Writing Clear and Concise Research Papers

✍️ Submitting to a top CS conference or journal? I just released an open-source, concrete, and opinionated checklist for CS paper writing — designed to prevent desk rejects, improve clarity, and save co-authors and reviewers a lot of pain. I have asked all my lab papers to go through this checklist before submission. 🧠 Inspired by real (painful) examples: 1. Forgot to include a co-author before submission (happened much more frequently than you can imagine) 😬 2. Revealed author identity via GitHub repo metadata 🔍 3. "Novel framework..." + no baselines + one giant equation = 🚫 4. Copy-pasted LLM citation hallucinations that don’t exist 🧨 ✅ The checklist covers many things (if not everything): title, abstract, method, experiments, figures, references, hallucinated citations, and final sanity checks. 📄 English & Chinese versions available. 🌐 GitHub: https://lnkd.in/gaQ85ChY Use it. Share it. Improve it. Save a paper (or a career). #CSResearch #AcademicWriting #MachineLearning #PhDLife #PeerReview #LLM #Reproducibility #OpenScience

No one is waking up at 7am, sipping coffee, thinking, “Wow, I really hope someone explains holistic wealth architecture today.” People want clarity. They want content that feels like a conversation, not a lecture. They want to understand what you’re saying the first time they read it. Write like you're talking to a real person. Not trying to win a Pulitzer. - Use short sentences. - Cut the jargon. - Sound like someone they’d trust with their money, not someone who spends weekends writing whitepapers for fun. Confused clients don’t ask for clarification. They move on. Here’s how to make your content clearer: 1. Ask yourself: Would my mom understand this? If the answer is “probably not,” simplify it until she would. No shade to your mom, she’s just a great clarity filter. 2. Use the “friend test.” Read it out loud. If it sounds weird or overly stiff, imagine explaining it to a friend at lunch. Rewrite it like that. 3. Replace jargon with real words. Say “retirement income you won’t outlive” instead of “longevity risk mitigation strategy.” Your clients are not Googling your vocabulary. 4. Stick to one idea per sentence. If your sentence is doing cartwheels and dragging a comma parade behind it, break it up. 5. Format like you actually want them to read it. Use line breaks. Add white space. Make it skimmable. No one wants to read a block of text the size of a mortgage document. Writing clearly isn’t dumbing it down. It’s respecting your audience enough to make content easy to understand. What’s the worst jargon-filled phrase you’ve seen in the wild? Let’s roast it.

𝐁𝐥𝐮𝐞𝐩𝐫𝐢𝐧𝐭 𝐟𝐨𝐫 𝐒𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐢𝐧𝐠 𝐚 𝐇𝐢𝐠𝐡-𝐈𝐦𝐩𝐚𝐜𝐭 𝐑𝐞𝐬𝐞𝐚𝐫𝐜𝐡 𝐏𝐚𝐩𝐞𝐫: Every section & what to address……… Writing a research paper that stands out requires more than great data → it requires structure, clarity, and impact. ⤷ This comprehensive guide walks you through every section of a high-impact research paper 𝐊𝐞𝐲 𝐒𝐞𝐜𝐭𝐢𝐨𝐧𝐬 & 𝐐𝐮𝐞𝐬𝐭𝐢𝐨𝐧𝐬 𝐭𝐨 𝐀𝐝𝐝𝐫𝐞𝐬𝐬: ✅ Abstract (200–350 words) ⤷ What’s the central problem? → Why is it significant? ⤷ Key results and their implications ✅ Introduction (800–1,200 words): ➺ Open with a hook: bold fact, unresolved question, or provocative statement. ➺ Summarize current advances. ➺ Highlight gaps ➣ State your objective and hypothesis. ✅ Literature Integration (1,200–1,800 words) ⤷ What theories or models frame your study? ⤷ What gaps does your work fill? Why does it matter? ✅ Methods (1,000–1,500 words) ➺ Why is this design the best to address your question? ➺ Detail your population, sampling, tools, and analytical plan. ✅ Results (800–1,500 words) ⤷ Use a narrative flow ⤷ Combine descriptive results with visual elements like charts and tables. ⤷ Quantify findings with key metrics and highlight their relevance. ✅ Discussion (1,200–2,000 words) ➺ Go beyond data ⤷ Connect findings to theory, practice, or policy. ➺ Compare with prior work and transparently state limitations. ➤ Conclusion (300–600 words) ⤷ Restate your findings and their impact in 1–2 sentences. ⤷ End with a call to action: How does this shape future research or practice? ************************** ✅ A structured paper ensures clarity and logical flow. What’s the most challenging part of structuring your research paper? ♻️ Repost to help fellow researchers create high-impact papers! #ResearchWriting #HighImpactPapers #AcademicSuccess #ResearchTips

On ruthlessly slashing content from your paper (or the worst pain is self-inflicted). Too often, in my own work, I find myself wanting to keep the perfect sentence - even though that sentence is no longer central to my paper. Less often, I find that I need to lose a whole section - or at least a review panel tells me that I do. When I receive that request, I use a simple process to decide if the reviewers are right. First, I ask: does this section serve the central argument? If a section doesn’t directly support your research question, hypothesis, or key takeaway, it’s probably a distraction. Just because something is interesting doesn’t mean it’s essential. Tip: Write a one-sentence summary of your paper’s main contribution. Then assess each section or paragraph: does it build toward that contribution? Second, I ask is this a “nice-to-have?” If so, I cut it ruthlessly. Tangents in the literature review, extended background details, or exploratory results may feel valuable but often dilute your argument. Keep your focus tight. Tip: If a section interrupts the flow or distracts from the core story, consider moving it to an appendix or cutting it entirely. Third, I focus on my logic, not just length I don’t remove content just to meet a word count. The goal is clarity, not just brevity. So I focus on making sure the argument still makes sense and flows logically. Tip: Start by trimming redundancy and repetition before cutting depth or nuance. Fourth, I save major cuts in a separate file. Some material might work better in a future paper, talk, or grant proposal. Tip: Use a separate document (e.g., “PaperName_Cuts.docx”) to store anything you remove. Name it topically. That way, you can find it more easily later. Finally, I read the paper like a reviewer, not a writer I know what I'm trying to say—but would a reviewer? Ask colleagues what felt unclear or unnecessary. If they skim a section or get lost, that’s a sign. Tip: If multiple readers question the same part, strongly consider revising or removing it. Final thought: Cutting isn’t about making your paper smaller—it’s about making it stronger, focused, and publishable. And the pain? that's self-inflicted? That is what happens when I don't listen to the reviewers - bc the rejection is swift if I fail to either change the paper or offer an adequate explanation for why I did not change it! Best of luck! #academicwriting

With the CVPR deadline approaching, there are many great tips out there about paper writing. Here's my personal take on how I use LLMs after writing the initial draft (yes, write your draft first!) 🧵 I'll focus on two main aspects of how I use LLMs in my paper writing process: Grammar & language refinement Structure & core message Let's dive in! First up: Grammar & Language 🔍 Obviously, LLMs are great for proofreading, but there's a specific way I use them that I find particularly effective... I ask the LLM to LIST all sentences that need improvement, along with suggested changes and explanations. This is crucial - don't let it make all changes at once! Why list format? Because it lets you evaluate each suggestion individually. You want to maintain your voice and only accept changes that make your sentences clearer and simpler :) Important: I explicitly tell the LLM not to change sentences unless they genuinely need improvement. In my experience, this prevents over-editing and keeps the paper authentic to your style ✍️ Now for the trickier part: Structure & Core Message. Here's where I use LLMs as a preliminary reviewer... After completing my draft, I ask the LLM to: - Summarize the key ideas - Outline the paper's flow This helps me check if what I wrote matches what I intended to convey. The magic happens in the iteration process. I compare the LLM's understanding with my intended message. Often, this reveals gaps between what I thought I wrote and what I actually communicated 🤔 When there's a mismatch, I revise and repeat. This iterative process helps refine both the message and its delivery 🔄 I also ask for structural suggestions. Sometimes, an outside perspective can spot better ways to organize the content. But remember - you're the expert on your research! Key takeaway: Use LLMs as a tool for reflection and refinement, not as the primary writer. Your original draft is the foundation; LLMs help polish and structure. Remember that these are just my personal strategies - find what works for you! And good luck with those CVPR submissions!

This week one of my main goals has been to conduct a final review of a paper prior to journal submission - thus I thought it would be helpful to do a short post about the mental checklist I run through when submitting a paper. (This is a first submission to the target journal, not a submission of a revision; and applies primarily to empirical papers.) 1) Carefully review the title and abstract. Are both informative and accurate? Does the abstract meet word limits (usually 100 or 150 words, depending on the journal)? Note that any errors in the abstract make a hugely negative first impression. Check it again! 2) Check the bibliographic compilation. Careful copy-editing of every entry is not usually necessary at this stage (if you want to do this, AI tools can help) but ensure that there are no missing references, "ADD REFERENCE HERE" notes, "???" compilations in Latex, etc. 3) Review the footnotes. Often, footnotes accumulate during writing as a parking lot for extra notes that someone may think are important or wants to remember. Pruning of footnotes is wise at this stage. Longer or more complex background information is often more appropriately placed in an appendix (where it is more clearly separate, and less distracting) compared to a footnote. 4) Review the exhibits and the notes. Does every exhibit have appropriate notes that are complete and readable? Are the exhibit titles logical, clear, and of generally similar structure? Different people have different preferences, but it is not wise, for example, to have one table named "Results" and one named "Robustness check: Alternate construction of the roads variable." They should be roughly similar in length and structure. A reader who goes straight to the exhibits and reads them alone should be able to understand them and understand the primary story of the paper. 5) Review, quickly, the section and subsection titles. Again, preferences differ - there is no one structure of a paper that is always preferable - but ensure that the titles are logical and internally coherent. 6) Review the acknowledgments and ensure that funders, partners, and others are appropriately acknowledged. If original data was collected, ensure that information about ethical approvals and any pre-registrations is provided (I prefer to provide this in the main text but some provide it an acknowledgments footnote.) 7) Return to the journal requirements and note if there are any other required documents (conflict of interest statements, etc.) Cover letters are generally optional at economics journals and if optional, I usually do not provide them; as editor, I only scan them quickly. The primary goal of a cover letter should be to convey information other than "this is a paper about X", information that can be gleaned from the abstract. For example, if the analysis uses proprietary data, or if there is some important information about the composition of the team. Good luck with your submissions!

Crafting Qualitative Research Questions: The Key to Your Study 🎯 Your qualitative research question is the compass 🧭 of your study. It guides every aspect—from data collection to analysis—and ensures your findings address the core phenomenon you're exploring. Here’s a breakdown of how to design effective qualitative research questions: 1. Purpose-Driven Questions ✨ Start with the intent of your study. Qualitative research seeks to explore, understand, or describe experiences, processes, or phenomena. Use open-ended phrases like: "What are the experiences of..." 🗣️ "How do individuals perceive..." 👀 "Why do people engage in..." ❓ 2. Key Characteristics 🗝️ Open-Ended: Avoid yes/no formats to enable depth. Exploratory: Focus on understanding rather than measuring. Contextual: Embed questions within the specific cultural, social, or situational settings of your study. 3. Common Frameworks 🏗️ Qualitative research questions often align with methodological traditions: Phenomenology: "What is the lived experience of [X]?" 🌅 Grounded Theory: "How do [participants] navigate [phenomenon]?" 🛤️ Ethnography: "What are the cultural practices of [group] in [context]?" 🏘️ Narrative: "How do individuals construct stories about [experience]?" 📖 4. Examples 💡 "What are the factors influencing community trust in healthcare systems? 🏥" "How do teachers adapt to online learning in resource-limited settings? 💻" "What strategies do caregivers use to manage stress? 💆♀️" 5. Iterative Refinement 🔄 As you immerse yourself in the research, refine your questions to reflect new insights and ensure they remain aligned with your study's purpose. 🛠️ Whether you're embarking on your first study or honing expertise, mastering research questions ensures clarity and focus. What qualitative inquiry will you design next? 🤔

The field of AI/ML is moving extremely quickly and as AI is applied to other fields they also accelerate. The result is a deluge of papers. In the past, the number of high-quality interesting papers (i.e. papers that I wanted to read) coming out in a year was pretty manageable if I read one or two papers a weeks. But lately it seems that there are one or two, or more, interesting papers coming out every day. Even for a fast reader that has a strong knowledge foundation, this deluge is hard to keep up with. The result is that many papers just sit forever in my reading list, getting buried until they lose relevance. In many cases the papers are getting ignored because they are excessively long and don't present their contribution clearly and concisely. For example, I just finished was a nearly 40 page paper (by authors I won't name/shame) where the contribution could have been summarized in a couple paragraphs of text and a few lines of math. Instead the description of the contribution was jumbled in with an overview of the field, motivation, and prior art, all of which I already know and don't need to read again. Similarly, the part of the paper with the math didn't get to the point and just show the new idea. Instead there was lots of discussion of alternate ideas that didn't work, redundant motivation, and analysis mixed in. Those things are good to have in a paper, but they should not be obscuring the presentation of the key ideas. The result was that it took an hour to read 40 dense pages to get to something that could have been clearly presented a single page. Ugg! If you're writing papers, make sure your contribution is clear, concise, and separate from motivation, analysis, comparisons, and alternatives. Those things still belong in a good paper, but put them in separate sections. Your paper will be better for it. Most importantly, other researchers will be more likely to read a clearly written concise paper, understand your contribution, and cite your work. I can think of many examples where two people/groups published equivalent ideas at about the same time, but one is highly cited and the other is mostly unknown. Everyone assumes this is because of some bias toward top schools, well-known researchers, or some other unfair reputational preference. That does happen, but more often it's because one paper is well written and the other is not. This isn't really surprising because the top researchers in a field got there by writing good papers, so there is a strong correlation between paper quality and reputation. It's a positive feedback loop. So make your contribution clear and keep it separate from tangential stuff! A fellow expert in your field should be able to flip to the key section, read it in a few minutes, and come away understanding the core of what you did. In a world with new work showing up on arXiv every day, you need to be clear and concise if you want others to pay attention to your work.

Mastering the Research Article – My Step-by-Step Approach One of the most important skills for any scientific writer is knowing how to write a research article. It’s the foundation of what we do—and learning how to structure it well is crucial. Here’s exactly how I approach it: 1️⃣ Start with the figures – Once all figures are finalized, I begin by writing the figure legends. Take this seriously—figures should be self-explanatory. If someone looks only at the figures, they should still understand your story. 2️⃣ Write the Methods – Explain how the results were obtained. 3️⃣ Then the Results – Present what you found. Stay factual and link directly to each figure. 4️⃣ Now the Discussion – This is where you explain what the results mean. Connect your findings to the bigger picture. Compare to existing literature, mention limitations, and offer hypotheses. 5️⃣ Go back to the Introduction – Now that you’ve written the core, you can clearly explain why this study matters. Set up the context and the research question. 6️⃣ Finally: Title, Abstract, and the rest – These come last: the abstract, acknowledgments, conflicts of interest, funding info, etc. Save them for when you have the full picture in place. How do you approach writing a research article? Do you follow a similar order? #ScientificWriting #ResearchArticle #WritingTips #AcademicWriting #ScienceCommunication

Guide to Essay 5 - Commonwealth Shared Scholarship PROMPT Provide a short outline of your proposed study/research. This should be written for an audience that does not have any prior knowledge of the subject, and must not be overly technical. Any abbreviations or specialist terms must be explained. 1️⃣ Consider the reviewer as you write Assume your reviewer has little or no background knowledge in the subject area. This means you should avoid ‘too much’ jargon, technical language, and abbreviations. 👉 Simplify: Your explanation should be clear and accessible. Aim to explain your topic in a way that an intelligent high school student could understand. 2️⃣ Identify and focus on the core of your research. Be clear in defining your research and the central focus of your study 👉Instead of saying "I will research epigenetic modifications," say, "I want to further understand how certain changes in our DNA, which do not alter the genetic code itself, can affect the way our body functions as we age." 3️⃣ Frame the importance of your research Why it matters: Explain why this topic is important in a simple way. Highlight the relevance of your study to real-world challenges. 4️⃣ Outline the Objectives Your Goals: State your research objectives in a straightforward manner. Example: "The goal of my study is to identify markers in our genes that can tell us more about how people age. With these markers, we may be able to create new therapies to keep people healthier as they grow older.” 5️⃣ Describe the methods in layman’s terms Explain your approach: Provide a simple explanation of how you will conduct your research. Use analogies if needed to make it more relatable. Avoid technical jargon: Replace words like "sequencing" with more general descriptions, like “analyzing the DNA to look for changes." Example: "I will be studying different kinds of cells from people of various ages to find changes that might explain why some people develop age-related diseases." 6️⃣ Include potential outcomes and impact expected results: Briefly state what you hope to find. 7️⃣ Explain the broader impact: Connect your findings to a real-world impact. How could your research improve people's lives? Example: This research could help create treatments that slow down the aging process, allowing people to stay healthier longer. 9️⃣ Keep it concise Short and Clear - Make sure each part of your outline is to the point. 👉 Aim to provide just enough information to make your research understandable 👉 Avoid too many details - Your goal is to communicate the essence of your research, not every detail. #share #repost #commonwealth #scholarship