研究人员发布用语言模型审计科学文献的完整构建方案
来源标题:Auditing the scientific literature with language models
一套可在小型网页面板上运行的系统方案被公开,用于自动发现已发表研究中的错误与不端行为。该系统按论文计算成本低廉,已经过校准,可由 Claude Code、Codex 等编码智能体直接读取并搭建。方案仅描述方法,文中数字均为构建系统时的校准或成本测量,不涉及任何具体论文或个人的结论。
以下正文同步自 Hacker News 最新,版权归原站所有,已转换为易读排版。
This is a complete recipe for a system that finds errors and misconduct in published research. It is cheap per paper, calibrated, and served to its operators from a small web dashboard. It is written so that a coding agent (Claude Code, Codex or similar) can read it and build a working version without further instructions. It describes methods only. Every number quoted below is a calibration or cost measurement from building the system, not a finding about any paper or person.
┌──────────── corpus (PostgreSQL) ────────────┐ │ works, authorships, persons, institutions, │ │ topics/fields, citations, locations (OA │ │ links), retractions/notices, grants │ └─────────────┬───────────────────────────────┘ query ──► resolve ──► seed set ──► Tier 0 metadata ──► Tier 1 full-text, CPU/GPU ──► Tier 2 cheap read │ │ │ └──── fan-out to co-authors, cited/citing, shared figures/text ▼ Tier 3 verify (sceptic / re-derive / recompute) ──► human adjudication ▼ evidence packets ──► dashboard ──► due process
Suggested stack:
Bibliographic backbone. Load a full OpenAlex snapshot (works, authorships, authors, institutions, topics, citations, locations) into PostgreSQL. It gives you, for free:
Note that the snapshot's abstracts may be empty. Fetch abstracts from the OpenAlex API in batches of about 50 when a classifier needs them, using the anonymous tier.
Identity is the hard part. Author profiles in bibliographic databases merge namesakes and split one person across several ids. Resolve in this order:
For name-based pools (all papers by a named person):
De-duplicate works. Working-paper, preprint and journal versions of one paper often appear as separate works. Drop a work whose normalised title repeats an earlier one, and treat identical files filed under two keys as one paper.
Try routes in order and stop at the first verified hit:
Never use: Sci-Hub, LibGen and similar shadow libraries; SSRN (it refuses scripts); ResearchGate or Academia downloads; paywall or CAPTCHA workarounds; accepting consent banners. Also never take a publisher's typeset version from a third-party site such as a course page or mirror, unless the publisher itself makes it free. Third-party copies of working papers and author drafts are fine.
One shared downloader. Every fetch, including those made by agents, goes through one small script. It:
Verify again after download. Short files are often abstracts, comments or summaries rather than papers, so review anything of 1–4 pages. Recognise and exclude:
Never delete; rename by status: .wrong_file, .duplicate, .ineligible, .unlicensed, .notext. Downstream code must understand these suffixes. The search step should skip papers marked ineligible or duplicate and keep looking for the rest.
Search agents. For misses, hand an LLM agent a written brief and a chunk of about 40 papers. The brief covers:
Use one controlled vocabulary of problems ("sins") so that detectors, findings, costs and dashboards share keys. Record these fields for each sin:
Methods catalogue. Keep a parallel list of about 20 detection methods. Each has a cost per paper and a status, and is mapped to the sins it can see. That method × sin matrix is a useful dashboard page:
Epistemic-quality sins need special care. Their detectors must be structural tests that apply equally to any field or ideology, and they must be calibrated on balanced sets before use.
All of these run on CPU or GPU in seconds per paper. Run them before any model read.
Author-neighbourhood index. The catchable image fraud is usually one group reusing an image across its own papers. For each target paper, index the open-access papers of its first, last and corresponding authors from 5 years before to 3 years after it.
Credit check for cross-paper matches. Treat the later paper as the copier and search its full text for the earlier paper's identifiers, title, or a first-three-author surname near its year. Then classify the match:
For self-uncited pairs, a small model reading the two captions only can say whether they describe the same sample (honest) or different materials or conditions (the fraud pattern).
The tools find copies; they cannot say whether a copy is legitimate. Insets, zooms, shared controls drawn in every panel, channel splits, and main-versus-supplementary repeats are all honest. That judgement needs captions and a reader (§6.4).
For long, number-heavy papers such as economics, policy or epidemiology:
Tier 0 (free): extract the text and split it into front matter (the first ~12%), the conclusion (the last "Conclusion" heading or the last 7%), and numeric sentences from the body. Rank those by count of numbers and keywords ("we estimate", "assume", "total", "rate"), up to about 26k characters. Count overclaim words (rule out, robust to all, unambiguous, prove) and uncertainty words (confidence interval, sensitivity, bounds).
Tier 1 (small model, ~16–24k tokens): returns typed flags, each with a verbatim quote, an impact level (high/medium/low) and a type. Types:
It also returns arithmetic tuples (stated inputs, operation, stated result), which Python recomputes; a mismatch beyond rounding adds to the score. Score = weighted flags + 3 × verified arithmetic mismatches.
Tier 2 (mid model, full text, ~30–40k tokens) on the top ~40% by tier-1 score. It checks each high-impact flag against the whole paper and returns confirmed, refuted or unclear. Use a strict rule: disclosure is not correction. A caveat in a footnote does not cure a headline that ignores it.
For detector hits (image duplicates, plot reuse), use one call per paper with a mid model. It sees the two regions with the matched area boxed, the captions, and the sentences that cite the figures, and it tries hardest to explain each hit away. Only what survives is escalated. This removes most honest repeats cheaply.
A general "verifier" that re-reads a few cited pages is not a precision gain. It un-flags real problems as fast as false ones, especially image findings it cannot see. Verify with the evidence in hand, or not at all.
An agent with tools (fetch, Python, pdftotext) audits one or two papers against a written brief:
Use this to ask whether errors push conclusions toward one side of a policy debate:
Input: an author, paper, institution, department (institution × subfield) or topic. Output: ranked evidence packets and a human review queue, never a bare verdict.
Every node in a graph of papers and people carries a suspicion score and an influence weight. At each step, run the cheapest test with the highest expected yield on the highest-priority node. Escalate only when a cheaper tier fires. Stop when expected confirmed findings per dollar falls below a threshold or the budget runs out.
Resolve the query to a seed set (free). An author maps to their ids and works; a paper to the paper, its authors, references and citers; a topic to its top works and most prolific authors.
Tier 0, metadata only ($0), on the seed and its 1-hop neighbourhood. Covers retractions and notices, author anomalies (one-off authors with no history or affiliation trail), output velocity, journal red flags, citation anomalies, grant cross-checks, reference resolution and interests.
Tier 1, full text, CPU/GPU (~$0), on nodes above a low threshold. All of §5, with the injection scan first.
Tier 2, cheap model read ($0.02–0.20 a unit), where Tier 0 or 1 fired or influence is high. This covers per-paper reads (§6.2/§6.3) and claim checks (§6.6).
Tier 3, verification ($0.20–1 a flag), on Tier 2 flags only. This covers the sceptic pass, re-reading with a different model or prompt, symbolic re-derivation (sympy) for math flags, image forensics and statistics recomputed from reported data. Only confirmed flags proceed. As a reference point, roughly 800 first-pass flags became about 100 checked and about 45 confirmed, and a hand check still found a false positive.
Fan-out, continuous. From any confirmed node, enqueue its neighbours:
Priority = suspicion × influence ÷ expected cost. Misconduct clusters in people and labs, so person edges get the highest prior after a confirmation. Follow the senior author and lab, not students.
Tier 4, rare. Rerun released code and data, request raw data, human expert adjudication.
Expect a blind text read to score fabricated papers well on internal support, because fabrication is designed to look like good evidence. Reliability has to come from the record: later replication, contradiction and retraction.
Re-run the benchmark on every prompt, model or threshold change. Publish recall and false-flag rate per check.
Targeted audits find problems but say nothing about rates. To compare groups (departments, schools of thought, institutions, individuals), pre-register one rule, write it down before drawing, and log every later change with its reason.
Direction: for every problem, record whether fixing it weakens the paper's own thesis, strengthens it, or is neutral. Across a set, the share in the "weakens" direction measures motivated reasoning. A share near 50% means noise; a share far above it means errors that systematically favour the authors' conclusion.
Survival-weighted influence: discount each paper's citations by a weight for each problem found, for example:
Expose the weights as sliders, because they are judgment calls. This works as an exploratory view, but it shows mainly that most papers lose something. The direction measures are more informative.
Merit grade (longer term): extract each paper's 1–5 main claims and score them on four axes, reported as a profile with uncertainty rather than one number:
Net contribution = Σ novelty × importance × reliability − harm, where harm is the uptake of claims later shown false.
Per-field prior. For each subfield, compute:
Combine them with editable weights. A field with a lot of money, low transparency and few detections is where undiscovered problems are likeliest. Replace the prior with measured prevalence from random audits as soon as you have them.
Leads from serial offenders. Fraud is serial and clustered by person. For people with several retractions for reasons your detectors can see (image or data duplication, falsification, fabrication), take their unretracted open-access papers from 3 years before their first such retraction to 3 years after their last, nearest years first. Exclude people whose retractions are all peer-review purges, since many co-authors there are innocent.
Contagion around a case. Score everyone who co-authored with the core people by exposure:
A small HTTP server (Python http.server is enough) serves static pages plus JSON endpoints. It needs no build step. Keep it on a private network and include no authentication secrets in pages.
Measured unit costs (subscription token use; dollar figures are API-equivalent and indicative):
{"key": "...", "model": "small", "tokens": 0, "score": 0.0, "t0": {"overclaim_hits": 0, "uncertainty_hits": 0, "chars": 0}, "t1": {"risk": "low|medium|high", "flags": [{"type": "single_assumption|overclaim|aggregation|baseline|mislabel|arithmetic|other", "impact": "high|medium|low", "quote": "verbatim", "why": "..."}], "arith": [{"inputs": [1.0, 2.0], "op": "sum|ratio|pct|diff|product", "stated": 3.0, "quote": "..."}]}, "arith_checked": [{"implied": 3.0, "stated": 3.0, "rel_err": 0.0}], "verify": {"model": "mid", "flags": [{"i": 0, "verdict": "confirmed|refuted|unclear", "impact": "high", "reason": "..."}]}}
Full audit (per paper)
{"paper": "...", "thesis": "one sentence", "source_urls": [], "headline_claims": [], "external_critiques_used": [], "could_not_access": [], "problems": [{"id": 1, "title": "...", "location": "p. 12, Table 3", "severity": "major|moderate|minor", "kind": "error|debatable", "direction": "overstates|understates|neutral", "thesis_direction": "overstates|understates|neutral", "confidence": "high|medium|low", "claimed_value": "...", "corrected_value_or_null": "...", "reasoning":
正文由 探How 从来源站点 RSS 同步,内容未经改写;遇到排版缺失或需要图片、视频时请以原文为准。