paper-search
Search OpenAlex + arXiv and rank by impact relative to the paper's own field and age —
not raw citation counts, which always favor old papers and always bury the work published
last month. No API key, no auth.
scripts/paper_search.py "lost in the middle long context" # topic search
scripts/paper_search.py "kv cache compression" --since 2025-06 # only recent work
scripts/paper_search.py "agent memory" --fresh 60 # + arXiv, last 60 days
scripts/paper_search.py --after arXiv:2307.03172 --about "position bias" # what built on it
scripts/paper_search.py "raft consensus" --field any --limit 25 # non-CS, or wider
scripts/paper_search.py --selftest # offline; no network
--after takes arXiv:ID, DOI:x, or an OpenAlex Wxxx. --json for machine output;
--help for all flags. Set OPENALEX_MAILTO=<your email> for OpenAlex's polite pool.
Reading the output
Papers are bucketed, best first. The buckets are the point — recency and quality are
independent axes, and collapsing them into one score hides exactly the tradeoff that
matters in a fast-moving field.
| Bucket |
Means |
LANDMARK |
cited fast and far above its field — read this first |
STRONG |
peer-reviewed, comfortably above field average |
RISING |
recent and being picked up quickly |
FRESH+ |
too new to be cited, but credible venue or authors |
OK |
real, unremarkable |
FRESH? |
too new to be cited and unvetted — verify it yourself |
THIN |
uncited preprint, authors with no track record — usually skip |
Each line shows date (age) · citations (velocity) · fwci · [tier] venue · authors h=<max h-index>. FWCI is field-weighted citation impact: 1.0 = exactly the average for that
field and year, so it lets a systems paper with 90 citations correctly outrank an LLM
paper with 300. [TOP] = top-tier venue (NeurIPS/ICML/ACL/OSDI/SOSP/VLDB/…).
Traps
- Citation counts are a lagging indicator. In AI/infra, the paper that matters may be
6 weeks old with zero citations. Never conclude "nothing exists" from a citation-ranked
list — run
--fresh 60 before saying a topic is unexplored.
- FWCI is noise below ~18 months — the expected-citation denominator is near zero, so
7 citations can score FWCI 108. The script ignores it below that age; don't reintroduce
it by reading the raw number off a young paper.
- Preprint date ≠ publication date. A 2024 TACL paper may be a 2023 arXiv paper; the
idea landed 8 months before the venue date, and in a fast field that lead is the story.
Output shows
preprint YYYY-MM when they differ — cite the earlier one for priority.
- Author credibility is the weakest signal here; don't lean on it. OpenAlex conflates
common names ("Kevin Lin": h=75 across 825 works, several people), and
h= is the max
across authors — the value conflation inflates. Prefer the @ Stanford, Berkeley
affiliation beside it. On --fresh arXiv hits both are absent by design: arXiv exposes
no institutions, and name lookup is worse than useless ("Feng Wang" → 4,865 authors, so
you'd attach a stranger's h-index). That is what FRESH? means — open the PDF.
- A landmark is cited by every field.
--after on a famous paper returns medical and
legal applications too; pass --about "<keywords>" to keep the frontier on topic.
--field cs is the default. Pass --field any for anything else, or results look
mysteriously empty. Venue metadata is imperfect regardless — a paper published at EMNLP
may still read [PREPRINT] arXiv, so trust the citation numbers over the tier label.
Workflow for a fast-moving topic
paper_search.py "<topic>" — find the LANDMARK and what is established.paper_search.py --after <landmark-id> --about "<topic>" --since <~12mo ago> — what
built on it since, ranked by impact. This is how you avoid citing a superseded result.paper_search.py "<topic>" --fresh 45 — what dropped in the last few weeks, which
step 1 structurally cannot see.- Read the abstracts, then the two or three papers that actually earned it.
1---2name: paper-search3description: Find and rank research papers by recency and field-adjusted impact. Use when searching literature, recent or superseding work, or evidence for AI/ML/systems claims. Do NOT use for general web search or GitHub repository evaluation.4---56# paper-search78Search OpenAlex + arXiv and rank by **impact relative to the paper's own field and age** —9not raw citation counts, which always favor old papers and always bury the work published10last month. No API key, no auth.1112```bash13scripts/paper_search.py "lost in the middle long context" # topic search14scripts/paper_search.py "kv cache compression" --since 2025-06 # only recent work15scripts/paper_search.py "agent memory" --fresh 60 # + arXiv, last 60 days16scripts/paper_search.py --after arXiv:2307.03172 --about "position bias" # what built on it17scripts/paper_search.py "raft consensus" --field any --limit 25 # non-CS, or wider18scripts/paper_search.py --selftest # offline; no network19```2021`--after` takes `arXiv:ID`, `DOI:x`, or an OpenAlex `Wxxx`. `--json` for machine output;22`--help` for all flags. Set `OPENALEX_MAILTO=<your email>` for OpenAlex's polite pool.2324## Reading the output2526Papers are bucketed, best first. The buckets are the point — **recency and quality are27independent axes**, and collapsing them into one score hides exactly the tradeoff that28matters in a fast-moving field.2930| Bucket | Means |31|---|---|32| `LANDMARK` | cited fast *and* far above its field — read this first |33| `STRONG` | peer-reviewed, comfortably above field average |34| `RISING` | recent and being picked up quickly |35| `FRESH+` | too new to be cited, but credible venue or authors |36| `OK` | real, unremarkable |37| `FRESH?` | too new to be cited **and** unvetted — verify it yourself |38| `THIN` | uncited preprint, authors with no track record — usually skip |3940Each line shows `date (age) · citations (velocity) · fwci · [tier] venue · authors h=<max41h-index>`. **FWCI** is field-weighted citation impact: 1.0 = exactly the average for that42field and year, so it lets a systems paper with 90 citations correctly outrank an LLM43paper with 300. `[TOP]` = top-tier venue (NeurIPS/ICML/ACL/OSDI/SOSP/VLDB/…).4445## Traps4647- **Citation counts are a lagging indicator.** In AI/infra, the paper that matters may be48 6 weeks old with zero citations. Never conclude "nothing exists" from a citation-ranked49 list — run `--fresh 60` before saying a topic is unexplored.50- **FWCI is noise below ~18 months** — the expected-citation denominator is near zero, so51 7 citations can score FWCI 108. The script ignores it below that age; don't reintroduce52 it by reading the raw number off a young paper.53- **Preprint date ≠ publication date.** A 2024 TACL paper may be a 2023 arXiv paper; the54 idea landed 8 months before the venue date, and in a fast field that lead is the story.55 Output shows `preprint YYYY-MM` when they differ — cite the earlier one for priority.56- **Author credibility is the weakest signal here; don't lean on it.** OpenAlex conflates57 common names ("Kevin Lin": h=75 across 825 works, several people), and `h=` is the max58 across authors — the value conflation inflates. Prefer the `@ Stanford, Berkeley`59 affiliation beside it. On `--fresh` arXiv hits both are absent by design: arXiv exposes60 no institutions, and name lookup is worse than useless ("Feng Wang" → 4,865 authors, so61 you'd attach a stranger's h-index). That is what `FRESH?` means — open the PDF.62- **A landmark is cited by every field.** `--after` on a famous paper returns medical and63 legal applications too; pass `--about "<keywords>"` to keep the frontier on topic.64- **`--field cs` is the default.** Pass `--field any` for anything else, or results look65 mysteriously empty. Venue metadata is imperfect regardless — a paper published at EMNLP66 may still read `[PREPRINT] arXiv`, so trust the citation numbers over the tier label.6768## Workflow for a fast-moving topic69701. `paper_search.py "<topic>"` — find the LANDMARK and what is established.712. `paper_search.py --after <landmark-id> --about "<topic>" --since <~12mo ago>` — what72 built on it since, ranked by impact. This is how you avoid citing a superseded result.733. `paper_search.py "<topic>" --fresh 45` — what dropped in the last few weeks, which74 step 1 structurally cannot see.754. Read the abstracts, then the two or three papers that actually earned it.
wilbeibi