Evaluating Context Quality

123456789101112
from dataclasses import dataclass
from datetime import datetime, timedelta
import numpy as np

@dataclass
class ContextQualityScore:
    relevance: float      # 0-1, semantic + learned relevance
    completeness: float   # 0-1, required info coverage
    freshness: float      # 0-1, time-decayed score
    accuracy: float       # 0-1, verification score
    coherence: float      # 0-1, chunk boundary quality
    terseness: float      # 0-1, signal-to-noise ratio

123456789101112
from dataclasses import dataclass
from typing import List, Dict, Optional
import numpy as np
from sentence_transformers import SentenceTransformer

@dataclass
class ContextQualityMetrics:
    relevance_score: float  # 0-1, semantic similarity to query
    coverage_score: float   # 0-1, % of required info present
    freshness_score: float  # 0-1, based on data age
    density_score: float    # 0-1, signal-to-noise ratio
    redundancy_score: float # 0-1, lower is better

123456789101112
from dataclasses import dataclass
from typing import List, Dict
import numpy as np
from sentence_transformers import SentenceTransformer

@dataclass
class ContextChunk:
    content: str
    source: str
    timestamp: float
    metadata: Dict

123456789101112
interface ContextMetrics {
  // Composition metrics
  totalTokens: number;
  componentBreakdown: Record<string, number>; // tokens per component
  retrievedChunks: number;
  filteredChunks: number;
  
  // Quality metrics
  avgRelevanceScore: number;
  contextCoverage: number; // % of query topics covered
  redundancyRatio: number; // duplicate information
  freshnessScore: number;

123456789101112
from dataclasses import dataclass
from typing import List, Optional
import statistics
from datetime import datetime, timedelta

@dataclass
class QualityAlert:
    severity: str  # 'warning', 'critical'
    metric: str
    current_value: float
    threshold: float
    baseline: float