Spaces:
Sleeping
Sleeping
File size: 11,507 Bytes
55ed80a | 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 | """
Reliability Testing System for AI-Enhanced Music Recommender.
Measures:
1. Consistency: Do repeated recommendations match?
2. Fairness: Are different demographics treated fairly?
3. Robustness: How do recommendations change with small input variations?
4. Explainability: Are explanations aligned with scores?
"""
from typing import List, Dict, Tuple, Optional
import json
from statistics import stdev, mean
class ReliabilityTester:
"""Test suite for validating recommendation system reliability."""
def __init__(self, recommender, rag_system=None):
"""Initialize tester with recommender and optional RAG system."""
self.recommender = recommender
self.rag_system = rag_system
self.test_results = []
def test_consistency(
self,
user_prefs: Dict,
songs: List[Dict],
num_runs: int = 5,
) -> Dict:
"""
Test consistency: Do repeated recommendations produce the same results?
Args:
user_prefs: User preferences
songs: Available songs
num_runs: Number of times to run recommendations
Returns:
Consistency test results
"""
from .recommender import recommend_songs
results = []
recommendation_sets = []
for run in range(num_runs):
recs = recommend_songs(
user_prefs=user_prefs,
songs=songs,
k=5,
)
recommendation_sets.append(
[song["id"] for song, _, _ in recs]
)
results.append(recs)
# Calculate consistency score
# Measure how many recommendations appear in all runs
if recommendation_sets:
all_ids = set(recommendation_sets[0])
for id_set in recommendation_sets[1:]:
all_ids = all_ids.intersection(set(id_set))
consistency_score = len(all_ids) / 5 # 5 recommendations total
# Check if top recommendation is always the same
top_ids = [rec_set[0] if rec_set else None for rec_set in recommendation_sets]
top_consistency = (
sum(1 for tid in top_ids if tid == top_ids[0]) / num_runs
)
else:
consistency_score = 0.0
top_consistency = 0.0
test_result = {
"test_name": "consistency",
"num_runs": num_runs,
"consistency_score": round(consistency_score, 2),
"top_recommendation_consistency": round(top_consistency, 2),
"recommendation_sets": recommendation_sets,
"status": "PASS" if consistency_score > 0.6 else "FAIL",
}
self.test_results.append(test_result)
return test_result
def test_robustness(
self,
base_prefs: Dict,
songs: List[Dict],
variation_amount: float = 0.1,
) -> Dict:
"""
Test robustness: How much do recommendations change with small input variations?
Args:
base_prefs: Base user preferences
songs: Available songs
variation_amount: Percentage to vary numeric preferences (0-1)
Returns:
Robustness test results
"""
from .recommender import recommend_songs
# Get baseline recommendations
baseline_recs = recommend_songs(
user_prefs=base_prefs,
songs=songs,
k=5,
)
baseline_ids = [song["id"] for song, _, _ in baseline_recs]
# Vary preferences slightly
variations = []
varied_recs_list = []
# Vary energy
if "energy" in base_prefs:
varied_energy = base_prefs.copy()
varied_energy["energy"] = max(
0.0,
min(1.0, base_prefs["energy"] * (1 + variation_amount))
)
variations.append("energy increased")
recs = recommend_songs(user_prefs=varied_energy, songs=songs, k=5)
varied_recs_list.append([song["id"] for song, _, _ in recs])
# Calculate robustness: what % of recommendations remain the same?
if varied_recs_list:
overlaps = [
len(set(baseline_ids) & set(varied_ids)) / 5
for varied_ids in varied_recs_list
]
robustness_score = mean(overlaps)
else:
robustness_score = 1.0
test_result = {
"test_name": "robustness",
"base_preferences": base_prefs,
"variations_tested": variations,
"baseline_recommendations": baseline_ids,
"varied_recommendations": varied_recs_list,
"robustness_score": round(robustness_score, 2),
"status": "PASS" if robustness_score > 0.4 else "WARN",
}
self.test_results.append(test_result)
return test_result
def test_fairness(
self,
user_profiles: List[Tuple[str, Dict]],
songs: List[Dict],
) -> Dict:
"""
Test fairness: Are artists and genres fairly represented across different user types?
Args:
user_profiles: List of (profile_name, preferences) tuples
songs: Available songs
Returns:
Fairness test results
"""
from .recommender import recommend_songs
profile_results = {}
all_genres = []
all_artists = []
for profile_name, prefs in user_profiles:
recs = recommend_songs(user_prefs=prefs, songs=songs, k=5)
genres = [song.get("genre") for song, _, _ in recs]
artists = [song.get("artist") for song, _, _ in recs]
all_genres.extend(genres)
all_artists.extend(artists)
profile_results[profile_name] = {
"recommended_genres": genres,
"recommended_artists": artists,
"genre_diversity": len(set(genres)) / len(genres),
"artist_diversity": len(set(artists)) / len(artists),
}
# Calculate overall fairness metrics
genre_distribution = {
g: all_genres.count(g) / len(all_genres)
for g in set(all_genres)
}
artist_distribution = {
a: all_artists.count(a) / len(all_artists)
for a in set(all_artists)
}
# Check for over-representation
over_represented = [
g for g, count in genre_distribution.items()
if count > 0.4
]
test_result = {
"test_name": "fairness",
"num_profiles": len(user_profiles),
"profile_results": profile_results,
"genre_distribution": {k: round(v, 2) for k, v in genre_distribution.items()},
"artist_distribution": {k: round(v, 2) for k, v in artist_distribution.items()},
"over_represented_genres": over_represented,
"fairness_issues": len(over_represented) > 0,
"status": "PASS" if len(over_represented) == 0 else "WARN",
}
self.test_results.append(test_result)
return test_result
def test_explanation_alignment(
self,
user_prefs: Dict,
recommended_song: Dict,
score: float,
reasons: List[str],
) -> Dict:
"""
Test explanation alignment: Do the reasons match the recommendation score?
Args:
user_prefs: User preferences
recommended_song: The recommended song
score: The recommendation score
reasons: Reasons for the recommendation
Returns:
Alignment test results
"""
# Score quality metrics
num_reasons = len(reasons)
alignment_checks = {
"has_reasons": num_reasons > 0,
"multiple_reasons": num_reasons >= 2,
"reasons_specific": any(
"match" in r.lower() or "close" in r.lower()
for r in reasons
),
}
# Check if score aligns with reasons
# Higher score should have more/stronger reasons
alignment_score = sum(alignment_checks.values()) / len(alignment_checks)
test_result = {
"test_name": "explanation_alignment",
"song": f"{recommended_song['title']} by {recommended_song['artist']}",
"score": score,
"num_reasons": num_reasons,
"reasons": reasons,
"alignment_checks": alignment_checks,
"alignment_score": round(alignment_score, 2),
"status": "PASS" if alignment_score > 0.5 else "FAIL",
}
self.test_results.append(test_result)
return test_result
def run_full_test_suite(
self,
user_prefs: Dict,
songs: List[Dict],
user_profiles: Optional[List[Tuple[str, Dict]]] = None,
) -> Dict:
"""
Run all reliability tests and generate a comprehensive report.
Args:
user_prefs: Primary user preferences
songs: Available songs
user_profiles: Optional list of different user profiles for fairness testing
Returns:
Complete test report
"""
from .recommender import recommend_songs
print("Running Reliability Test Suite...")
# Test 1: Consistency
print(" - Testing consistency...")
consistency = self.test_consistency(user_prefs, songs, num_runs=5)
# Test 2: Robustness
print(" - Testing robustness...")
robustness = self.test_robustness(user_prefs, songs)
# Test 3: Fairness (if multiple profiles provided)
fairness = None
if user_profiles:
print(" - Testing fairness...")
fairness = self.test_fairness(user_profiles, songs)
# Test 4: Explanation Alignment
print(" - Testing explanation alignment...")
recs = recommend_songs(user_prefs=user_prefs, songs=songs, k=1)
if recs:
song, score, reasons = recs[0]
alignment = self.test_explanation_alignment(
user_prefs, song, score, reasons
)
else:
alignment = {"status": "SKIP", "reason": "No recommendations generated"}
# Generate summary
test_summary = {
"timestamp": "2026-04-24",
"test_suite": "Reliability Testing",
"overall_status": "PASS",
"tests_run": 4 if fairness else 3,
"tests_passed": sum(
1 for t in [consistency, robustness, fairness, alignment]
if t and t.get("status") == "PASS"
),
"consistency": consistency,
"robustness": robustness,
"fairness": fairness,
"explanation_alignment": alignment,
}
# Determine overall status
statuses = [t.get("status") for t in [consistency, robustness, fairness, alignment] if t]
if "FAIL" in statuses:
test_summary["overall_status"] = "FAIL"
elif "WARN" in statuses:
test_summary["overall_status"] = "WARN"
return test_summary
def export_results(self) -> str:
"""Export all test results as JSON."""
return json.dumps(self.test_results, indent=2)
|