Spaces:
Running
Running
File size: 15,446 Bytes
b5c2bb1 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 | # src/scipeerai/modules/reproducibility_scanner.py
#
# Reproducibility Scanner
# -----------------------
# The reproducibility crisis exists largely because
# researchers cannot access the code, data, and exact
# methods used in published papers.
#
# This module scans paper text for reproducibility
# signals β what is present and what is critically
# missing for independent replication.
import re
from dataclasses import dataclass, field
# ββ data structures βββββββββββββββββββββββββββββββββββββββββββ
@dataclass
class ReproducibilityFlag:
flag_type: str
severity: str
description: str
evidence: str
suggestion: str
@dataclass
class ReproducibilityResult:
# what was found
has_code_link: bool
has_data_link: bool
has_software_versions: bool
has_statistical_software: bool
has_preregistration: bool
has_ethics_statement: bool
has_conflict_statement: bool
has_sample_size_justification: bool
# scoring
reproducibility_score: float # 0.0 = not reproducible, 1.0 = fully
flags: list
risk_level: str
summary: str
# ββ main class ββββββββββββββββββββββββββββββββββββββββββββββββ
class ReproducibilityScanner:
"""
Scans paper text for reproducibility indicators.
Two layers:
1. Presence checks β what good papers SHOULD have
2. Absence flags β what is missing and how serious
Scoring is inverted from other modules:
HIGH reproducibility score = LOW risk.
We report both for clarity.
"""
# code/data sharing signals
CODE_PATTERNS = [
r'github\.com/\S+',
r'gitlab\.com/\S+',
r'bitbucket\.org/\S+',
r'code.*available.*at',
r'code.*provided.*at',
r'source code.*available',
r'scripts.*available',
r'zenodo\.org/\S+',
r'osf\.io/\S+',
r'code ocean',
r'figshare\.com/\S+',
]
DATA_PATTERNS = [
r'data.*available.*at',
r'dataset.*available',
r'data.*deposited',
r'data.*repository',
r'data.*doi',
r'supplementary data',
r'data.*provided',
r'open data',
r'zenodo\.org/\S+',
r'osf\.io/\S+',
r'dryad',
r'figshare',
r'harvard dataverse',
r'data.*upon.*request', # weaker β noted separately
]
SOFTWARE_PATTERNS = [
r'r\s+version\s+\d',
r'python\s+\d+\.\d+',
r'spss\s+version',
r'stata\s+\d+',
r'matlab\s+r\d+',
r'sas\s+version',
r'scipy\s+\d',
r'numpy\s+\d',
r'sklearn\s+\d',
r'tensorflow\s+\d',
r'pytorch\s+\d',
]
STAT_SOFTWARE = [
'r software', 'rstudio', 'spss', 'stata',
'sas', 'matlab', 'python', 'excel', 'graphpad'
]
PREREG_PATTERNS = [
r'pre.?registered',
r'preregistered',
r'clinicaltrials\.gov',
r'osf\.io',
r'aspredicted\.org',
r'registered report',
r'trial registration',
r'isrctn',
r'anzctr',
]
def __init__(self):
self._code_re = [re.compile(p, re.IGNORECASE) for p in self.CODE_PATTERNS]
self._data_re = [re.compile(p, re.IGNORECASE) for p in self.DATA_PATTERNS]
self._sw_re = [re.compile(p, re.IGNORECASE) for p in self.SOFTWARE_PATTERNS]
self._prereg_re = [re.compile(p, re.IGNORECASE) for p in self.PREREG_PATTERNS]
# ββ public method βββββββββββββββββββββββββββββββββββββββββ
def analyze(self, text: str) -> ReproducibilityResult:
"""
Full reproducibility scan.
Returns what is present, what is missing, and risk level.
"""
t = text.lower()
# presence checks
has_code = self._check_patterns(text, self._code_re)
has_data = self._check_patterns(text, self._data_re)
has_sw_version = self._check_patterns(text, self._sw_re)
has_stat_sw = any(sw in t for sw in self.STAT_SOFTWARE)
has_prereg = self._check_patterns(text, self._prereg_re)
has_ethics = self._has_ethics_statement(t)
has_conflict = self._has_conflict_statement(t)
has_n_justify = self._has_sample_size_justification(t)
# build flags for what is missing
flags = []
flags.extend(self._flag_missing_code(has_code, t))
flags.extend(self._flag_missing_data(has_data, t))
flags.extend(self._flag_missing_software(has_sw_version, has_stat_sw, t))
flags.extend(self._flag_missing_prereg(has_prereg, t))
flags.extend(self._flag_missing_ethics(has_ethics, t))
flags.extend(self._flag_data_on_request(text))
# reproducibility score: percentage of key items present
checklist = [
has_code, has_data, has_sw_version,
has_stat_sw, has_prereg, has_ethics,
has_conflict, has_n_justify
]
repro_score = sum(checklist) / len(checklist)
# risk is inverse of reproducibility
risk_score = round(1.0 - repro_score, 3)
risk_level = self._get_risk_level(risk_score)
return ReproducibilityResult(
has_code_link=has_code,
has_data_link=has_data,
has_software_versions=has_sw_version,
has_statistical_software=has_stat_sw,
has_preregistration=has_prereg,
has_ethics_statement=has_ethics,
has_conflict_statement=has_conflict,
has_sample_size_justification=has_n_justify,
reproducibility_score=round(repro_score, 3),
flags=flags,
risk_level=risk_level,
summary=self._write_summary(
repro_score, risk_level, flags,
has_code, has_data
),
)
# ββ presence detectors ββββββββββββββββββββββββββββββββββββ
def _check_patterns(self, text: str, patterns: list) -> bool:
return any(p.search(text) for p in patterns)
def _has_ethics_statement(self, text: str) -> bool:
markers = [
'ethics committee', 'institutional review board',
'irb approval', 'ethics approval', 'ethical approval',
'helsinki declaration', 'informed consent',
'ethical clearance', 'ethics board'
]
return any(m in text for m in markers)
def _has_conflict_statement(self, text: str) -> bool:
markers = [
'conflict of interest', 'competing interest',
'no conflict', 'declare no', 'disclose',
'funding source', 'financial disclosure'
]
return any(m in text for m in markers)
def _has_sample_size_justification(self, text: str) -> bool:
markers = [
'power analysis', 'sample size calculation',
'power calculation', 'statistical power',
'a priori power', 'effect size calculation',
'g*power', 'gpower'
]
return any(m in text for m in markers)
# ββ flag generators βββββββββββββββββββββββββββββββββββββββ
def _flag_missing_code(self, has_code: bool, text: str) -> list:
"""
Code absence is critical for computational papers.
We detect if the paper is computational first.
"""
flags = []
is_computational = any(w in text for w in [
'algorithm', 'code', 'software', 'script',
'simulation', 'model', 'neural network',
'machine learning', 'deep learning'
])
if is_computational and not has_code:
flags.append(ReproducibilityFlag(
flag_type="missing_code_availability",
severity="high",
description=(
"Computational study does not provide a link to "
"source code or analysis scripts. Independent "
"replication is not possible without this."
),
evidence="Computational methods detected β no code link found",
suggestion=(
"Deposit code on GitHub/GitLab/Zenodo and include "
"the URL in a 'Code Availability' section."
),
))
return flags
def _flag_missing_data(self, has_data: bool, text: str) -> list:
flags = []
has_empirical = any(w in text for w in [
'dataset', 'data', 'sample', 'participants',
'measurements', 'observations', 'collected'
])
if has_empirical and not has_data:
flags.append(ReproducibilityFlag(
flag_type="missing_data_availability",
severity="high",
description=(
"Empirical study does not specify where raw data "
"can be accessed. Results cannot be independently verified."
),
evidence="Empirical data detected β no data availability statement found",
suggestion=(
"Deposit raw data in a repository (OSF, Zenodo, Dryad, "
"Harvard Dataverse) and include a Data Availability statement."
),
))
return flags
def _flag_missing_software(
self, has_versions: bool, has_sw: bool, text: str
) -> list:
flags = []
is_quantitative = any(w in text for w in [
'statistical', 'analysis', 'test', 'regression',
'anova', 'correlation', 't-test', 'chi-square'
])
if is_quantitative and not has_versions:
flags.append(ReproducibilityFlag(
flag_type="missing_software_versions",
severity="medium",
description=(
"Statistical analysis performed but software name and "
"version number not reported. Results may not replicate "
"across different software versions."
),
evidence="Statistical analysis detected β no software version found",
suggestion=(
"Specify the exact software and version used "
"(e.g., 'R version 4.3.1', 'Python 3.10.12 with "
"scikit-learn 1.3.0')."
),
))
return flags
def _flag_missing_prereg(self, has_prereg: bool, text: str) -> list:
flags = []
is_clinical_or_experimental = any(w in text for w in [
'clinical trial', 'randomized', 'experiment',
'intervention', 'treatment', 'placebo',
'hypothesis', 'we predicted', 'we hypothesized'
])
if is_clinical_or_experimental and not has_prereg:
flags.append(ReproducibilityFlag(
flag_type="missing_preregistration",
severity="medium",
description=(
"Experimental or clinical study with no preregistration "
"detected. Without preregistration, it is difficult to "
"distinguish confirmatory from exploratory analyses."
),
evidence="Experimental design detected β no preregistration link",
suggestion=(
"For future studies, preregister hypotheses on OSF "
"(osf.io) or ClinicalTrials.gov before data collection."
),
))
return flags
def _flag_missing_ethics(self, has_ethics: bool, text: str) -> list:
flags = []
involves_humans = any(w in text for w in [
'participants', 'subjects', 'patients', 'volunteers',
'respondents', 'human', 'children', 'adults'
])
if involves_humans and not has_ethics:
flags.append(ReproducibilityFlag(
flag_type="missing_ethics_statement",
severity="high",
description=(
"Human participants study with no ethics approval "
"or IRB statement detected. This is required by "
"most journals and funding bodies."
),
evidence="Human participants detected β no ethics statement found",
suggestion=(
"Include an Ethics Statement specifying the approving "
"body, protocol number, and that informed consent was obtained."
),
))
return flags
def _flag_data_on_request(self, text: str) -> list:
"""
'Data available upon request' is widely considered
a reproducibility red flag β studies show that
most such requests are never fulfilled.
"""
flags = []
if re.search(
r'data.*available.*upon.*request|'
r'data.*available.*on.*request|'
r'available.*from.*corresponding.*author',
text, re.IGNORECASE
):
flags.append(ReproducibilityFlag(
flag_type="data_available_on_request",
severity="medium",
description=(
"'Data available upon request' is a reproducibility "
"risk. Research shows that over 80% of such requests "
"go unfulfilled or receive no response."
),
evidence="'Data available upon request' language detected",
suggestion=(
"Deposit data in a public repository instead. "
"This increases citation rates and research trust."
),
))
return flags
# ββ scoring βββββββββββββββββββββββββββββββββββββββββββββββ
def _get_risk_level(self, risk_score: float) -> str:
if risk_score >= 0.7: return "critical"
elif risk_score >= 0.4: return "high"
elif risk_score >= 0.2: return "medium"
return "low"
def _write_summary(
self,
repro_score: float,
risk_level: str,
flags: list,
has_code: bool,
has_data: bool,
) -> str:
pct = round(repro_score * 100)
if not flags:
return (
f"Reproducibility score: {pct}%. "
f"All key reproducibility indicators detected."
)
missing = []
if not has_code: missing.append("code")
if not has_data: missing.append("data")
high = sum(1 for f in flags if f.severity == "high")
med = sum(1 for f in flags if f.severity == "medium")
parts = []
if high: parts.append(f"{high} critical gap{'s' if high > 1 else ''}")
if med: parts.append(f"{med} concern{'s' if med > 1 else ''}")
return (
f"Reproducibility score: {pct}%. "
f"Flagged {', '.join(parts)}. "
f"Risk level: {risk_level.upper()}."
) |