File size: 17,351 Bytes
484e3bc |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 |
"""
Structured Event Extraction Pipeline
Converts unstructured intelligence (PDFs, articles, reports) into structured,
timestamped events suitable for:
- Causal graph construction and updates
- Time-series analysis
- Panel data modeling
- Temporal feature engineering
Event schema:
- Timestamp (normalized to UTC)
- Event type (conflict, diplomacy, economic, etc.)
- Actors (countries, organizations)
- Location (geospatial)
- Magnitude/severity
- Source and confidence
- Causal attributes (preconditions, effects)
"""
import re
from datetime import datetime, timezone
from typing import List, Dict, Optional, Tuple, Any
import numpy as np
from dataclasses import dataclass, field
from enum import Enum
import json
class EventType(Enum):
"""Event taxonomy."""
CONFLICT = "conflict"
DIPLOMACY = "diplomacy"
ECONOMIC = "economic"
MILITARY_MOBILIZATION = "military_mobilization"
SANCTIONS = "sanctions"
ALLIANCE_FORMATION = "alliance_formation"
TREATY_SIGNING = "treaty_signing"
PROTEST = "protest"
ELECTION = "election"
COUP = "coup"
TERROR_ATTACK = "terror_attack"
CYBER_ATTACK = "cyber_attack"
TRADE_AGREEMENT = "trade_agreement"
ARMS_DEAL = "arms_deal"
HUMANITARIAN_CRISIS = "humanitarian_crisis"
OTHER = "other"
@dataclass
class GeopoliticalEvent:
"""
Structured geopolitical event.
Attributes
----------
event_id : str
Unique event identifier
timestamp : datetime
Event timestamp (normalized to UTC)
event_type : EventType
Type of event
actors : List[str]
Involved actors (countries, organizations)
target : Optional[str]
Target of action (if applicable)
location : Optional[str]
Geographic location
magnitude : float
Event magnitude/severity (0-1)
confidence : float
Extraction confidence (0-1)
source : str
Source document/article
text : str
Original text describing event
causal_preconditions : List[str]
Identified preconditions
causal_effects : List[str]
Identified effects
metadata : Dict[str, Any]
Additional metadata
"""
event_id: str
timestamp: datetime
event_type: EventType
actors: List[str]
target: Optional[str] = None
location: Optional[str] = None
magnitude: float = 0.5
confidence: float = 0.5
source: str = ""
text: str = ""
causal_preconditions: List[str] = field(default_factory=list)
causal_effects: List[str] = field(default_factory=list)
metadata: Dict[str, Any] = field(default_factory=dict)
def to_dict(self) -> Dict[str, Any]:
"""Convert to dictionary."""
return {
'event_id': self.event_id,
'timestamp': self.timestamp.isoformat(),
'event_type': self.event_type.value,
'actors': self.actors,
'target': self.target,
'location': self.location,
'magnitude': self.magnitude,
'confidence': self.confidence,
'source': self.source,
'text': self.text,
'causal_preconditions': self.causal_preconditions,
'causal_effects': self.causal_effects,
'metadata': self.metadata
}
@classmethod
def from_dict(cls, data: Dict[str, Any]) -> 'GeopoliticalEvent':
"""Load from dictionary."""
data['timestamp'] = datetime.fromisoformat(data['timestamp'])
data['event_type'] = EventType(data['event_type'])
return cls(**data)
class EventExtractor:
"""
Extract structured events from unstructured text.
Uses rule-based patterns and NLP to identify:
- Event mentions
- Temporal expressions
- Actor identification
- Event type classification
"""
def __init__(self):
"""Initialize event extractor."""
self.country_names = self._load_country_names()
self.organization_names = self._load_organization_names()
self.event_patterns = self._compile_event_patterns()
def _load_country_names(self) -> List[str]:
"""Load list of country names."""
# Extended list of countries
return [
'United States', 'USA', 'China', 'Russia', 'India', 'Pakistan',
'Iran', 'North Korea', 'South Korea', 'Japan', 'Germany', 'France',
'United Kingdom', 'UK', 'Israel', 'Saudi Arabia', 'Turkey', 'Egypt',
'Syria', 'Iraq', 'Afghanistan', 'Ukraine', 'Poland', 'Italy', 'Spain',
'Canada', 'Australia', 'Brazil', 'Mexico', 'South Africa', 'Nigeria'
]
def _load_organization_names(self) -> List[str]:
"""Load list of international organizations."""
return [
'NATO', 'UN', 'United Nations', 'EU', 'European Union',
'OPEC', 'ASEAN', 'African Union', 'Arab League', 'G7', 'G20',
'IMF', 'World Bank', 'WTO', 'WHO', 'ICC'
]
def _compile_event_patterns(self) -> Dict[EventType, List[re.Pattern]]:
"""Compile regex patterns for event types."""
patterns = {
EventType.CONFLICT: [
re.compile(r'\b(attack|strike|bomb|missile|war|combat|clash|battle)\b', re.I),
re.compile(r'\b(invasion|offensive|assault|raid)\b', re.I)
],
EventType.DIPLOMACY: [
re.compile(r'\b(negotiation|talk|summit|meeting|dialogue)\b', re.I),
re.compile(r'\b(diplomatic|embassy|ambassador)\b', re.I)
],
EventType.SANCTIONS: [
re.compile(r'\b(sanction|embargo|restriction|ban)\b', re.I)
],
EventType.MILITARY_MOBILIZATION: [
re.compile(r'\b(mobiliz|deploy|troop|force|military)\b', re.I)
],
EventType.ALLIANCE_FORMATION: [
re.compile(r'\b(alliance|partnership|coalition|pact)\b', re.I)
],
EventType.TREATY_SIGNING: [
re.compile(r'\b(treaty|agreement|accord|convention)\b', re.I)
],
EventType.ELECTION: [
re.compile(r'\b(election|vote|ballot|referendum)\b', re.I)
],
EventType.COUP: [
re.compile(r'\b(coup|overthrow|takeover|regime change)\b', re.I)
],
EventType.TERROR_ATTACK: [
re.compile(r'\b(terror|terrorist|extremist|bombing)\b', re.I)
],
EventType.CYBER_ATTACK: [
re.compile(r'\b(cyber|hack|breach|malware|ransomware)\b', re.I)
]
}
return patterns
def extract_events(
self,
text: str,
source: str = "",
default_timestamp: Optional[datetime] = None
) -> List[GeopoliticalEvent]:
"""
Extract events from text.
Parameters
----------
text : str
Input text
source : str
Source identifier
default_timestamp : datetime, optional
Default timestamp if none found
Returns
-------
list
List of extracted events
"""
events = []
# Split into sentences
sentences = self._split_sentences(text)
for i, sentence in enumerate(sentences):
# Detect event type
event_type = self._classify_event_type(sentence)
if event_type != EventType.OTHER:
# Extract actors
actors = self._extract_actors(sentence)
# Extract timestamp
timestamp = self._extract_timestamp(sentence, default_timestamp)
# Extract location
location = self._extract_location(sentence)
# Compute magnitude
magnitude = self._estimate_magnitude(sentence, event_type)
# Create event
event = GeopoliticalEvent(
event_id=f"{source}_{i}",
timestamp=timestamp,
event_type=event_type,
actors=actors,
location=location,
magnitude=magnitude,
confidence=0.7, # Rule-based confidence
source=source,
text=sentence
)
events.append(event)
return events
def _split_sentences(self, text: str) -> List[str]:
"""Split text into sentences."""
# Simple sentence splitting
sentences = re.split(r'[.!?]+', text)
return [s.strip() for s in sentences if len(s.strip()) > 20]
def _classify_event_type(self, text: str) -> EventType:
"""Classify event type using patterns."""
for event_type, patterns in self.event_patterns.items():
for pattern in patterns:
if pattern.search(text):
return event_type
return EventType.OTHER
def _extract_actors(self, text: str) -> List[str]:
"""Extract actor entities."""
actors = []
# Check for countries
for country in self.country_names:
if country.lower() in text.lower():
actors.append(country)
# Check for organizations
for org in self.organization_names:
if org.lower() in text.lower():
actors.append(org)
return list(set(actors)) # Remove duplicates
def _extract_timestamp(
self,
text: str,
default: Optional[datetime] = None
) -> datetime:
"""Extract timestamp from text."""
# Try to find date patterns
date_patterns = [
r'(\d{4})-(\d{2})-(\d{2})', # YYYY-MM-DD
r'(\d{1,2})/(\d{1,2})/(\d{4})', # MM/DD/YYYY
r'(January|February|March|April|May|June|July|August|September|October|November|December)\s+(\d{1,2}),?\s+(\d{4})'
]
for pattern in date_patterns:
match = re.search(pattern, text)
if match:
# Parse date (simplified)
try:
date_str = match.group(0)
# Try multiple formats
for fmt in ['%Y-%m-%d', '%m/%d/%Y']:
try:
return datetime.strptime(date_str, fmt).replace(tzinfo=timezone.utc)
except:
continue
except:
pass
# Default to current time or provided default
return default or datetime.now(timezone.utc)
def _extract_location(self, text: str) -> Optional[str]:
"""Extract location from text."""
# Check for country names as locations
for country in self.country_names:
if country.lower() in text.lower():
return country
return None
def _estimate_magnitude(self, text: str, event_type: EventType) -> float:
"""Estimate event magnitude/severity."""
# Keywords indicating severity
high_severity_words = ['major', 'massive', 'large-scale', 'significant', 'devastating']
low_severity_words = ['minor', 'small', 'limited', 'isolated']
text_lower = text.lower()
if any(word in text_lower for word in high_severity_words):
return 0.8
elif any(word in text_lower for word in low_severity_words):
return 0.3
else:
return 0.5 # Default
class TemporalNormalizer:
"""
Normalize timestamps to consistent format (UTC).
Handles:
- Time zone conversion
- Temporal granularity (day, week, month)
- Missing timestamps (imputation)
"""
@staticmethod
def normalize_to_utc(dt: datetime) -> datetime:
"""
Normalize datetime to UTC.
Parameters
----------
dt : datetime
Input datetime
Returns
-------
datetime
UTC datetime
"""
if dt.tzinfo is None:
# Assume local time
return dt.replace(tzinfo=timezone.utc)
else:
return dt.astimezone(timezone.utc)
@staticmethod
def round_to_day(dt: datetime) -> datetime:
"""Round to start of day."""
return dt.replace(hour=0, minute=0, second=0, microsecond=0)
@staticmethod
def round_to_week(dt: datetime) -> datetime:
"""Round to start of week (Monday)."""
day_of_week = dt.weekday()
days_to_subtract = day_of_week
week_start = dt - datetime.timedelta(days=days_to_subtract)
return TemporalNormalizer.round_to_day(week_start)
class CausalFeatureExtractor:
"""
Extract causal features from events for modeling.
Constructs features suitable for:
- Causal graph learning
- Structural equation modeling
- Time-series forecasting
"""
def __init__(self):
"""Initialize causal feature extractor."""
pass
def extract_features(
self,
events: List[GeopoliticalEvent],
time_window: int = 30
) -> Dict[str, np.ndarray]:
"""
Extract causal features from event sequence.
Parameters
----------
events : list
List of events
time_window : int
Time window in days
Returns
-------
dict
Feature dictionary
"""
import numpy as np
# Sort events by timestamp
sorted_events = sorted(events, key=lambda e: e.timestamp)
# Count events by type
event_counts = {}
for event_type in EventType:
event_counts[event_type.value] = sum(
1 for e in sorted_events if e.event_type == event_type
)
# Actor involvement matrix
all_actors = list(set(actor for e in sorted_events for actor in e.actors))
actor_indices = {actor: i for i, actor in enumerate(all_actors)}
# Event-actor matrix
n_events = len(sorted_events)
n_actors = len(all_actors)
actor_matrix = np.zeros((n_events, n_actors))
for i, event in enumerate(sorted_events):
for actor in event.actors:
if actor in actor_indices:
actor_matrix[i, actor_indices[actor]] = 1
# Temporal features
if sorted_events:
time_deltas = []
for i in range(1, len(sorted_events)):
delta = (sorted_events[i].timestamp - sorted_events[i-1].timestamp).total_seconds() / 86400 # days
time_deltas.append(delta)
mean_time_delta = np.mean(time_deltas) if time_deltas else 0
else:
mean_time_delta = 0
features = {
'event_counts': np.array([event_counts[et.value] for et in EventType]),
'actor_matrix': actor_matrix,
'mean_time_delta': mean_time_delta,
'total_events': n_events,
'unique_actors': n_actors
}
return features
def construct_panel_data(
self,
events: List[GeopoliticalEvent],
actors: List[str],
time_granularity: str = 'day'
) -> Dict[str, Any]:
"""
Construct panel data structure from events.
Panel data format: (actor, time) -> features
Parameters
----------
events : list
List of events
actors : list
List of actors
time_granularity : str
Time granularity ('day', 'week', 'month')
Returns
-------
dict
Panel data structure
"""
import pandas as pd
import numpy as np
# Create time index
if not events:
return {}
sorted_events = sorted(events, key=lambda e: e.timestamp)
start_time = sorted_events[0].timestamp
end_time = sorted_events[-1].timestamp
# Generate time grid
if time_granularity == 'day':
time_index = pd.date_range(start_time, end_time, freq='D')
elif time_granularity == 'week':
time_index = pd.date_range(start_time, end_time, freq='W')
elif time_granularity == 'month':
time_index = pd.date_range(start_time, end_time, freq='M')
else:
raise ValueError(f"Unknown granularity: {time_granularity}")
# Initialize panel
panel = {}
for actor in actors:
panel[actor] = pd.DataFrame(index=time_index, columns=['event_count', 'avg_magnitude'])
panel[actor] = panel[actor].fillna(0)
# Fill panel with events
for event in sorted_events:
for actor in event.actors:
if actor in panel:
# Find closest time point
event_date = pd.Timestamp(event.timestamp)
closest_idx = time_index[np.argmin(np.abs(time_index - event_date))]
panel[actor].loc[closest_idx, 'event_count'] += 1
panel[actor].loc[closest_idx, 'avg_magnitude'] += event.magnitude
# Normalize magnitudes
for actor in actors:
mask = panel[actor]['event_count'] > 0
panel[actor].loc[mask, 'avg_magnitude'] /= panel[actor].loc[mask, 'event_count']
return panel
|