File size: 9,859 Bytes
46df5f0 |
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 |
"""
Terminology consistency checker.
Validates:
- Consistent spelling of the same term
- Consistent hyphenation
- Consistent capitalization of technical terms
"""
import re
from typing import List, Dict, Set
from collections import defaultdict
from .base import BaseChecker, CheckResult, CheckSeverity
class ConsistencyChecker(BaseChecker):
"""Check terminology and spelling consistency."""
name = "consistency"
display_name = "Consistency"
description = "Check for inconsistent terminology and spelling"
# Known variant pairs (canonical -> variants)
KNOWN_VARIANTS = {
# Hyphenation variants
'self-supervised': ['self supervised', 'selfsupervised'],
'pre-trained': ['pre trained', 'pretrained'],
'fine-tuned': ['fine tuned', 'finetuned'],
'state-of-the-art': ['state of the art', 'stateoftheart'],
'real-world': ['real world', 'realworld'],
'end-to-end': ['end to end', 'endtoend', 'e2e'],
'large-scale': ['large scale', 'largescale'],
'long-term': ['long term', 'longterm'],
'short-term': ['short term', 'shortterm'],
'multi-task': ['multi task', 'multitask'],
'multi-modal': ['multi modal', 'multimodal'],
'cross-lingual': ['cross lingual', 'crosslingual'],
'zero-shot': ['zero shot', 'zeroshot'],
'few-shot': ['few shot', 'fewshot'],
'in-context': ['in context', 'incontext'],
# American vs British English (comprehensive list)
# -or/-our endings
'color': ['colour'],
'behavior': ['behaviour'],
'favor': ['favour'],
'honor': ['honour'],
'labor': ['labour'],
'neighbor': ['neighbour'],
'rumor': ['rumour'],
'vapor': ['vapour'],
# -ize/-ise endings
'analyze': ['analyse'],
'characterize': ['characterise'],
'generalize': ['generalise'],
'initialize': ['initialise'],
'maximize': ['maximise'],
'minimize': ['minimise'],
'normalize': ['normalise'],
'optimize': ['optimise'],
'organize': ['organise'],
'realize': ['realise'],
'recognize': ['recognise'],
'specialize': ['specialise'],
'standardize': ['standardise'],
'summarize': ['summarise'],
'utilize': ['utilise'],
'visualize': ['visualise'],
'categorize': ['categorise'],
'emphasize': ['emphasise'],
'hypothesize': ['hypothesise'],
'prioritize': ['prioritise'],
'synchronize': ['synchronise'],
# -ization/-isation endings
'generalization': ['generalisation'],
'initialization': ['initialisation'],
'maximization': ['maximisation'],
'minimization': ['minimisation'],
'normalization': ['normalisation'],
'optimization': ['optimisation'],
'organization': ['organisation'],
'realization': ['realisation'],
'regularization': ['regularisation'],
'specialization': ['specialisation'],
'standardization': ['standardisation'],
'summarization': ['summarisation'],
'utilization': ['utilisation'],
'visualization': ['visualisation'],
'categorization': ['categorisation'],
'characterization': ['characterisation'],
'parametrization': ['parametrisation'],
'quantization': ['quantisation'],
# -er/-re endings
'center': ['centre'],
'fiber': ['fibre'],
'meter': ['metre'],
'liter': ['litre'],
# -l-/-ll- (American single, British double)
'modeling': ['modelling'],
'labeled': ['labelled'],
'labeling': ['labelling'],
'traveled': ['travelled'],
'traveling': ['travelling'],
'canceled': ['cancelled'],
'canceling': ['cancelling'],
'signaled': ['signalled'],
'signaling': ['signalling'],
# -og/-ogue endings
'analog': ['analogue'],
'catalog': ['catalogue'],
'dialog': ['dialogue'],
# -ense/-ence endings
'defense': ['defence'],
'license': ['licence'],
'offense': ['offence'],
# Other common differences
'gray': ['grey'],
'artifact': ['artefact'],
'program': ['programme'], # Note: 'program' is standard in computing
'skeptical': ['sceptical'],
'aluminum': ['aluminium'],
# Verb forms
'learned': ['learnt'],
'burned': ['burnt'],
'spelled': ['spelt'],
# Common term variants
'dataset': ['data set', 'data-set'],
'benchmark': ['bench mark', 'bench-mark'],
'baseline': ['base line', 'base-line'],
'downstream': ['down stream', 'down-stream'],
'upstream': ['up stream', 'up-stream'],
'encoder': ['en-coder'],
'decoder': ['de-coder'],
}
# Capitalization variants to track
CAPITALIZATION_TERMS = [
'transformer', 'attention', 'bert', 'gpt', 'lstm', 'cnn', 'rnn',
'encoder', 'decoder', 'embedding', 'softmax', 'sigmoid', 'relu',
]
def check(self, tex_content: str, config: dict = None) -> List[CheckResult]:
results = []
# Remove comments
content = re.sub(r'(?<!\\)%.*$', '', tex_content, flags=re.MULTILINE)
content_lower = content.lower()
# Check for known variant inconsistencies
for canonical, variants in self.KNOWN_VARIANTS.items():
found_forms = []
# Check canonical form
if re.search(rf'\b{re.escape(canonical)}\b', content, re.IGNORECASE):
found_forms.append(canonical)
# Check variants
for variant in variants:
if re.search(rf'\b{re.escape(variant)}\b', content, re.IGNORECASE):
found_forms.append(variant)
if len(found_forms) > 1:
results.append(self._create_result(
passed=False,
severity=CheckSeverity.WARNING,
message=f"Inconsistent spelling: {', '.join(found_forms)}",
suggestion=f"Use '{canonical}' consistently throughout"
))
# Check hyphenated word consistency
hyphen_results = self._check_hyphenation_consistency(content)
results.extend(hyphen_results)
# Check capitalization consistency
cap_results = self._check_capitalization_consistency(content)
results.extend(cap_results)
return results
def _check_hyphenation_consistency(self, content: str) -> List[CheckResult]:
"""Find words that appear both hyphenated and non-hyphenated."""
results = []
# Common terms that should always be hyphenated (exceptions)
ALWAYS_HYPHENATED = {
'state-of-the-art', 'end-to-end', 'real-time', 'real-world',
'fine-tuning', 'fine-grained', 'large-scale', 'small-scale',
'multi-task', 'multi-modal', 'cross-domain', 'cross-lingual',
'self-supervised', 'self-attention', 'co-training', 'pre-training',
'post-processing', 'pre-processing', 'well-known', 'well-defined',
'high-quality', 'low-quality', 'long-term', 'short-term'
}
# Find all hyphenated words
hyphenated = set(re.findall(r'\b([a-z]+-[a-z]+(?:-[a-z]+)*)\b', content, re.IGNORECASE))
for hyph_word in hyphenated:
# Skip if it's a known compound that should always be hyphenated
if hyph_word.lower() in ALWAYS_HYPHENATED:
continue
# Create non-hyphenated version
non_hyph = hyph_word.replace('-', ' ')
combined = hyph_word.replace('-', '')
# Check if non-hyphenated version exists
if re.search(rf'\b{re.escape(non_hyph)}\b', content, re.IGNORECASE):
results.append(self._create_result(
passed=False,
severity=CheckSeverity.INFO,
message=f"Inconsistent hyphenation: '{hyph_word}' vs '{non_hyph}'",
suggestion="Choose one form and use it consistently"
))
elif re.search(rf'\b{re.escape(combined)}\b', content, re.IGNORECASE):
results.append(self._create_result(
passed=False,
severity=CheckSeverity.INFO,
message=f"Inconsistent hyphenation: '{hyph_word}' vs '{combined}'",
suggestion="Choose one form and use it consistently"
))
return results
def _check_capitalization_consistency(self, content: str) -> List[CheckResult]:
"""Check if technical terms have consistent capitalization."""
results = []
for term in self.CAPITALIZATION_TERMS:
# Find all case variations
pattern = re.compile(rf'\b{term}\b', re.IGNORECASE)
matches = pattern.findall(content)
if len(matches) > 1:
# Check if there are mixed capitalizations
unique_forms = set(matches)
if len(unique_forms) > 1:
forms_str = ', '.join(f"'{f}'" for f in unique_forms)
results.append(self._create_result(
passed=False,
severity=CheckSeverity.INFO,
message=f"Inconsistent capitalization: {forms_str}",
suggestion="Use consistent capitalization for technical terms"
))
return results
|