的情况，需标记出来，而非直接修改。
不得承诺录用。需将成果描述为风格一致性和期刊适配度，而非保证录用。
遵守期刊的AI使用披露政策（见期刊配置文件）；绝不得使用风格对齐工具规避AI检测或披露要求。

Memory

记忆功能

Uses the shared

.paper-memory/

convention in the user's paper directory (full spec:

paper-memory-convention.md

At start: read
```
.paper-memory/profile.yml
```
—
```
writing_preferences
```
records the chosen style signature (author voice, venue register, or merged) and spelling/voice, so align to it instead of re-deciding. Read
```
lessons.md
```
for
```
recurring
```
style deviations to watch for.
At end: append durable findings in the shared format
```
- [YYYY-MM-DD] (match-style | <scope>) deviation -> recommendation
```
(via
```
reflect-and-improve
```
's
```
reflect_log.py append
```
, which dedupes and dates); persist the resolved author-vs-venue conflicts so later runs stay consistent. Cross-draft deviations are
```
recurring
```
; one-offs are
```
this-paper
```
.
Create
```
.paper-memory/
```
on demand if absent and offer to add it to the project
```
.gitignore
```
. It is local-only; never upload it or copy it into this repo.

使用用户论文目录中的共享

.paper-memory/

约定（完整规范：

paper-memory-convention.md

）。

启动时：读取
```
.paper-memory/profile.yml
```
——
```
writing_preferences
```
记录了已选择的风格特征（作者风格、期刊风格或合并风格）以及拼写/风格偏好，需据此进行对齐，无需重新决策。读取
```
lessons.md
```
中的
```
recurring
```
（反复出现的）风格偏差，以便重点关注。
结束时：以共享格式
```
- [YYYY-MM-DD] (match-style | <scope>) deviation -> recommendation
```
添加持久化发现（通过
```
reflect-and-improve
```
的
```
reflect_log.py append
```
命令，该命令会去重并添加日期）；保存已解决的作者风格与期刊风格冲突，确保后续运行保持一致性。跨草稿的偏差标记为
```
recurring
```
；仅单篇草稿存在的偏差标记为
```
this-paper
```
。
若
```
.paper-memory/
```
不存在，需按需创建，并建议用户将其添加至项目
```
.gitignore
```
。该目录仅本地存储；不得上传至本仓库或复制到其他位置。