Spaces:
Running
Running
File size: 10,354 Bytes
5e1a30c |
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 |
"""
Weighted Fusion implementation for Modular Retriever Architecture.
This module provides a direct implementation of score-based weighted fusion
as an alternative to RRF for improved modularity and flexibility.
"""
import logging
from typing import List, Dict, Any, Tuple
from .base import FusionStrategy
logger = logging.getLogger(__name__)
class WeightedFusion(FusionStrategy):
"""
Weighted score fusion implementation.
This is a direct implementation of score-based fusion that combines
dense and sparse retrieval results using direct score weighting.
No external dependencies are required.
Score Formula: final_score = dense_weight * dense_score + sparse_weight * sparse_score
Features:
- Direct score weighting (not rank-based)
- Optional score normalization
- Configurable weights for dense and sparse retrieval
- Handles empty result sets gracefully
- Preserves original score information
Example:
config = {
"weights": {
"dense": 0.7,
"sparse": 0.3
},
"normalize": True
}
fusion = WeightedFusion(config)
results = fusion.fuse_results(dense_results, sparse_results)
"""
def __init__(self, config: Dict[str, Any]):
"""
Initialize weighted fusion strategy.
Args:
config: Configuration dictionary with:
- weights: Dictionary with dense and sparse weights
- dense: Weight for dense retrieval (default: 0.7)
- sparse: Weight for sparse retrieval (default: 0.3)
- normalize: Whether to normalize scores to [0,1] range (default: True)
"""
self.config = config
# Extract weights
weights = config.get("weights", {})
self.dense_weight = weights.get("dense", 0.7)
self.sparse_weight = weights.get("sparse", 0.3)
self.normalize = config.get("normalize", True)
# Validation
if not 0 <= self.dense_weight <= 1:
raise ValueError("dense_weight must be between 0 and 1")
if not 0 <= self.sparse_weight <= 1:
raise ValueError("sparse_weight must be between 0 and 1")
# Normalize weights if they don't sum to 1
weight_sum = self.dense_weight + self.sparse_weight
if weight_sum > 0:
self.dense_weight /= weight_sum
self.sparse_weight /= weight_sum
else:
self.dense_weight = 0.7
self.sparse_weight = 0.3
logger.info(f"WeightedFusion initialized with dense_weight={self.dense_weight:.3f}, normalize={self.normalize}")
def fuse_results(
self,
dense_results: List[Tuple[int, float]],
sparse_results: List[Tuple[int, float]]
) -> List[Tuple[int, float]]:
"""
Fuse dense and sparse retrieval results using weighted scoring.
Args:
dense_results: List of (document_index, score) from dense retrieval
sparse_results: List of (document_index, score) from sparse retrieval
Returns:
List of (document_index, fused_score) tuples sorted by score
"""
# Handle empty results
if not dense_results and not sparse_results:
return []
if not dense_results:
return sparse_results[:] if sparse_results else []
if not sparse_results:
return dense_results[:] if dense_results else []
# Normalize scores if requested
normalized_dense = self._normalize_scores(dense_results) if self.normalize else dense_results
normalized_sparse = self._normalize_scores(sparse_results) if self.normalize else sparse_results
# Convert to dictionaries for efficient lookup
dense_scores = dict(normalized_dense)
sparse_scores = dict(normalized_sparse)
# Get all unique document IDs
all_docs = set(dense_scores.keys()) | set(sparse_scores.keys())
# Calculate weighted scores
weighted_results = []
for doc_id in all_docs:
dense_score = dense_scores.get(doc_id, 0.0)
sparse_score = sparse_scores.get(doc_id, 0.0)
final_score = self.dense_weight * dense_score + self.sparse_weight * sparse_score
weighted_results.append((doc_id, final_score))
# Sort by final score (descending)
weighted_results.sort(key=lambda x: x[1], reverse=True)
return weighted_results
def get_strategy_info(self) -> Dict[str, Any]:
"""
Get information about the weighted fusion strategy.
Returns:
Dictionary with strategy configuration and statistics
"""
return {
"algorithm": "weighted_score_fusion",
"dense_weight": self.dense_weight,
"sparse_weight": self.sparse_weight,
"normalize": self.normalize,
"parameters": {
"weights": {
"dense": self.dense_weight,
"sparse": self.sparse_weight
},
"normalize": self.normalize
}
}
def _normalize_scores(self, results: List[Tuple[int, float]]) -> List[Tuple[int, float]]:
"""
Normalize scores to [0,1] range.
Args:
results: List of (document_index, score) tuples
Returns:
List of (document_index, normalized_score) tuples
"""
if not results:
return []
scores = [score for _, score in results]
max_score = max(scores)
min_score = min(scores)
score_range = max_score - min_score
if score_range == 0:
# All scores are the same, return as-is
return results
# Normalize to [0,1] range
normalized_results = [
(doc_id, (score - min_score) / score_range)
for doc_id, score in results
]
return normalized_results
def update_weights(self, dense_weight: float, sparse_weight: float) -> None:
"""
Update fusion weights dynamically.
Args:
dense_weight: New weight for dense retrieval
sparse_weight: New weight for sparse retrieval
"""
if not 0 <= dense_weight <= 1:
raise ValueError("dense_weight must be between 0 and 1")
if not 0 <= sparse_weight <= 1:
raise ValueError("sparse_weight must be between 0 and 1")
# Normalize weights
weight_sum = dense_weight + sparse_weight
if weight_sum > 0:
self.dense_weight = dense_weight / weight_sum
self.sparse_weight = sparse_weight / weight_sum
else:
raise ValueError("At least one weight must be positive")
logger.info(f"Updated weighted fusion weights: dense={self.dense_weight:.3f}, sparse={self.sparse_weight:.3f}")
def set_normalize(self, normalize: bool) -> None:
"""
Update the normalization setting.
Args:
normalize: Whether to normalize scores
"""
self.normalize = normalize
logger.info(f"Updated normalization setting to {normalize}")
def calculate_individual_scores(
self,
dense_results: List[Tuple[int, float]],
sparse_results: List[Tuple[int, float]]
) -> Dict[int, Dict[str, float]]:
"""
Calculate individual weighted scores for debugging purposes.
Args:
dense_results: List of (document_index, score) from dense retrieval
sparse_results: List of (document_index, score) from sparse retrieval
Returns:
Dictionary mapping document_index to individual score components
"""
# Normalize scores if requested
normalized_dense = self._normalize_scores(dense_results) if self.normalize else dense_results
normalized_sparse = self._normalize_scores(sparse_results) if self.normalize else sparse_results
dense_scores = dict(normalized_dense)
sparse_scores = dict(normalized_sparse)
all_docs = set(dense_scores.keys()) | set(sparse_scores.keys())
scores = {}
for doc_id in all_docs:
dense_score = dense_scores.get(doc_id, 0.0)
sparse_score = sparse_scores.get(doc_id, 0.0)
weighted_dense = self.dense_weight * dense_score
weighted_sparse = self.sparse_weight * sparse_score
scores[doc_id] = {
"dense": dense_score,
"sparse": sparse_score,
"weighted_dense": weighted_dense,
"weighted_sparse": weighted_sparse,
"total": weighted_dense + weighted_sparse
}
return scores
def get_score_statistics(
self,
dense_results: List[Tuple[int, float]],
sparse_results: List[Tuple[int, float]]
) -> Dict[str, Any]:
"""
Get score statistics for analysis.
Args:
dense_results: List of (document_index, score) from dense retrieval
sparse_results: List of (document_index, score) from sparse retrieval
Returns:
Dictionary with score statistics
"""
stats = {}
if dense_results:
dense_scores = [score for _, score in dense_results]
stats["dense"] = {
"min": min(dense_scores),
"max": max(dense_scores),
"mean": sum(dense_scores) / len(dense_scores),
"count": len(dense_scores)
}
if sparse_results:
sparse_scores = [score for _, score in sparse_results]
stats["sparse"] = {
"min": min(sparse_scores),
"max": max(sparse_scores),
"mean": sum(sparse_scores) / len(sparse_scores),
"count": len(sparse_scores)
}
return stats |