File size: 6,843 Bytes
d520909 |
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 |
"""
Document Classification Schemas
Pydantic models for document type classification and categorization.
"""
from enum import Enum
from typing import List, Dict, Any, Optional
from pydantic import BaseModel, Field
from .core import EvidenceRef
class DocumentType(str, Enum):
"""
Common document types for classification.
Extensible for domain-specific types.
"""
# Legal & Business
CONTRACT = "contract"
INVOICE = "invoice"
RECEIPT = "receipt"
PURCHASE_ORDER = "purchase_order"
AGREEMENT = "agreement"
NDA = "nda"
TERMS_OF_SERVICE = "terms_of_service"
# Technical & Scientific
PATENT = "patent"
RESEARCH_PAPER = "research_paper"
TECHNICAL_REPORT = "technical_report"
SPECIFICATION = "specification"
DATASHEET = "datasheet"
USER_MANUAL = "user_manual"
# Financial
FINANCIAL_REPORT = "financial_report"
BANK_STATEMENT = "bank_statement"
TAX_FORM = "tax_form"
BALANCE_SHEET = "balance_sheet"
INCOME_STATEMENT = "income_statement"
# Identity & Administrative
ID_DOCUMENT = "id_document"
PASSPORT = "passport"
DRIVERS_LICENSE = "drivers_license"
CERTIFICATE = "certificate"
FORM = "form"
APPLICATION = "application"
# Medical
MEDICAL_RECORD = "medical_record"
PRESCRIPTION = "prescription"
LAB_REPORT = "lab_report"
INSURANCE_CLAIM = "insurance_claim"
# General
LETTER = "letter"
EMAIL = "email"
MEMO = "memo"
PRESENTATION = "presentation"
SPREADSHEET = "spreadsheet"
REPORT = "report"
ARTICLE = "article"
BOOK = "book"
# Catch-all
OTHER = "other"
UNKNOWN = "unknown"
class ClassificationScore(BaseModel):
"""Score for a single document type classification."""
document_type: DocumentType = Field(..., description="Document type")
confidence: float = Field(..., ge=0.0, le=1.0, description="Classification confidence")
reasoning: Optional[str] = Field(default=None, description="Reasoning for classification")
class DocumentClassification(BaseModel):
"""
Document classification result with confidence scores.
"""
document_id: str = Field(..., description="Document identifier")
# Primary classification
primary_type: DocumentType = Field(..., description="Most likely document type")
primary_confidence: float = Field(
...,
ge=0.0,
le=1.0,
description="Confidence in primary classification"
)
# All classification scores
scores: List[ClassificationScore] = Field(
default_factory=list,
description="Scores for all considered types"
)
# Evidence
evidence: List[EvidenceRef] = Field(
default_factory=list,
description="Evidence supporting classification"
)
# Classification metadata
method: str = Field(
default="llm",
description="Classification method used (llm/rule-based/hybrid)"
)
model_used: Optional[str] = Field(default=None, description="Model used for classification")
# Warnings and flags
is_confident: bool = Field(
default=True,
description="Whether classification meets confidence threshold"
)
warnings: List[str] = Field(default_factory=list, description="Classification warnings")
needs_human_review: bool = Field(
default=False,
description="Whether human review is recommended"
)
# Additional attributes detected
attributes: Dict[str, Any] = Field(
default_factory=dict,
description="Additional detected attributes (language, domain, etc.)"
)
def get_top_k(self, k: int = 3) -> List[ClassificationScore]:
"""Get top k classifications by confidence."""
sorted_scores = sorted(self.scores, key=lambda x: x.confidence, reverse=True)
return sorted_scores[:k]
def is_type(self, doc_type: DocumentType, min_confidence: float = 0.5) -> bool:
"""Check if document is classified as a specific type with minimum confidence."""
for score in self.scores:
if score.document_type == doc_type and score.confidence >= min_confidence:
return True
return False
class DocumentCategoryRule(BaseModel):
"""
Rule for rule-based document classification.
"""
name: str = Field(..., description="Rule name")
document_type: DocumentType = Field(..., description="Target document type")
# Matching criteria
title_keywords: List[str] = Field(
default_factory=list,
description="Keywords to match in title"
)
content_keywords: List[str] = Field(
default_factory=list,
description="Keywords to match in content"
)
required_sections: List[str] = Field(
default_factory=list,
description="Required section headings"
)
file_patterns: List[str] = Field(
default_factory=list,
description="Filename patterns (regex)"
)
# Confidence adjustment
base_confidence: float = Field(
default=0.8,
ge=0.0,
le=1.0,
description="Base confidence when rule matches"
)
keyword_boost: float = Field(
default=0.05,
ge=0.0,
le=0.2,
description="Confidence boost per matched keyword"
)
# Priority
priority: int = Field(
default=0,
description="Rule priority (higher = checked first)"
)
class ClassificationConfig(BaseModel):
"""
Configuration for document classification.
"""
# Confidence thresholds
min_confidence: float = Field(
default=0.6,
ge=0.0,
le=1.0,
description="Minimum confidence for classification"
)
human_review_threshold: float = Field(
default=0.7,
ge=0.0,
le=1.0,
description="Below this, flag for human review"
)
# Classification method
use_llm: bool = Field(default=True, description="Use LLM for classification")
use_rules: bool = Field(default=True, description="Use rule-based classification")
hybrid_mode: str = Field(
default="llm_primary",
description="Hybrid mode: llm_primary, rules_primary, or ensemble"
)
# Custom rules
custom_rules: List[DocumentCategoryRule] = Field(
default_factory=list,
description="Custom classification rules"
)
# Document types to consider
enabled_types: List[DocumentType] = Field(
default_factory=lambda: list(DocumentType),
description="Document types to consider"
)
# Evidence requirements
require_evidence: bool = Field(
default=True,
description="Require evidence for classification"
)
max_evidence_snippets: int = Field(
default=3,
description="Maximum evidence snippets to include"
)
|