Learning · Community report
Curate top research questions in a field
@DeryaTR_ says Muse, using a tool called Jev, picked the top 100 unanswered immunology research questions from 10,000 literature-grounded candidates in a few minutes, each with cited sources. How the ranking was produced or validated is not shown, so check the citations before using the list.
What you can get
What it is
Ranked question shortlist
A ranked shortlist of open research questions, each with a stated subfield and the papers it was drawn from.
It came back right when
- Every listed question has at least one cited source paper.
- The field or subfield matches what you asked for.
- The count of questions matches what you requested.
- A domain expert has reviewed the list before it's used for anything beyond a first pass.
The checks are editorial, not from the source.
- Define the pool
Field, subfield, and roughly how many candidates to draw from.
- Set the target count
How many top questions you want returned.
- Review with citationsStops for your approval
Each question comes back with its supporting papers.
- Expert reviewStops for your approval
A domain expert checks the shortlist before it's used.
Editorial, not from the source
Works when
- You want a first-pass shortlist from a large pool of candidate questions.
- You have someone with domain expertise to review the result.
- You're using this for literature review, not a final research decision.
Does not work when
- You need the ranking method itself validated before trusting the order.
- The field has little indexed literature for it to draw from.
- You plan to act on the list without expert review.
Curate top research questions: review template
| Item | What it is based on | Status |
|---|---|---|
| [Field/subfield scope] | [Your request] | [Confirmed matches intent] |
| [Ranked questions] | [Returned shortlist] | [Count matches request] |
| [Cited sources] | [Papers listed per question] | [Checked by a domain expert] |
The source case
On X, @DeryaTR_ reports asking Muse, using a tool called Jev, to select the top 100 unanswered questions in immunology from 10,000 literature-grounded candidates, and says it came back a few minutes later with a ranked list and cited sources for each question. The post includes example questions with subfield tags, two numeric scores per question, and supporting paper titles; checked on 2026-10-09, this is one individual's account of a single run, not a description of how the ranking was produced or validated.
A public report by someone outside this library. Not independently reproduced here.
From the source
Source excerpt
I asked @Muse from @Meta to use Jev to select the top 100 unanswered questions in immunology from 10,000 literature-grounded candidates. A few minutes later, it came back with some of the best immunology questions I had ever seen!
Quoted from the original posts. Not written by this guide, and not a result produced here.
Suggested prompt
Bring Field or subfield, Candidate pool size or source material, Number of questions wanted. Acts Produces a ranked list of candidate questions with cited sources. Does not submit, publish, or act on any question itself. Source community report, checked 2026-10-09.
Written by this guide as a starting point, not quoted from the source. Your edits stay in this browser.
Other ways to ask
Editorial rewrites of the brief above, for a different emphasis. Copy them separately.
What you need
Editorial, not from the source
You supply
- Field or subfield
- Candidate pool size or source material
- Number of questions wanted
Connected
- Whatever literature or search tool the ranking draws from
Approved by you
- None beyond generating the list for your review
How to try it
Editorial, not from the source
- Name the field or subfield and roughly how large a candidate pool to draw from.
- State how many top questions you want returned, since a count narrows an otherwise open-ended ranking.
- Ask for the source papers behind each question, not just the question text, so the reasoning can be checked.
- Have someone with domain expertise review the shortlist before treating any question as settled or worth pursuing.
Start from a defined candidate pool
The source report describes a large starting point, 10,000 literature-grounded candidate questions in immunology, narrowed down to a top 100. That scale matters: a ranking is only as useful as the pool it's drawn from, and a vague request to 'find good research questions' gives far less to work with than a defined field, a rough candidate count, and a clear target number. State the field or subfield as specifically as you can, since immunology alone spans dozens of distinct subareas, and the source questions returned span several of them with different subfield tags. The more precisely the starting pool is bounded, the more the resulting shortlist reflects a genuine survey of that area rather than a thin sample dressed up as comprehensive.
Ask for the sources, not just the questions
A ranked list of research questions is far more useful when each one comes with the papers it's grounded in, and the source report describes exactly that: supporting paper titles attached to each example question. That citation trail is what turns a list of interesting-sounding questions into something checkable, letting you or a domain expert trace why a particular question was flagged as open rather than taking the ranking on faith. Ask explicitly for this when requesting a shortlist, since a result without citations is much harder to evaluate. If the sources returned don't seem to support the question as framed, that's worth flagging rather than assuming the citation was chosen correctly.
Treat it as a first pass, not a verdict
The source report shows two numeric scores attached to each question, described only by stage labels, with no explanation of how those numbers were produced. Without that explanation, the scores are a sorting mechanism you can't fully audit, which is a reasonable basis for a first-pass shortlist but not for a final decision about what's worth pursuing. Have someone with real expertise in the field review the list before treating any question as settled, since judging what's genuinely unanswered and important in a research area is exactly the kind of call that benefits from domain knowledge the ranking itself doesn't demonstrate having. Use the output to start a conversation, not to end one.
Open Muse to try it. That is the product's own site; this guide is not part of it.
Limits and confirmation points
Limits to confirm
- This is one report of a single run; nothing here establishes how the ranking method works or how reliable it is across fields.
- The post shows two numeric scores per question without explaining how they were calculated.
- This is a curation aid for literature review, not a substitute for a domain expert's judgment about what is actually worth researching.
Stop and check
Editorial, not from the source
- Before it starts: define the field and how many questions you want.
- After it returns: check that every question has a cited source.
- Before acting on it: have a domain expert review the shortlist.
If it goes wrong
Editorial, not from the source
- The ranking order seems arbitrary
- The source report doesn't explain how the numeric scores are calculated; treat the order as a starting point for expert review, not a settled ranking.
- Some cited sources don't fully support the question
- Check each citation yourself; this directory has not verified that the papers shown in the source report actually support the stated questions.
- The field has limited literature to draw from
- A smaller candidate pool will likely produce a thinner or less confident shortlist; narrow or broaden the field accordingly.
Common questions
How does the ranking actually work?
The source report doesn't explain it. It shows two numeric scores per question without describing how they were calculated, so treat the order as a starting point rather than a validated method.
Can I trust the cited sources without checking them?
No. This directory has not verified that the papers shown in the source report actually support each stated question; check citations yourself before relying on them.
Does this work for any research field?
The source report covers immunology specifically. Fields with less indexed literature to draw from will likely produce a thinner or less confident shortlist.
How many questions can it shortlist?
The source report describes narrowing 10,000 candidates down to a top 100. State your own target number rather than assuming that scale is fixed.
Is this a substitute for expert review?
No. Treat it as a curation aid for a first pass through the literature, and have a domain expert review the shortlist before acting on any of it.
Next
Related to this task
More in Learning
Get recurring task recommendations — ask your own agent to check for relevant updates.