How to find available .com domains in research preprints
Scientists and engineers name new ideas the moment they have them, and preprint servers publish that work daily, before peer review and long before journalists write about it. A term that appears in several preprints this month may be in product names next year.
That lag is the opportunity. By the time a research term is a headline, its .com is usually gone; while it's still in abstracts, it often isn't.
Where to look
- arXiv: AI new submissions
Artificial intelligence, updated every weekday.
- arXiv: Computation and Language
Language models and linguistics; heavy on new jargon.
- arXiv: Machine Learning
The biggest arXiv category by volume.
- bioRxiv
Biology preprints. It limits automated access, so browse it with the extension.
- medRxiv
Health sciences preprints.
Why this works
Preprints are where a field argues about names. A technique often arrives with two or three competing labels, and over a few months one wins. Scanning listings repeatedly shows you which one is winning, because it starts appearing in papers from groups that didn't invent it.
That's also the honest limit of the method: you're betting on vocabulary, not on the science. A brilliant result with an unpronounceable name is worth nothing here, while a modest idea with a memorable label can carry a domain for years.
Step by step
- 1
Pick the field that matches the kind of names you want: AI and machine learning coin terms fastest, but biology and physics produce durable ones.
- 2
Open the day's 'new submissions' listing for that field, not a single paper. A listing carries dozens of abstracts at once.
- 3
Scan the listing with NameHawk. Terms that several teams use at once rise to the top of the results.
- 4
Repeat over a week. A word that keeps appearing across different groups' papers is being adopted, not just coined.
- 5
Check the finalists' .coms, then read one abstract for each to make sure you understand what the word means.
What NameHawk found
A selection of the words NameHawk flagged as new on these pages when they were scanned on 2026-09-19. Pages change, and so does availability: scan them yourself for what's there today.
arXiv Computation and Language new submissions
- reranker
- diarization
- tokenization
- harmfulness
- auditability
arXiv Machine Learning new submissions
- unlearning
- hypernetwork
- identifiability
- staleness
- regularizer
What makes a good find here
- Short, pronounceable words that describe a technique or property (like 'agentic' or 'unlearning'), not a single team's acronym.
- Terms that show up across several fields' listings: those are escaping their niche.
- Two-word phrases that name a problem people will build products around.
What to watch out for
- Much research vocabulary stays in the lab. A word used by one group for one paper is a weak bet.
- Watch for dataset and model names: they belong to their creators.
- Listings mix in math and markup; NameHawk filters common LaTeX commands, but skip anything that's notation rather than a word.
Common questions
- Do I need to understand the research?
- Enough to know what the word means and who would use it. If you can't explain a term in a sentence, you can't build a site on it either.
- How often should I scan?
- The listings change every weekday. Once or twice a week is enough to see which words repeat, which is the signal that matters.
The other nine methods
- Technical white papers
- Social media and meme trackers
- Crowd-sourced slang dictionaries
- Fan fiction
- Dictionary word-of-the-year lists
- Game wikis and patch notes
- Laws and regulations
- Software release notes
- Job postings