🔬 Research
Academic writing, literature review, paper analysis, methodology design
📚 Literature Review
找某领域最新论文
Search for the 10 most cited papers on [topic] published in the last 2 years from top venues (CHI, CSCW, UIST). Summarize each in 2-3 sentences.
找 Research Gap
I'm reading papers about [topic]. Based on these summaries, identify potential research gaps and underexplored directions: [paste summaries]
生成 Related Work 表格
Organize these papers into a comparison table with columns: Paper, Year, Method, Key Finding, Limitation. Papers: [list]
写 Related Work 段落
Write a Related Work section (~500 words) that synthesizes these papers into 2-3 thematic groups. Papers: [list with summaries]
快速总结一篇论文
Summarize this paper in a structured format: (1) Research Question, (2) Method, (3) Key Findings, (4) Limitations, (5) Relevance to my work on [your topic]. Paper: [paste abstract or content]
✍️ Paper Writing
生成论文 Outline
Create a detailed outline for a CHI/CSCW paper about [topic]. Include all standard sections with 2-3 bullet points of what each should cover.
写 Introduction
Write an Introduction (~800 words) for a paper about [topic]. Follow this structure: (1) broad motivation, (2) specific problem, (3) what we did, (4) contributions list.
写 Method Section
Write a Method section describing: participants (N=[n], recruited via [method]), study design ([between/within]-subjects), procedure, and analysis approach. Details: [provide details]
写 Discussion
Based on these findings: [list findings]. Write a Discussion section that: (1) interprets results, (2) compares with prior work, (3) discusses implications for design, (4) acknowledges limitations.
学术英语润色
Polish this paragraph for academic writing. Improve clarity, flow, and precision while keeping the meaning. Flag any logical gaps: [paste text]
缩减字数
Shorten this section by 30% while keeping all key information. Current word count: [n]. Target: [n]. Text: [paste]
🧪 Experiment Design
细化 Research Questions
My broad topic is [topic]. Help me formulate 3 specific, testable research questions (RQs) that are scoped for a single paper.
设计用户实验
I want to evaluate [system/method] for [task]. Design a within-subjects user study: participants, conditions, tasks, measures (both quantitative and qualitative), and procedure.
设计问卷
Design a post-task questionnaire for evaluating [system] on: usability, usefulness, trust, and satisfaction. Use 7-point Likert scales. Include 2-3 open-ended questions.
写 IRB Protocol
Draft an IRB protocol summary for a study that involves [brief description]. Include: purpose, participants, procedures, risks, data handling, and consent process.
🔍 Paper Review
UIST Paper Review — System Contribution, Presentation & Related Work
You are a senior reviewer for ACM UIST. Please review the following paper with a focus on three key dimensions. Be specific, constructive, and cite page/section/figure numbers where possible.
Dimension 1: System Contribution(系统贡献)
Evaluate whether the paper clearly articulates its system-level contributions:
1. 核心贡献是否清晰? The paper should ideally present at least two distinct contributions (e.g., a novel interaction technique + a system architecture, or a design framework + a working prototype, or a new algorithm + a novel UI paradigm). Are these contributions explicitly stated and easy to identify?
2. 贡献的层次感: Are the contributions at different levels (e.g., conceptual/design level + technical/implementation level)? Or are they redundant and overlapping?
3. 新颖性: For each claimed contribution, is it genuinely novel compared to existing systems? Or is it an incremental combination of known techniques?
4. 技术深度: Is there sufficient technical detail to understand how the system works, not just what it does? Are key design decisions justified?
Evaluation fit: Does the evaluation (if any) actually validate the claimed system contributions? Or is there a mismatch between what is claimed and what is evaluated?
Please list:
- What you believe are the paper's top 1–2 system contributions
- Whether the authors have communicated them effectively
- What is missing or underclaimed/overclaimed
Dimension 2: Presentation(表达与传达)
Evaluate whether the paper helps a reader quickly grasp the key contributions:
1. 前两页的信息密度: Can a reader understand the core problem, approach, and contribution within the first 2 pages (abstract + intro + teaser figure)? Or does the paper bury the key message?
2. Teaser / Overview Figure: Is there a clear system overview or teaser figure that visually communicates the pipeline, interaction flow, or key idea? Is it effective?
3. 故事线(Narrative Flow): Does the paper follow a logical and compelling structure? Is there a clear problem → insight → approach → system → evaluation → findings arc?
4. Figure quality: Are figures self-explanatory? Do they have clear captions? Can someone skimming the paper understand the system just from figures + captions?
5. Writing clarity: Is the writing concise and precise? Are there unnecessary repetitions, vague claims, or jargon without definition?
6. 快速定位能力: If an AC only has 10 minutes to skim this paper, would they be able to identify the key contributions and assessment? If not, what needs to change?
Dimension 3: Related Work(相关工作)
Evaluate the comprehensiveness and currency of the related work:
1. 覆盖面: Does the related work cover all relevant research threads? Are there obvious missing areas or communities (e.g., HCI, NLP, VIS, SE) that should be cited?
2. 对比深度: Does the paper go beyond listing related work and actually compare and differentiate its approach from prior systems? Is there a clear articulation of "what exists vs. what we do differently"?
3. 最新文献(2023–2026): Does the paper cite recent and relevant work from UIST 2023–2025 and CHI 2023–2026? Are there obvious recent papers that are missing?
4. 定位准确性: Does the related work help the reader understand where this paper sits in the research landscape? Or does it feel like a disconnected literature dump?
5. Table or comparison: Would the paper benefit from a comparison table summarizing how it differs from the most closely related 3–5 systems?
Final Summary:
| Dimension | Rating (Weak / Fair / Good / Strong) | Key Issue |
|---|---|---|
| System Contribution | | |
| Presentation | | |
| Related Work | | |
Top 3 actionable suggestions for the authors (prioritized)
🔄 Revision & Rebuttal
回复 Reviewer
Reviewer comment: "[paste comment]". Draft a professional rebuttal response that: (1) thanks the reviewer, (2) addresses the concern directly, (3) describes what we changed, (4) quotes the revised text if applicable.
分析 Review 意见
Here are the reviews for my paper. Categorize each comment as: (a) must-fix, (b) should-address, (c) optional/stylistic. Identify common themes across reviewers. Reviews: [paste]
格式检查
Check this LaTeX source against [ACM/IEEE] formatting requirements. Flag any issues with: margins, font sizes, citation format, figure placement, page limit.
生成 Talk Outline
Create a 15-minute presentation outline for this paper. Structure: motivation (2 min), background (2 min), method (3 min), results (5 min), discussion (2 min), Q&A (1 min). Paper: [summary]
🎯 Self-Critique Workflow
A reusable workflow for stress-testing research ideas before presenting them to advisors, reviewers, or search committees. By pre-rehearsing critiques, external feedback becomes validation rather than attack.
Step 1: Generate the Idea
I have a research idea about [X]. Help me flesh it out into a 1-page description including:
- Motivation: Why does this problem matter?
- Approach: What will I do?
- Expected Contribution: What's new?
- Novelty: How is this different from existing work?
Step 2: Generate the Harshest Review
You are a senior HCI/NLP/SE reviewer known for being extremely tough and specific. Write a detailed review of this research idea with:
- 3 major weaknesses
- 2 minor weaknesses
Be specific, harsh, and realistic. Consider:
- Novelty: Has this been done before? What's truly new?
- Technical depth: Is the approach rigorous enough?
- Evaluation plan: How would you actually validate this?
- Scope: Is this too broad or too narrow?
- Positioning: How does this compare to [specific related work]?
Do NOT be nice. I need to hear the worst.
Step 2b: Domain-Specific — NLP
Now review this as an ACL/EMNLP reviewer. Focus on: dataset quality, baseline comparisons, reproducibility, and whether the NLP contribution is substantial or just an application.
Step 2b: Domain-Specific — SE
Now review this as an ICSE/FSE reviewer. Focus on: practical relevance, scalability, developer study design, and whether the SE community would find this impactful.
Step 2b: Domain-Specific — HCI
Now review this as a CHI reviewer. Focus on: user study rigor, design rationale, theoretical grounding, and whether the contribution goes beyond "we built a system."
⚠️ Step 3: Write Your Own Rebuttal
Do this yourself. Do NOT let AI write it.
For each weakness: ✅ Can defend → Write your response. ❌ Can't defend → This is a real weakness.
The ones you can't rebut are the ones that need fixing.
For each weakness: ✅ Can defend → Write your response. ❌ Can't defend → This is a real weakness.
The ones you can't rebut are the ones that need fixing.
Step 4: Revise the Idea
Based on my rebuttal below, help me revise the original idea to address the weaknesses I couldn't fully defend:
[Paste your rebuttal]
Focus on strengthening the weak points while preserving the core contribution.
Step 5: Search Committee Simulation
You are a search committee member at a top CS department. I just gave a 45-minute job talk based on this work.
What would you ask during Q&A? What concerns would you raise in the closed-door committee discussion afterward? Consider:
- Does this person have a clear research identity?
- Is there a viable 5-year research agenda here?
- Would this person get tenure?
- How does this compare to other candidates on the market?
Step 6: Advisor Meeting Simulation
You are a senior professor who gives high-level feedback but rarely does hands-on work. Critique this idea the way you would in a 1-on-1 meeting: focus on framing, positioning, and whether the story is compelling. Point out what's unclear or unconvincing.
📋 Research TODO
☐ First-pass qualitative coding — eat your own dog food
☐ 论文初稿生成 — related work、methods、results 结构
☐ 研究原型代码 — 80% 的前后端代码可以 AI 写
☐ 数据清洗预处理 — 标准化 pipeline
☐ 提交前 self-review — argument consistency checker
🚫 永远不要交给 AI 的:选题判断、数据的深层解释、导师/合作者关系、论文叙事框架、mentoring。这些是你的不可替代核心。
📊 Data
Data analysis, visualization, statistics, machine learning
🔢 Quantitative Analysis
选择统计方法
I have [describe data: e.g., 2 conditions × 20 participants, measuring task completion time and error rate]. What statistical tests should I use? Explain assumptions to check.
写 Results 段落
Write a Results paragraph reporting these statistics in APA format: [paste stats]. Include effect sizes and confidence intervals.
🗣️ Qualitative Analysis
Thematic Analysis 编码
Here are interview excerpts about [topic]. Perform open coding: identify initial codes, then group them into themes. Excerpts: [paste]
生成 Codebook
Based on these codes and themes, create a codebook table with columns: Theme, Code, Definition, Example Quote. Codes: [list]
📈 Visualization
推荐图表类型
I need to visualize [describe data and comparison]. Suggest the best chart type and provide Python (matplotlib/seaborn) code to create it.
🎨 Creative
Image generation, design concepts, visual prompts
🎨 Design References
React Visualized — Design Style Reference
↗ react.gg/visualizedAn interactive, scroll-driven visual exploration of core React concepts.
- Scroll-driven storytelling — content unfolds as user scrolls
- Dark theme with vibrant accents
- Progressive disclosure — complex topics in digestible sections
- Minimal text, maximum visual
💻 Coding
Programming assistance, debugging, code review, architecture design
✨ No prompts yet — this collection is growing!
✍️ Writing
Creative writing, editing, summarization, translation
✨ No prompts yet — this collection is growing!
🧠 Thinking
Problem solving, brainstorming, critical analysis, decision making
✨ No prompts yet — this collection is growing!