File size: 39,028 Bytes
bb53f16 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 |
"""
Module: agents.oxossi
Codinome: Oxóssi - Caçador de Fraudes
Description: Agent specialized in fraud detection and tracking with precision hunting capabilities
Author: Anderson H. Silva
Date: 2025-01-25
License: Proprietary - All rights reserved
"""
import asyncio
from datetime import datetime, timedelta
from typing import Any, Dict, List, Optional, Tuple
from dataclasses import dataclass
from enum import Enum
import numpy as np
import pandas as pd
from pydantic import BaseModel, Field as PydanticField
from src.agents.deodoro import BaseAgent, AgentContext, AgentMessage, AgentResponse, AgentStatus
from src.core import get_logger
from src.core.exceptions import AgentExecutionError, DataAnalysisError
logger = get_logger(__name__)
class FraudType(Enum):
"""Types of fraud detected by Oxóssi."""
BID_RIGGING = "bid_rigging"
PRICE_FIXING = "price_fixing"
PHANTOM_VENDOR = "phantom_vendor"
INVOICE_FRAUD = "invoice_fraud"
KICKBACK_SCHEME = "kickback_scheme"
CONFLICT_OF_INTEREST = "conflict_of_interest"
MONEY_LAUNDERING = "money_laundering"
FALSE_CLAIMS = "false_claims"
PAYROLL_FRAUD = "payroll_fraud"
PROCUREMENT_FRAUD = "procurement_fraud"
class FraudSeverity(Enum):
"""Severity levels for fraud detection."""
LOW = "low"
MEDIUM = "medium"
HIGH = "high"
CRITICAL = "critical"
@dataclass
class FraudIndicator:
"""Individual fraud indicator detected."""
indicator_type: str
description: str
confidence: float # 0.0 to 1.0
evidence: List[Dict[str, Any]]
risk_score: float # 0.0 to 10.0
@dataclass
class FraudPattern:
"""Fraud pattern detection result."""
fraud_type: FraudType
severity: FraudSeverity
confidence: float
indicators: List[FraudIndicator]
entities_involved: List[str]
estimated_impact: float
recommendations: List[str]
evidence_trail: Dict[str, Any]
class OxossiAgent(BaseAgent):
"""
Oxóssi - The Fraud Hunter Agent.
Specialized in detecting and tracking various types of fraud in government
contracts and financial transactions with hunter-like precision.
Cultural reference: Oxóssi is the Yoruba deity of the hunt, known for
his precision, focus, and ability to track targets through any terrain.
"""
def __init__(self):
"""Initialize Oxóssi agent."""
super().__init__(
agent_id="oxossi",
name="Oxóssi",
description="Fraud detection specialist with precision tracking capabilities",
capabilities=[
"fraud_detection",
"pattern_recognition",
"entity_relationship_analysis",
"financial_forensics",
"risk_assessment",
"evidence_tracking"
]
)
# Fraud detection thresholds
self.fraud_thresholds = {
"bid_similarity": 0.85, # Similarity threshold for bid rigging
"price_deviation": 2.5, # Standard deviations for price fixing
"vendor_activity": 0.1, # Minimum activity for legitimate vendor
"invoice_anomaly": 0.7, # Confidence threshold for invoice fraud
"relationship_strength": 0.6 # Threshold for suspicious relationships
}
# Known fraud patterns database
self.fraud_patterns = self._initialize_fraud_patterns()
def _initialize_fraud_patterns(self) -> Dict[str, Any]:
"""Initialize known fraud pattern templates."""
return {
FraudType.BID_RIGGING: {
"indicators": [
"identical_bid_amounts",
"sequential_bid_numbers",
"rotation_pattern",
"last_minute_withdrawals"
],
"min_confidence": 0.7
},
FraudType.PRICE_FIXING: {
"indicators": [
"uniform_price_increases",
"identical_pricing_across_vendors",
"price_stability_anomaly",
"market_share_stability"
],
"min_confidence": 0.65
},
FraudType.PHANTOM_VENDOR: {
"indicators": [
"no_physical_address",
"recent_registration",
"single_contract_only",
"no_web_presence",
"shared_contact_info"
],
"min_confidence": 0.75
},
FraudType.INVOICE_FRAUD: {
"indicators": [
"duplicate_invoices",
"sequential_invoice_numbers",
"rounded_amounts",
"unusual_descriptions",
"timing_anomalies"
],
"min_confidence": 0.7
}
}
async def process(self, message: AgentMessage, context: AgentContext) -> AgentResponse:
"""Process fraud detection request."""
start_time = datetime.now()
try:
self.status = AgentStatus.THINKING
logger.info(f"Oxóssi starting fraud hunt: {message.content}")
# Extract data for analysis
data = message.data
if not data:
raise AgentExecutionError("No data provided for fraud detection")
# Determine analysis type
if "contracts" in data:
fraud_patterns = await self._analyze_contract_fraud(data["contracts"], context)
elif "transactions" in data:
fraud_patterns = await self._analyze_transaction_fraud(data["transactions"], context)
elif "vendors" in data:
fraud_patterns = await self._analyze_vendor_fraud(data["vendors"], context)
elif "invoices" in data:
fraud_patterns = await self._analyze_invoice_fraud(data["invoices"], context)
else:
# Comprehensive fraud analysis
fraud_patterns = await self._comprehensive_fraud_analysis(data, context)
# Generate fraud report
report = self._generate_fraud_report(fraud_patterns)
# Calculate processing time
processing_time = (datetime.now() - start_time).total_seconds()
self.status = AgentStatus.IDLE
return AgentResponse(
success=True,
data={
"fraud_analysis": report,
"patterns_detected": len(fraud_patterns),
"high_risk_entities": self._identify_high_risk_entities(fraud_patterns),
"total_estimated_impact": sum(p.estimated_impact for p in fraud_patterns),
"processing_time": processing_time
},
metadata={
"agent": self.name,
"analysis_timestamp": datetime.now().isoformat(),
"confidence_score": self._calculate_overall_confidence(fraud_patterns)
}
)
except Exception as e:
self.status = AgentStatus.ERROR
logger.error(f"Oxóssi fraud detection failed: {str(e)}")
return AgentResponse(
success=False,
data={"error": str(e)},
metadata={"agent": self.name, "error_type": type(e).__name__}
)
async def _analyze_contract_fraud(
self, contracts: List[Dict[str, Any]], context: AgentContext
) -> List[FraudPattern]:
"""Analyze contracts for fraud patterns."""
fraud_patterns = []
# Check for bid rigging
bid_rigging = await self._detect_bid_rigging(contracts)
if bid_rigging:
fraud_patterns.extend(bid_rigging)
# Check for price fixing
price_fixing = await self._detect_price_fixing(contracts)
if price_fixing:
fraud_patterns.extend(price_fixing)
# Check for phantom vendors
phantom_vendors = await self._detect_phantom_vendors(contracts)
if phantom_vendors:
fraud_patterns.extend(phantom_vendors)
return fraud_patterns
async def _detect_bid_rigging(self, contracts: List[Dict[str, Any]]) -> List[FraudPattern]:
"""Detect bid rigging patterns."""
patterns = []
# Group contracts by bidding process
bidding_groups = {}
for contract in contracts:
bid_id = contract.get("bidding_process_id")
if bid_id:
if bid_id not in bidding_groups:
bidding_groups[bid_id] = []
bidding_groups[bid_id].append(contract)
# Analyze each bidding group
for bid_id, group_contracts in bidding_groups.items():
if len(group_contracts) < 3: # Need multiple bids to detect rigging
continue
indicators = []
# Check for identical or very similar bid amounts
bid_amounts = [c.get("bid_amount", 0) for c in group_contracts]
if self._check_bid_similarity(bid_amounts):
indicators.append(FraudIndicator(
indicator_type="identical_bid_amounts",
description="Multiple bids with identical or nearly identical amounts",
confidence=0.8,
evidence=[{"bid_amounts": bid_amounts}],
risk_score=7.5
))
# Check for bid rotation patterns
if self._check_bid_rotation(group_contracts):
indicators.append(FraudIndicator(
indicator_type="rotation_pattern",
description="Vendors appear to be taking turns winning bids",
confidence=0.75,
evidence=[{"pattern": "rotation_detected"}],
risk_score=8.0
))
if indicators:
patterns.append(FraudPattern(
fraud_type=FraudType.BID_RIGGING,
severity=FraudSeverity.HIGH,
confidence=max(ind.confidence for ind in indicators),
indicators=indicators,
entities_involved=[c.get("vendor_name", "Unknown") for c in group_contracts],
estimated_impact=sum(c.get("contract_value", 0) for c in group_contracts) * 0.1,
recommendations=[
"Investigate bidding process for collusion",
"Review communications between vendors",
"Check for common ownership or management"
],
evidence_trail={"bidding_process_id": bid_id}
))
return patterns
async def _detect_phantom_vendors(self, contracts: List[Dict[str, Any]]) -> List[FraudPattern]:
"""Detect phantom vendor patterns."""
patterns = []
# Analyze vendors
vendor_contracts = {}
for contract in contracts:
vendor_id = contract.get("vendor_id")
if vendor_id:
if vendor_id not in vendor_contracts:
vendor_contracts[vendor_id] = []
vendor_contracts[vendor_id].append(contract)
for vendor_id, vendor_contracts_list in vendor_contracts.items():
indicators = []
# Check for single contract vendors
if len(vendor_contracts_list) == 1:
indicators.append(FraudIndicator(
indicator_type="single_contract_only",
description="Vendor has only one contract in the system",
confidence=0.6,
evidence=[{"contract_count": 1}],
risk_score=5.0
))
# Check for recent registration before contract
for contract in vendor_contracts_list:
vendor_reg_date = contract.get("vendor_registration_date")
contract_date = contract.get("contract_date")
if vendor_reg_date and contract_date:
days_diff = (pd.to_datetime(contract_date) - pd.to_datetime(vendor_reg_date)).days
if days_diff < 30: # Registered less than 30 days before contract
indicators.append(FraudIndicator(
indicator_type="recent_registration",
description="Vendor registered shortly before receiving contract",
confidence=0.7,
evidence=[{"days_before_contract": days_diff}],
risk_score=6.5
))
if indicators:
patterns.append(FraudPattern(
fraud_type=FraudType.PHANTOM_VENDOR,
severity=FraudSeverity.HIGH if len(indicators) > 1 else FraudSeverity.MEDIUM,
confidence=max(ind.confidence for ind in indicators),
indicators=indicators,
entities_involved=[vendor_contracts_list[0].get("vendor_name", f"Vendor_{vendor_id}")],
estimated_impact=sum(c.get("contract_value", 0) for c in vendor_contracts_list),
recommendations=[
"Verify vendor physical existence",
"Check vendor registration details",
"Validate vendor tax records",
"Conduct site visits if necessary"
],
evidence_trail={"vendor_id": vendor_id}
))
return patterns
async def _detect_price_fixing(self, contracts: List[Dict[str, Any]]) -> List[FraudPattern]:
"""Detect price fixing patterns."""
patterns = []
# Group contracts by category/type
category_contracts = {}
for contract in contracts:
category = contract.get("category", "Unknown")
if category not in category_contracts:
category_contracts[category] = []
category_contracts[category].append(contract)
for category, contracts_list in category_contracts.items():
if len(contracts_list) < 5: # Need sufficient data for price analysis
continue
# Extract prices and vendors
price_data = []
for contract in contracts_list:
price_data.append({
"vendor": contract.get("vendor_name"),
"price": contract.get("unit_price", contract.get("contract_value", 0)),
"date": contract.get("contract_date")
})
df = pd.DataFrame(price_data)
if df.empty:
continue
indicators = []
# Check for identical pricing across vendors
price_groups = df.groupby("vendor")["price"].mean()
if len(price_groups) > 1:
price_variance = price_groups.std() / price_groups.mean()
if price_variance < 0.05: # Less than 5% variance
indicators.append(FraudIndicator(
indicator_type="identical_pricing_across_vendors",
description="Multiple vendors have nearly identical pricing",
confidence=0.75,
evidence=[{"price_variance": price_variance}],
risk_score=7.0
))
# Check for synchronized price increases
df["date"] = pd.to_datetime(df["date"])
df = df.sort_values("date")
# Group by time periods and check for uniform increases
monthly_avg = df.groupby(pd.Grouper(key="date", freq="M"))["price"].mean()
if len(monthly_avg) > 3:
price_changes = monthly_avg.pct_change().dropna()
if (price_changes > 0.05).sum() > 1: # Multiple 5%+ increases
if price_changes.std() < 0.02: # Very similar increase rates
indicators.append(FraudIndicator(
indicator_type="uniform_price_increases",
description="Synchronized price increases across time periods",
confidence=0.7,
evidence=[{"price_increases": price_changes.to_list()}],
risk_score=6.5
))
if indicators:
patterns.append(FraudPattern(
fraud_type=FraudType.PRICE_FIXING,
severity=FraudSeverity.HIGH,
confidence=max(ind.confidence for ind in indicators),
indicators=indicators,
entities_involved=df["vendor"].unique().tolist(),
estimated_impact=df["price"].sum() * 0.15, # Estimated 15% overcharge
recommendations=[
"Compare prices with market rates",
"Investigate vendor communications",
"Review industry pricing trends",
"Consider antitrust investigation"
],
evidence_trail={"category": category}
))
return patterns
async def _analyze_transaction_fraud(
self, transactions: List[Dict[str, Any]], context: AgentContext
) -> List[FraudPattern]:
"""Analyze financial transactions for fraud."""
patterns = []
# Check for money laundering patterns
laundering = await self._detect_money_laundering(transactions)
if laundering:
patterns.extend(laundering)
# Check for kickback schemes
kickbacks = await self._detect_kickback_schemes(transactions)
if kickbacks:
patterns.extend(kickbacks)
return patterns
async def _analyze_vendor_fraud(
self, vendors: List[Dict[str, Any]], context: AgentContext
) -> List[FraudPattern]:
"""Analyze vendor data for fraud patterns."""
patterns = []
# Check for shell companies
for vendor in vendors:
indicators = []
# Check for shared addresses
address = vendor.get("address")
if address:
same_address_vendors = [v for v in vendors if v.get("address") == address]
if len(same_address_vendors) > 2:
indicators.append(FraudIndicator(
indicator_type="shared_address",
description="Multiple vendors share the same address",
confidence=0.8,
evidence=[{"vendors": [v.get("name") for v in same_address_vendors]}],
risk_score=7.0
))
# Check for shared contact information
phone = vendor.get("phone")
email = vendor.get("email")
if phone or email:
same_contact_vendors = [
v for v in vendors
if (phone and v.get("phone") == phone) or (email and v.get("email") == email)
]
if len(same_contact_vendors) > 1:
indicators.append(FraudIndicator(
indicator_type="shared_contact_info",
description="Multiple vendors share contact information",
confidence=0.85,
evidence=[{"vendors": [v.get("name") for v in same_contact_vendors]}],
risk_score=8.0
))
if indicators:
patterns.append(FraudPattern(
fraud_type=FraudType.PHANTOM_VENDOR,
severity=FraudSeverity.HIGH,
confidence=max(ind.confidence for ind in indicators),
indicators=indicators,
entities_involved=[vendor.get("name", "Unknown")],
estimated_impact=0, # Unknown until contracts analyzed
recommendations=[
"Investigate vendor relationships",
"Verify vendor legitimacy",
"Check for common ownership"
],
evidence_trail={"vendor_id": vendor.get("id")}
))
return patterns
async def _analyze_invoice_fraud(
self, invoices: List[Dict[str, Any]], context: AgentContext
) -> List[FraudPattern]:
"""Analyze invoices for fraud patterns."""
patterns = []
# Check for duplicate invoices
invoice_hashes = {}
for invoice in invoices:
# Create hash of key invoice attributes
hash_key = f"{invoice.get('vendor_id')}_{invoice.get('amount')}_{invoice.get('date')}"
if hash_key in invoice_hashes:
patterns.append(FraudPattern(
fraud_type=FraudType.INVOICE_FRAUD,
severity=FraudSeverity.HIGH,
confidence=0.9,
indicators=[FraudIndicator(
indicator_type="duplicate_invoices",
description="Duplicate invoice detected",
confidence=0.9,
evidence=[{
"invoice_1": invoice_hashes[hash_key],
"invoice_2": invoice.get("invoice_number")
}],
risk_score=8.5
)],
entities_involved=[invoice.get("vendor_name", "Unknown")],
estimated_impact=invoice.get("amount", 0),
recommendations=[
"Reject duplicate invoice",
"Investigate vendor billing practices",
"Review payment controls"
],
evidence_trail={"invoice_numbers": [invoice_hashes[hash_key], invoice.get("invoice_number")]}
))
else:
invoice_hashes[hash_key] = invoice.get("invoice_number")
# Check for sequential invoice patterns
vendor_invoices = {}
for invoice in invoices:
vendor_id = invoice.get("vendor_id")
if vendor_id:
if vendor_id not in vendor_invoices:
vendor_invoices[vendor_id] = []
vendor_invoices[vendor_id].append(invoice)
for vendor_id, vendor_invoice_list in vendor_invoices.items():
# Sort by invoice number if numeric
try:
sorted_invoices = sorted(
vendor_invoice_list,
key=lambda x: int(x.get("invoice_number", "0").replace("-", ""))
)
# Check for suspicious sequential patterns
invoice_numbers = [int(inv.get("invoice_number", "0").replace("-", "")) for inv in sorted_invoices]
if len(invoice_numbers) > 5:
# Check if all invoices are perfectly sequential
expected_sequence = list(range(min(invoice_numbers), max(invoice_numbers) + 1))
if invoice_numbers == expected_sequence:
patterns.append(FraudPattern(
fraud_type=FraudType.INVOICE_FRAUD,
severity=FraudSeverity.MEDIUM,
confidence=0.7,
indicators=[FraudIndicator(
indicator_type="sequential_invoice_numbers",
description="Perfectly sequential invoice numbers suggest manipulation",
confidence=0.7,
evidence=[{"sequence": invoice_numbers[:10]}], # First 10 as example
risk_score=6.0
)],
entities_involved=[sorted_invoices[0].get("vendor_name", f"Vendor_{vendor_id}")],
estimated_impact=sum(inv.get("amount", 0) for inv in sorted_invoices) * 0.05,
recommendations=[
"Audit vendor invoicing practices",
"Verify actual services delivered",
"Review invoice approval process"
],
evidence_trail={"vendor_id": vendor_id}
))
except:
pass # Skip if invoice numbers aren't numeric
return patterns
async def _comprehensive_fraud_analysis(
self, data: Dict[str, Any], context: AgentContext
) -> List[FraudPattern]:
"""Perform comprehensive fraud analysis across all data types."""
all_patterns = []
# Analyze each data type if present
if "contracts" in data:
patterns = await self._analyze_contract_fraud(data["contracts"], context)
all_patterns.extend(patterns)
if "transactions" in data:
patterns = await self._analyze_transaction_fraud(data["transactions"], context)
all_patterns.extend(patterns)
if "vendors" in data:
patterns = await self._analyze_vendor_fraud(data["vendors"], context)
all_patterns.extend(patterns)
if "invoices" in data:
patterns = await self._analyze_invoice_fraud(data["invoices"], context)
all_patterns.extend(patterns)
# Cross-reference patterns for complex fraud schemes
complex_patterns = await self._detect_complex_fraud_schemes(all_patterns, data)
all_patterns.extend(complex_patterns)
return all_patterns
async def _detect_money_laundering(self, transactions: List[Dict[str, Any]]) -> List[FraudPattern]:
"""Detect potential money laundering patterns."""
patterns = []
# Look for structuring (smurfing) patterns
daily_transactions = {}
for trans in transactions:
date = trans.get("date", "").split("T")[0] # Extract date part
entity = trans.get("entity_id")
amount = trans.get("amount", 0)
key = f"{date}_{entity}"
if key not in daily_transactions:
daily_transactions[key] = []
daily_transactions[key].append(amount)
# Check for multiple transactions just below reporting threshold
threshold = 10000 # Common reporting threshold
for key, amounts in daily_transactions.items():
if len(amounts) > 2: # Multiple transactions in one day
below_threshold = [a for a in amounts if threshold * 0.8 < a < threshold]
if len(below_threshold) >= 2:
patterns.append(FraudPattern(
fraud_type=FraudType.MONEY_LAUNDERING,
severity=FraudSeverity.HIGH,
confidence=0.75,
indicators=[FraudIndicator(
indicator_type="structuring",
description="Multiple transactions just below reporting threshold",
confidence=0.75,
evidence=[{"transactions": below_threshold}],
risk_score=8.0
)],
entities_involved=[key.split("_")[1]],
estimated_impact=sum(below_threshold),
recommendations=[
"File suspicious activity report",
"Review all transactions by entity",
"Check for related accounts"
],
evidence_trail={"date_entity": key}
))
return patterns
async def _detect_kickback_schemes(self, transactions: List[Dict[str, Any]]) -> List[FraudPattern]:
"""Detect potential kickback schemes."""
patterns = []
# Look for payments shortly after contract awards
# This is a simplified detection - real implementation would be more sophisticated
return patterns
async def _detect_complex_fraud_schemes(
self, patterns: List[FraudPattern], data: Dict[str, Any]
) -> List[FraudPattern]:
"""Detect complex fraud schemes involving multiple fraud types."""
complex_patterns = []
# Look for entities involved in multiple fraud types
entity_fraud_types = {}
for pattern in patterns:
for entity in pattern.entities_involved:
if entity not in entity_fraud_types:
entity_fraud_types[entity] = set()
entity_fraud_types[entity].add(pattern.fraud_type)
# Entities with multiple fraud types indicate complex schemes
for entity, fraud_types in entity_fraud_types.items():
if len(fraud_types) >= 2:
# Combine evidence from all patterns involving this entity
entity_patterns = [p for p in patterns if entity in p.entities_involved]
complex_patterns.append(FraudPattern(
fraud_type=FraudType.PROCUREMENT_FRAUD, # Complex multi-type fraud
severity=FraudSeverity.CRITICAL,
confidence=0.85,
indicators=[
FraudIndicator(
indicator_type="complex_scheme",
description=f"Entity involved in multiple fraud types: {', '.join(f.value for f in fraud_types)}",
confidence=0.85,
evidence=[{"fraud_types": [f.value for f in fraud_types]}],
risk_score=9.5
)
],
entities_involved=[entity],
estimated_impact=sum(p.estimated_impact for p in entity_patterns),
recommendations=[
"Launch comprehensive investigation",
"Freeze entity accounts",
"Coordinate with law enforcement",
"Review all entity transactions and contracts"
],
evidence_trail={"related_patterns": [p.fraud_type.value for p in entity_patterns]}
))
return complex_patterns
def _check_bid_similarity(self, bid_amounts: List[float]) -> bool:
"""Check if bids are suspiciously similar."""
if len(bid_amounts) < 2:
return False
# Calculate pairwise similarities
similarities = []
for i in range(len(bid_amounts)):
for j in range(i + 1, len(bid_amounts)):
if bid_amounts[i] > 0 and bid_amounts[j] > 0:
similarity = 1 - abs(bid_amounts[i] - bid_amounts[j]) / max(bid_amounts[i], bid_amounts[j])
similarities.append(similarity)
# If any pair is too similar, flag it
return any(s > self.fraud_thresholds["bid_similarity"] for s in similarities)
def _check_bid_rotation(self, contracts: List[Dict[str, Any]]) -> bool:
"""Check for bid rotation patterns."""
# Sort by date
sorted_contracts = sorted(contracts, key=lambda x: x.get("contract_date", ""))
# Track winning vendors
winners = [c.get("vendor_id") for c in sorted_contracts if c.get("is_winner", False)]
if len(winners) < 4:
return False
# Check for rotation pattern (simplified check)
unique_winners = list(dict.fromkeys(winners)) # Preserve order, remove duplicates
if len(unique_winners) >= 3:
# Check if winners cycle through vendors
for i in range(len(winners) - len(unique_winners)):
if winners[i:i+len(unique_winners)] == unique_winners:
return True
return False
def _identify_high_risk_entities(self, patterns: List[FraudPattern]) -> List[Dict[str, Any]]:
"""Identify high-risk entities from fraud patterns."""
entity_risks = {}
for pattern in patterns:
for entity in pattern.entities_involved:
if entity not in entity_risks:
entity_risks[entity] = {
"entity": entity,
"risk_score": 0,
"fraud_types": set(),
"total_impact": 0
}
# Update risk score (max of all patterns)
max_risk = max(ind.risk_score for ind in pattern.indicators)
entity_risks[entity]["risk_score"] = max(entity_risks[entity]["risk_score"], max_risk)
# Add fraud type
entity_risks[entity]["fraud_types"].add(pattern.fraud_type.value)
# Add to total impact
entity_risks[entity]["total_impact"] += pattern.estimated_impact
# Convert to list and sort by risk
high_risk_entities = []
for entity_data in entity_risks.values():
entity_data["fraud_types"] = list(entity_data["fraud_types"])
high_risk_entities.append(entity_data)
high_risk_entities.sort(key=lambda x: x["risk_score"], reverse=True)
return high_risk_entities[:10] # Top 10 high-risk entities
def _calculate_overall_confidence(self, patterns: List[FraudPattern]) -> float:
"""Calculate overall confidence score for the fraud analysis."""
if not patterns:
return 0.0
# Weighted average based on estimated impact
total_impact = sum(p.estimated_impact for p in patterns)
if total_impact == 0:
# Simple average if no impact estimates
return sum(p.confidence for p in patterns) / len(patterns)
# Weighted average
weighted_confidence = sum(p.confidence * p.estimated_impact for p in patterns)
return weighted_confidence / total_impact
def _generate_fraud_report(self, patterns: List[FraudPattern]) -> Dict[str, Any]:
"""Generate comprehensive fraud analysis report."""
if not patterns:
return {
"summary": "No fraud patterns detected",
"risk_level": "LOW",
"patterns": [],
"recommendations": ["Continue routine monitoring"]
}
# Categorize by severity
critical_patterns = [p for p in patterns if p.severity == FraudSeverity.CRITICAL]
high_patterns = [p for p in patterns if p.severity == FraudSeverity.HIGH]
medium_patterns = [p for p in patterns if p.severity == FraudSeverity.MEDIUM]
low_patterns = [p for p in patterns if p.severity == FraudSeverity.LOW]
# Determine overall risk level
if critical_patterns:
risk_level = "CRITICAL"
elif len(high_patterns) >= 3:
risk_level = "CRITICAL"
elif high_patterns:
risk_level = "HIGH"
elif medium_patterns:
risk_level = "MEDIUM"
else:
risk_level = "LOW"
# Compile all recommendations
all_recommendations = []
recommendation_set = set()
for pattern in patterns:
for rec in pattern.recommendations:
if rec not in recommendation_set:
recommendation_set.add(rec)
all_recommendations.append(rec)
# Create summary
summary = f"Detected {len(patterns)} fraud patterns: "
summary += f"{len(critical_patterns)} critical, {len(high_patterns)} high, "
summary += f"{len(medium_patterns)} medium, {len(low_patterns)} low severity"
return {
"summary": summary,
"risk_level": risk_level,
"total_estimated_impact": sum(p.estimated_impact for p in patterns),
"patterns_by_type": {
fraud_type.value: len([p for p in patterns if p.fraud_type == fraud_type])
for fraud_type in FraudType
},
"patterns": [self._pattern_to_dict(p) for p in patterns],
"recommendations": all_recommendations[:10], # Top 10 recommendations
"requires_immediate_action": risk_level in ["CRITICAL", "HIGH"]
}
def _pattern_to_dict(self, pattern: FraudPattern) -> Dict[str, Any]:
"""Convert FraudPattern to dictionary for reporting."""
return {
"fraud_type": pattern.fraud_type.value,
"severity": pattern.severity.value,
"confidence": pattern.confidence,
"entities_involved": pattern.entities_involved,
"estimated_impact": pattern.estimated_impact,
"indicators": [
{
"type": ind.indicator_type,
"description": ind.description,
"confidence": ind.confidence,
"risk_score": ind.risk_score
}
for ind in pattern.indicators
],
"evidence_trail": pattern.evidence_trail,
"recommendations": pattern.recommendations[:3] # Top 3 for each pattern
}
async def hunt_specific_fraud(
self, fraud_type: FraudType, data: Dict[str, Any], context: AgentContext
) -> AgentResponse:
"""Hunt for a specific type of fraud with focused analysis."""
# This method allows targeting specific fraud types for deeper analysis
message = AgentMessage(
role="user",
content=f"Hunt for {fraud_type.value} fraud",
data=data,
metadata={"target_fraud_type": fraud_type.value}
)
return await self.process(message, context) |