File size: 70,249 Bytes
2ec0d39 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 |
"""
Context Compression and Synthesis System
=======================================
Advanced context compression and efficient encoding system with synthesis capabilities
for integrating data from diverse sources and modalities within unified contextual models.
"""
import asyncio
import json
import logging
import pickle
import zlib
from datetime import datetime, timedelta
from typing import Dict, List, Any, Optional, Tuple, Union, Set
from dataclasses import dataclass, field
from enum import Enum
import numpy as np
from collections import defaultdict, deque
import hashlib
import base64
from ai_agent_framework.core.context_engineering_agent import (
ContextElement, ContextModality, ContextDimension, ContextVector
)
logger = logging.getLogger(__name__)
class CompressionStrategy(Enum):
"""Context compression strategies."""
HIERARCHICAL = "hierarchical"
SEMANTIC = "semantic"
TEMPORAL = "temporal"
RELEVANCE = "relevance"
MULTIMODAL = "multimodal"
ADAPTIVE = "adaptive"
class SynthesisMethod(Enum):
"""Context synthesis methods."""
FUSION = "fusion"
MESH = "mesh"
HIERARCHICAL = "hierarchical"
GRAPH_BASED = "graph_based"
ATTENTION_BASED = "attention_based"
ENSEMBLE = "ensemble"
@dataclass
class CompressionConfig:
"""Configuration for context compression."""
strategy: CompressionStrategy
compression_ratio: float
quality_threshold: float
adaptive_threshold: float
max_context_size: int
expiry_policy: Dict[str, Any]
@dataclass
class SynthesizedContext:
"""Represents synthesized context from multiple sources."""
id: str
source_elements: List[str]
synthesis_method: SynthesisMethod
compressed_representation: Dict[str, Any]
quality_score: float
confidence: float
temporal_scope: Tuple[datetime, datetime]
semantic_clusters: List[Dict[str, Any]]
def __post_init__(self):
if not self.id:
self.id = str(hashlib.md5(str(self.source_elements).encode()).hexdigest())
@dataclass
class ContextVector:
"""Vector representation of context for compression."""
element_id: str
vector: np.ndarray
vector_type: str # semantic, temporal, spatial, etc.
compression_level: int
quality_score: float
timestamp: datetime
def __post_init__(self):
if not self.timestamp:
self.timestamp = datetime.utcnow()
class ContextCompressionEngine:
"""Advanced context compression engine with multiple strategies."""
def __init__(self):
self.compression_configs = self._initialize_compression_configs()
self.vector_index = {} # Vector-to-element mapping
self.compression_cache = {} # Cached compressed representations
self.quality_metrics = {}
def _initialize_compression_configs(self) -> Dict[CompressionStrategy, CompressionConfig]:
"""Initialize compression configurations for different strategies."""
return {
CompressionStrategy.HIERARCHICAL: CompressionConfig(
strategy=CompressionStrategy.HIERARCHICAL,
compression_ratio=0.7,
quality_threshold=0.6,
adaptive_threshold=0.5,
max_context_size=100,
expiry_policy={"level_1": timedelta(minutes=30), "level_2": timedelta(hours=2)}
),
CompressionStrategy.SEMANTIC: CompressionConfig(
strategy=CompressionStrategy.SEMANTIC,
compression_ratio=0.8,
quality_threshold=0.7,
adaptive_threshold=0.6,
max_context_size=50,
expiry_policy={"core_concepts": timedelta(hours=6), "supporting": timedelta(hours=1)}
),
CompressionStrategy.TEMPORAL: CompressionConfig(
strategy=CompressionStrategy.TEMPORAL,
compression_ratio=0.6,
quality_threshold=0.5,
adaptive_threshold=0.4,
max_context_size=200,
expiry_policy={"immediate": timedelta(minutes=15), "recent": timedelta(hours=4)}
),
CompressionStrategy.RELEVANCE: CompressionConfig(
strategy=CompressionStrategy.RELEVANCE,
compression_ratio=0.9,
quality_threshold=0.8,
adaptive_threshold=0.7,
max_context_size=30,
expiry_policy={"high_relevance": timedelta(days=1), "medium": timedelta(hours=6)}
),
CompressionStrategy.MULTIMODAL: CompressionConfig(
strategy=CompressionStrategy.MULTIMODAL,
compression_ratio=0.5,
quality_threshold=0.6,
adaptive_threshold=0.5,
max_context_size=150,
expiry_policy={"text": timedelta(hours=2), "visual": timedelta(minutes=30)}
)
}
async def compress_context(
self,
context_elements: List[ContextElement],
strategy: CompressionStrategy = CompressionStrategy.ADAPTIVE,
quality_threshold: float = 0.6
) -> Dict[str, Any]:
"""Compress context using specified strategy."""
try:
if strategy == CompressionStrategy.ADAPTIVE:
strategy = await self._select_optimal_strategy(context_elements)
config = self.compression_configs.get(strategy)
if not config:
strategy = CompressionStrategy.HIERARCHICAL
config = self.compression_configs[strategy]
if strategy == CompressionStrategy.HIERARCHICAL:
return await self._hierarchical_compression(context_elements, config, quality_threshold)
elif strategy == CompressionStrategy.SEMANTIC:
return await self._semantic_compression(context_elements, config, quality_threshold)
elif strategy == CompressionStrategy.TEMPORAL:
return await self._temporal_compression(context_elements, config, quality_threshold)
elif strategy == CompressionStrategy.RELEVANCE:
return await self._relevance_compression(context_elements, config, quality_threshold)
elif strategy == CompressionStrategy.MULTIMODAL:
return await self._multimodal_compression(context_elements, config, quality_threshold)
else:
return await self._hierarchical_compression(context_elements, config, quality_threshold)
except Exception as e:
logger.error(f"Context compression failed: {e}")
return {"status": "error", "error": str(e)}
async def _select_optimal_strategy(self, context_elements: List[ContextElement]) -> CompressionStrategy:
"""Select optimal compression strategy based on context characteristics."""
if not context_elements:
return CompressionStrategy.HIERARCHICAL
# Analyze context characteristics
modalities = set(element.modality for element in context_elements)
temporal_spread = self._calculate_temporal_spread(context_elements)
relevance_variance = np.var([element.relevance_score for element in context_elements])
# Strategy selection logic
if len(modalities) > 3: # High modality diversity
return CompressionStrategy.MULTIMODAL
elif temporal_spread > timedelta(hours=2): # Wide temporal scope
return CompressionStrategy.TEMPORAL
elif relevance_variance > 0.1: # High relevance variance
return CompressionStrategy.RELEVANCE
else:
return CompressionStrategy.SEMANTIC
def _calculate_temporal_spread(self, context_elements: List[ContextElement]) -> timedelta:
"""Calculate temporal spread of context elements."""
if len(context_elements) < 2:
return timedelta(0)
timestamps = [element.timestamp for element in context_elements]
return max(timestamps) - min(timestamps)
async def _hierarchical_compression(
self,
context_elements: List[ContextElement],
config: CompressionConfig,
quality_threshold: float
) -> Dict[str, Any]:
"""Perform hierarchical context compression."""
# Level 1: High-importance core context
core_context = [
element for element in context_elements
if element.relevance_score > 0.8 and element.confidence > 0.7
]
# Level 2: Supporting context
supporting_context = [
element for element in context_elements
if 0.5 < element.relevance_score <= 0.8
]
# Level 3: Peripheral context
peripheral_context = [
element for element in context_elements
if element.relevance_score <= 0.5
]
# Apply compression based on size limits
compressed_core = await self._compress_element_group(core_context, 0.9, "core")
compressed_supporting = await self._compress_element_group(supporting_context, 0.7, "supporting")
compressed_peripheral = await self._compress_element_group(peripheral_context, 0.5, "peripheral")
return {
"strategy": "hierarchical",
"levels": {
"core": compressed_core,
"supporting": compressed_supporting,
"peripheral": compressed_peripheral
},
"compression_ratio": len(compressed_core) + len(compressed_supporting) + len(compressed_peripheral),
"quality_threshold": quality_threshold,
"total_elements": len(context_elements),
"compressed_elements": len(compressed_core) + len(compressed_supporting) + len(compressed_peripheral)
}
async def _semantic_compression(
self,
context_elements: List[ContextElement],
config: CompressionConfig,
quality_threshold: float
) -> Dict[str, Any]:
"""Perform semantic context compression."""
# Group elements by semantic similarity
semantic_groups = await self._group_by_semantic_similarity(context_elements)
compressed_groups = []
total_original_size = 0
total_compressed_size = 0
for group in semantic_groups:
# Compress each semantic group
group_elements = group["elements"]
group_theme = group["theme"]
compressed_group = await self._compress_semantic_group(group_elements, group_theme, quality_threshold)
total_original_size += len(group_elements)
total_compressed_size += len(compressed_group["compressed_elements"])
compressed_groups.append({
"theme": group_theme,
"elements": compressed_group["compressed_elements"],
"key_concepts": compressed_group["key_concepts"],
"representative_embedding": compressed_group["representative_embedding"],
"confidence": compressed_group["confidence"]
})
return {
"strategy": "semantic",
"groups": compressed_groups,
"compression_ratio": total_compressed_size / max(total_original_size, 1),
"quality_threshold": quality_threshold,
"semantic_coherence": self._calculate_semantic_coherence(compressed_groups)
}
async def _temporal_compression(
self,
context_elements: List[ContextElement],
config: CompressionConfig,
quality_threshold: float
) -> Dict[str, Any]:
"""Perform temporal context compression."""
# Sort elements by timestamp
sorted_elements = sorted(context_elements, key=lambda e: e.timestamp)
# Define temporal windows
current_time = datetime.utcnow()
immediate_window = current_time - timedelta(minutes=30)
recent_window = current_time - timedelta(hours=4)
temporal_windows = {
"immediate": [], # Last 30 minutes
"recent": [], # Last 4 hours
"historical": [] # Everything else
}
for element in sorted_elements:
if element.timestamp >= immediate_window:
temporal_windows["immediate"].append(element)
elif element.timestamp >= recent_window:
temporal_windows["recent"].append(element)
else:
temporal_windows["historical"].append(element)
# Compress each temporal window
compressed_windows = {}
for window_name, elements in temporal_windows.items():
if elements:
compressed_windows[window_name] = await self._compress_temporal_window(
elements, window_name, quality_threshold
)
return {
"strategy": "temporal",
"windows": compressed_windows,
"temporal_distribution": {name: len(elements) for name, elements in temporal_windows.items()},
"compression_ratio": sum(len(window.get("compressed_elements", []))
for window in compressed_windows.values()) / max(len(context_elements), 1)
}
async def _relevance_compression(
self,
context_elements: List[ContextElement],
config: CompressionConfig,
quality_threshold: float
) -> Dict[str, Any]:
"""Perform relevance-based context compression."""
# Sort by relevance score
sorted_elements = sorted(context_elements, key=lambda e: e.relevance_score, reverse=True)
# Create relevance tiers
high_relevance = [e for e in sorted_elements if e.relevance_score > 0.8]
medium_relevance = [e for e in sorted_elements if 0.5 < e.relevance_score <= 0.8]
low_relevance = [e for e in sorted_elements if e.relevance_score <= 0.5]
# Apply different compression levels based on relevance
compressed_high = await self._compress_with_minimal_loss(high_relevance, "high_relevance")
compressed_medium = await self._compress_with_standard_loss(medium_relevance, "medium_relevance")
compressed_low = await self._compress_aggressively(low_relevance, "low_relevance")
return {
"strategy": "relevance",
"relevance_tiers": {
"high": compressed_high,
"medium": compressed_medium,
"low": compressed_low
},
"preservation_rate": {
"high": len(compressed_high) / max(len(high_relevance), 1),
"medium": len(compressed_medium) / max(len(medium_relevance), 1),
"low": len(compressed_low) / max(len(low_relevance), 1)
},
"compression_ratio": (len(compressed_high) + len(compressed_medium) + len(compressed_low)) / max(len(context_elements), 1)
}
async def _multimodal_compression(
self,
context_elements: List[ContextElement],
config: CompressionConfig,
quality_threshold: float
) -> Dict[str, Any]:
"""Perform multimodal context compression."""
# Group by modality
modality_groups = defaultdict(list)
for element in context_elements:
modality_groups[element.modality].append(element)
compressed_modalities = {}
total_original = 0
total_compressed = 0
for modality, elements in modality_groups.items():
compressed_modality = await self._compress_modality_group(elements, modality, quality_threshold)
compressed_modalities[modality.value] = compressed_modality
total_original += len(elements)
total_compressed += len(compressed_modality["compressed_elements"])
return {
"strategy": "multimodal",
"modalities": compressed_modalities,
"cross_modal_correlations": await self._analyze_cross_modal_correlations(modality_groups),
"compression_ratio": total_compressed / max(total_original, 1)
}
async def _compress_element_group(
self,
elements: List[ContextElement],
compression_level: float,
group_type: str
) -> List[Dict[str, Any]]:
"""Compress a group of elements."""
if not elements:
return []
compressed = []
for element in elements:
if group_type == "core":
# Preserve full detail for core elements
compressed.append({
"id": element.id,
"content": element.content,
"metadata": element.metadata,
"full_representation": True
})
elif group_type == "supporting":
# Summarize supporting elements
compressed.append({
"id": element.id,
"summary": self._generate_summary(element.content),
"key_points": element.metadata.get("key_points", []),
"confidence": element.confidence
})
else: # peripheral
# Abstract peripheral elements
compressed.append({
"id": element.id,
"theme": element.metadata.get("theme", str(element.content)[:50]),
"relevance_score": element.relevance_score,
"abstract": True
})
return compressed
async def _group_by_semantic_similarity(self, context_elements: List[ContextElement]) -> List[Dict[str, Any]]:
"""Group context elements by semantic similarity."""
# Simple semantic grouping based on keywords and themes
groups = defaultdict(list)
for element in context_elements:
# Extract semantic theme
theme = self._extract_semantic_theme(element)
groups[theme].append(element)
# Convert to list format
semantic_groups = []
for theme, elements in groups.items():
semantic_groups.append({
"theme": theme,
"elements": elements,
"coherence_score": self._calculate_group_coherence(elements)
})
return semantic_groups
def _extract_semantic_theme(self, element: ContextElement) -> str:
"""Extract semantic theme from context element."""
content = str(element.content).lower()
metadata = element.metadata
# Simple theme extraction logic
if any(word in content for word in ["urgent", "asap", "deadline"]):
return "urgency"
elif any(word in content for word in ["team", "collaborate", "group"]):
return "collaboration"
elif any(word in content for word in ["data", "analysis", "metrics"]):
return "data_analysis"
elif any(word in content for word in ["customer", "client", "user"]):
return "customer_oriented"
else:
return "general"
def _calculate_group_coherence(self, elements: List[ContextElement]) -> float:
"""Calculate semantic coherence of a group."""
if len(elements) < 2:
return 1.0
# Calculate average similarity within group
similarities = []
for i, elem1 in enumerate(elements):
for j, elem2 in enumerate(elements[i+1:], i+1):
# Simple similarity based on metadata overlap
overlap = len(set(elem1.metadata.keys()) & set(elem2.metadata.keys()))
total_keys = len(set(elem1.metadata.keys()) | set(elem2.metadata.keys()))
similarity = overlap / max(total_keys, 1)
similarities.append(similarity)
return np.mean(similarities) if similarities else 0.5
def _generate_summary(self, content: Any) -> str:
"""Generate summary of content."""
content_str = str(content)
if len(content_str) <= 100:
return content_str
# Simple summarization
sentences = content_str.split('.')
if len(sentences) > 2:
return '. '.join(sentences[:2]) + '.'
else:
return content_str[:97] + "..."
async def _compress_semantic_group(
self,
elements: List[ContextElement],
theme: str,
quality_threshold: float
) -> Dict[str, Any]:
"""Compress a semantic group of elements."""
# Extract key concepts
key_concepts = await self._extract_key_concepts(elements)
# Generate representative embedding
representative_embedding = await self._generate_representative_embedding(elements)
# Select representative elements
representative_elements = sorted(
elements,
key=lambda e: e.relevance_score * e.confidence,
reverse=True
)[:min(5, len(elements))]
return {
"theme": theme,
"compressed_elements": [self._element_to_summary(e) for e in representative_elements],
"key_concepts": key_concepts,
"representative_embedding": representative_embedding.tolist() if isinstance(representative_embedding, np.ndarray) else representative_embedding,
"confidence": np.mean([e.confidence for e in elements])
}
async def _extract_key_concepts(self, elements: List[ContextElement]) -> List[str]:
"""Extract key concepts from semantic group."""
concepts = []
concept_scores = defaultdict(float)
for element in elements:
content = str(element.content).lower()
metadata = element.metadata
# Extract from content
words = content.split()
for word in words:
if len(word) > 3: # Skip short words
concept_scores[word] += element.relevance_score
# Extract from metadata
for key, value in metadata.items():
if isinstance(value, str):
concept_scores[value.lower()] += element.confidence * 0.5
# Select top concepts
sorted_concepts = sorted(concept_scores.items(), key=lambda x: x[1], reverse=True)
return [concept for concept, score in sorted_concepts[:10]]
async def _generate_representative_embedding(self, elements: List[ContextElement]) -> np.ndarray:
"""Generate representative embedding for semantic group."""
if not elements:
return np.zeros(128)
# Simple averaging approach (in production would use more sophisticated methods)
embeddings = []
for element in elements:
# Generate mock embedding
np.random.seed(hash(element.id) % (2**32))
embedding = np.random.rand(128)
embeddings.append(embedding * element.relevance_score)
if embeddings:
return np.mean(embeddings, axis=0)
else:
return np.zeros(128)
def _element_to_summary(self, element: ContextElement) -> Dict[str, Any]:
"""Convert element to summary format."""
return {
"id": element.id,
"content_summary": self._generate_summary(element.content),
"relevance_score": element.relevance_score,
"confidence": element.confidence,
"key_metadata": {k: v for k, v in list(element.metadata.items())[:3]}
}
async def _compress_temporal_window(
self,
elements: List[ContextElement],
window_name: str,
quality_threshold: float
) -> Dict[str, Any]:
"""Compress a temporal window of elements."""
# Apply time-based compression rules
if window_name == "immediate":
# Preserve detailed information for immediate context
compression_level = 0.9
elif window_name == "recent":
# Moderate compression for recent context
compression_level = 0.7
else:
# Aggressive compression for historical context
compression_level = 0.4
compressed_elements = []
for element in elements:
if element.relevance_score > quality_threshold:
if compression_level > 0.8:
compressed_elements.append(self._element_to_summary(element))
else:
compressed_elements.append({
"id": element.id,
"content_summary": self._generate_summary(element.content),
"relevance_score": element.relevance_score
})
return {
"window_name": window_name,
"compression_level": compression_level,
"compressed_elements": compressed_elements,
"time_range": {
"start": min(e.timestamp for e in elements).isoformat(),
"end": max(e.timestamp for e in elements).isoformat()
}
}
async def _compress_with_minimal_loss(self, elements: List[ContextElement], tier: str) -> List[Dict[str, Any]]:
"""Compress with minimal information loss."""
return [self._element_to_summary(e) for e in elements]
async def _compress_with_standard_loss(self, elements: List[ContextElement], tier: str) -> List[Dict[str, Any]]:
"""Compress with standard information loss."""
compressed = []
for element in elements:
compressed.append({
"id": element.id,
"summary": self._generate_summary(element.content),
"relevance_score": element.relevance_score,
"confidence": element.confidence
})
return compressed
async def _compress_aggressively(self, elements: List[ContextElement], tier: str) -> List[Dict[str, Any]]:
"""Compress aggressively with significant information loss."""
compressed = []
for element in elements:
compressed.append({
"id": element.id,
"theme": self._extract_semantic_theme(element),
"relevance_score": element.relevance_score,
"confidence": round(element.confidence, 2)
})
return compressed
async def _compress_modality_group(
self,
elements: List[ContextElement],
modality: ContextModality,
quality_threshold: float
) -> Dict[str, Any]:
"""Compress elements of a specific modality."""
# Modality-specific compression logic
if modality == ContextModality.TEXT:
return await self._compress_text_modality(elements, quality_threshold)
elif modality == ContextModality.VISUAL:
return await self._compress_visual_modality(elements, quality_threshold)
elif modality == ContextModality.AUDITORY:
return await self._compress_auditory_modality(elements, quality_threshold)
else:
return await self._compress_generic_modality(elements, quality_threshold)
async def _compress_text_modality(self, elements: List[ContextElement], quality_threshold: float) -> Dict[str, Any]:
"""Compress text modality elements."""
compressed_elements = []
for element in elements:
if element.relevance_score > quality_threshold:
compressed_elements.append({
"id": element.id,
"content_type": "text",
"content": self._generate_summary(element.content),
"metadata": element.metadata
})
return {
"modality": "text",
"compressed_elements": compressed_elements,
"compression_method": "text_summarization"
}
async def _compress_visual_modality(self, elements: List[ContextElement], quality_threshold: float) -> Dict[str, Any]:
"""Compress visual modality elements."""
compressed_elements = []
for element in elements:
if element.relevance_score > quality_threshold:
# For visual elements, preserve key visual descriptors
compressed_elements.append({
"id": element.id,
"content_type": "visual",
"visual_descriptor": element.metadata.get("visual_descriptor", str(element.content)[:50]),
"metadata": {k: v for k, v in element.metadata.items() if k in ["color", "shape", "size", "location"]}
})
return {
"modality": "visual",
"compressed_elements": compressed_elements,
"compression_method": "visual_feature_extraction"
}
async def _compress_auditory_modality(self, elements: List[ContextElement], quality_threshold: float) -> Dict[str, Any]:
"""Compress auditory modality elements."""
compressed_elements = []
for element in elements:
if element.relevance_score > quality_threshold:
compressed_elements.append({
"id": element.id,
"content_type": "auditory",
"audio_descriptor": element.metadata.get("audio_descriptor", str(element.content)[:50]),
"metadata": {k: v for k, v in element.metadata.items() if k in ["frequency", "amplitude", "duration"]}
})
return {
"modality": "auditory",
"compressed_elements": compressed_elements,
"compression_method": "audio_feature_extraction"
}
async def _compress_generic_modality(self, elements: List[ContextElement], quality_threshold: float) -> Dict[str, Any]:
"""Compress generic modality elements."""
compressed_elements = []
for element in elements:
if element.relevance_score > quality_threshold:
compressed_elements.append({
"id": element.id,
"content_summary": self._generate_summary(element.content),
"relevance_score": element.relevance_score,
"confidence": element.confidence
})
return {
"modality": "generic",
"compressed_elements": compressed_elements,
"compression_method": "generic_compression"
}
async def _analyze_cross_modal_correlations(self, modality_groups: Dict[ContextModality, List[ContextElement]]) -> Dict[str, Any]:
"""Analyze correlations between different modalities."""
correlations = {}
modalities = list(modality_groups.keys())
for i, mod1 in enumerate(modalities):
for mod2 in modalities[i+1:]:
elements1 = modality_groups[mod1]
elements2 = modality_groups[mod2]
# Calculate temporal correlation
temporal_corr = self._calculate_temporal_correlation(elements1, elements2)
# Calculate relevance correlation
relevance_corr = self._calculate_relevance_correlation(elements1, elements2)
correlations[f"{mod1.value}_{mod2.value}"] = {
"temporal_correlation": temporal_corr,
"relevance_correlation": relevance_corr,
"correlation_strength": (temporal_corr + relevance_corr) / 2
}
return correlations
def _calculate_temporal_correlation(self, elements1: List[ContextElement], elements2: List[ContextElement]) -> float:
"""Calculate temporal correlation between two sets of elements."""
if not elements1 or not elements2:
return 0.0
# Simple correlation based on timestamp proximity
timestamps1 = [e.timestamp for e in elements1]
timestamps2 = [e.timestamp for e in elements2]
# Calculate average time differences
avg_diff = 0
count = 0
for t1 in timestamps1:
min_diff = min(abs((t1 - t2).total_seconds()) for t2 in timestamps2)
avg_diff += min_diff
count += 1
if count == 0:
return 0.0
avg_diff /= count
# Convert to correlation score (closer timestamps = higher correlation)
return max(0, 1 - avg_diff / 3600) # Normalize by 1 hour
def _calculate_relevance_correlation(self, elements1: List[ContextElement], elements2: List[ContextElement]) -> float:
"""Calculate relevance correlation between two sets of elements."""
if not elements1 or not elements2:
return 0.0
# Simple correlation based on relevance score patterns
relevance1 = [e.relevance_score for e in elements1]
relevance2 = [e.relevance_score for e in elements2]
if len(relevance1) > 1 and len(relevance2) > 1:
# Calculate variance correlation
var1 = np.var(relevance1)
var2 = np.var(relevance2)
if var1 > 0 and var2 > 0:
# Higher variance correlation indicates similar patterns
return min(var1, var2) / max(var1, var2)
return 0.5 # Default moderate correlation
def _calculate_semantic_coherence(self, compressed_groups: List[Dict[str, Any]]) -> float:
"""Calculate overall semantic coherence of compressed context."""
if not compressed_groups:
return 0.0
coherence_scores = [group.get("confidence", 0.5) for group in compressed_groups]
return np.mean(coherence_scores)
class ContextSynthesisEngine:
"""Advanced context synthesis engine for integrating diverse data sources."""
def __init__(self):
self.synthesis_methods = {
SynthesisMethod.FUSION: self._fusion_synthesis,
SynthesisMethod.MESH: self._mesh_synthesis,
SynthesisMethod.HIERARCHICAL: self._hierarchical_synthesis,
SynthesisMethod.GRAPH_BASED: self._graph_based_synthesis,
SynthesisMethod.ATTENTION_BASED: self._attention_based_synthesis,
SynthesisMethod.ENSEMBLE: self._ensemble_synthesis
}
self.synthesis_cache = {}
async def synthesize_contexts(
self,
context_groups: List[List[ContextElement]],
synthesis_method: SynthesisMethod = SynthesisMethod.FUSION,
quality_threshold: float = 0.6
) -> SynthesizedContext:
"""Synthesize contexts from multiple sources."""
try:
if synthesis_method not in self.synthesis_methods:
synthesis_method = SynthesisMethod.FUSION
synthesis_func = self.synthesis_methods[synthesis_method]
result = await synthesis_func(context_groups, quality_threshold)
# Create synthesized context object
synthesized_context = SynthesizedContext(
id=f"synthesized_{int(time.time())}",
source_elements=[elem.id for group in context_groups for elem in group],
synthesis_method=synthesis_method,
compressed_representation=result,
quality_score=self._calculate_synthesis_quality(result),
confidence=self._calculate_synthesis_confidence(result),
temporal_scope=self._calculate_temporal_scope(context_groups),
semantic_clusters=self._identify_semantic_clusters(result)
)
return synthesized_context
except Exception as e:
logger.error(f"Context synthesis failed: {e}")
return SynthesizedContext(
id=f"error_{int(time.time())}",
source_elements=[],
synthesis_method=synthesis_method,
compressed_representation={"error": str(e)},
quality_score=0.0,
confidence=0.0,
temporal_scope=(datetime.utcnow(), datetime.utcnow()),
semantic_clusters=[]
)
async def _fusion_synthesis(
self,
context_groups: List[List[ContextElement]],
quality_threshold: float
) -> Dict[str, Any]:
"""Perform fusion-based context synthesis."""
# Flatten all elements
all_elements = []
for group in context_groups:
all_elements.extend(group)
# Remove duplicates based on content similarity
unique_elements = self._remove_duplicates(all_elements)
# Sort by relevance and confidence
sorted_elements = sorted(
unique_elements,
key=lambda e: e.relevance_score * e.confidence,
reverse=True
)
# Select top elements based on quality threshold
selected_elements = [
e for e in sorted_elements
if e.relevance_score * e.confidence > quality_threshold
]
# Create fused representation
fused_content = []
for element in selected_elements[:50]: # Limit to top 50
fused_content.append({
"id": element.id,
"content": element.content,
"source_group": self._identify_source_group(element, context_groups),
"relevance_score": element.relevance_score,
"confidence": element.confidence,
"metadata": element.metadata
})
return {
"method": "fusion",
"elements": fused_content,
"fusion_score": len(fused_content) / max(len(unique_elements), 1),
"quality_distribution": self._calculate_quality_distribution(selected_elements)
}
async def _mesh_synthesis(
self,
context_groups: List[List[ContextElement]],
quality_threshold: float
) -> Dict[str, Any]:
"""Perform mesh-based context synthesis."""
# Create connections between elements from different groups
connections = self._create_cross_group_connections(context_groups)
# Build mesh representation
mesh_nodes = []
mesh_edges = []
for group_idx, group in enumerate(context_groups):
for element in group:
if element.relevance_score > quality_threshold:
mesh_nodes.append({
"id": element.id,
"content": element.content,
"group": group_idx,
"relevance_score": element.relevance_score
})
# Add edges based on connections
for connection in connections:
if connection["source"] in [node["id"] for node in mesh_nodes] and \
connection["target"] in [node["id"] for node in mesh_nodes]:
mesh_edges.append(connection)
return {
"method": "mesh",
"nodes": mesh_nodes,
"edges": mesh_edges,
"mesh_density": len(mesh_edges) / max(len(mesh_nodes) * (len(mesh_nodes) - 1) / 2, 1),
"clustering_coefficient": self._calculate_clustering_coefficient(mesh_edges, mesh_nodes)
}
async def _hierarchical_synthesis(
self,
context_groups: List[List[ContextElement]],
quality_threshold: float
) -> Dict[str, Any]:
"""Perform hierarchical context synthesis."""
# Create hierarchy levels
hierarchy = {
"level_1": [], # High importance, cross-group elements
"level_2": [], # Medium importance, group-specific elements
"level_3": [] # Low importance, supporting elements
}
# Analyze cross-group presence
element_presence = self._analyze_cross_group_presence(context_groups)
for group_idx, group in enumerate(context_groups):
for element in group:
if element.relevance_score > quality_threshold:
presence_count = element_presence.get(element.id, 0)
if presence_count > 1: # Appears in multiple groups
hierarchy["level_1"].append({
"id": element.id,
"content": element.content,
"groups": self._find_element_groups(element, context_groups),
"cross_group_relevance": presence_count,
"relevance_score": element.relevance_score
})
elif element.relevance_score > 0.7:
hierarchy["level_2"].append({
"id": element.id,
"content": element.content,
"group": group_idx,
"relevance_score": element.relevance_score
})
else:
hierarchy["level_3"].append({
"id": element.id,
"content_summary": str(element.content)[:100],
"group": group_idx,
"relevance_score": element.relevance_score
})
return {
"method": "hierarchical",
"hierarchy": hierarchy,
"distribution": {
"level_1": len(hierarchy["level_1"]),
"level_2": len(hierarchy["level_2"]),
"level_3": len(hierarchy["level_3"])
}
}
async def _graph_based_synthesis(
self,
context_groups: List[List[ContextElement]],
quality_threshold: float
) -> Dict[str, Any]:
"""Perform graph-based context synthesis."""
# Build knowledge graph
nodes = []
edges = []
# Add nodes
for group_idx, group in enumerate(context_groups):
for element in group:
if element.relevance_score > quality_threshold:
nodes.append({
"id": element.id,
"content": element.content,
"group": group_idx,
"relevance_score": element.relevance_score,
"metadata": element.metadata
})
# Add edges based on semantic relationships
for i, node1 in enumerate(nodes):
for j, node2 in enumerate(nodes[i+1:], i+1):
relationship_strength = self._calculate_relationship_strength(node1, node2)
if relationship_strength > 0.3:
edges.append({
"source": node1["id"],
"target": node2["id"],
"strength": relationship_strength,
"relationship_type": "semantic"
})
# Calculate graph metrics
graph_metrics = self._calculate_graph_metrics(nodes, edges)
return {
"method": "graph_based",
"nodes": nodes,
"edges": edges,
"graph_metrics": graph_metrics,
"community_structure": self._detect_communities(nodes, edges)
}
async def _attention_based_synthesis(
self,
context_groups: List[List[ContextElement]],
quality_threshold: float
) -> Dict[str, Any]:
"""Perform attention-based context synthesis."""
# Calculate attention weights for each element
attention_weights = {}
total_elements = 0
for group_idx, group in enumerate(context_groups):
for element in group:
if element.relevance_score > quality_threshold:
# Calculate attention weight based on multiple factors
relevance_weight = element.relevance_score
confidence_weight = element.confidence
recency_weight = self._calculate_recency_weight(element.timestamp)
diversity_weight = self._calculate_diversity_weight(element, context_groups)
attention_score = (relevance_weight * 0.4 +
confidence_weight * 0.3 +
recency_weight * 0.2 +
diversity_weight * 0.1)
attention_weights[element.id] = attention_score
total_elements += 1
# Normalize attention weights
if attention_weights:
max_weight = max(attention_weights.values())
attention_weights = {k: v / max_weight for k, v in attention_weights.items()}
# Create attention-based representation
attended_elements = []
for group_idx, group in enumerate(context_groups):
for element in group:
if element.id in attention_weights and attention_weights[element.id] > 0.1:
attended_elements.append({
"id": element.id,
"content": element.content,
"group": group_idx,
"attention_weight": attention_weights[element.id],
"relevance_score": element.relevance_score
})
return {
"method": "attention_based",
"attended_elements": attended_elements,
"attention_distribution": self._calculate_attention_distribution(attention_weights),
"attention_entropy": self._calculate_attention_entropy(attention_weights)
}
async def _ensemble_synthesis(
self,
context_groups: List[List[ContextElement]],
quality_threshold: float
) -> Dict[str, Any]:
"""Perform ensemble-based context synthesis combining multiple methods."""
# Apply multiple synthesis methods
synthesis_results = {}
for method in [SynthesisMethod.FUSION, SynthesisMethod.HIERARCHICAL, SynthesisMethod.ATTENTION_BASED]:
try:
result = await self.synthesis_methods[method](context_groups, quality_threshold)
synthesis_results[method.value] = result
except Exception as e:
logger.warning(f"Ensemble synthesis method {method.value} failed: {e}")
# Combine results using voting/weighted approach
ensemble_result = self._combine_synthesis_results(synthesis_results)
return {
"method": "ensemble",
"individual_results": synthesis_results,
"ensemble_result": ensemble_result,
"method_agreement": self._calculate_method_agreement(synthesis_results)
}
def _remove_duplicates(self, elements: List[ContextElement]) -> List[ContextElement]:
"""Remove duplicate elements based on content similarity."""
unique_elements = []
seen_content_hashes = set()
for element in elements:
content_hash = hashlib.md5(str(element.content).encode()).hexdigest()
if content_hash not in seen_content_hashes:
seen_content_hashes.add(content_hash)
unique_elements.append(element)
return unique_elements
def _identify_source_group(self, element: ContextElement, context_groups: List[List[ContextElement]]) -> int:
"""Identify which group an element belongs to."""
for group_idx, group in enumerate(context_groups):
if element in group:
return group_idx
return -1
def _find_element_groups(self, element: ContextElement, context_groups: List[List[ContextElement]]) -> List[int]:
"""Find all groups an element appears in."""
groups = []
for group_idx, group in enumerate(context_groups):
if element in group:
groups.append(group_idx)
return groups
def _calculate_quality_distribution(self, elements: List[ContextElement]) -> Dict[str, float]:
"""Calculate quality score distribution."""
if not elements:
return {}
relevance_scores = [e.relevance_score for e in elements]
confidence_scores = [e.confidence for e in elements]
return {
"mean_relevance": np.mean(relevance_scores),
"std_relevance": np.std(relevance_scores),
"mean_confidence": np.mean(confidence_scores),
"std_confidence": np.std(confidence_scores)
}
def _create_cross_group_connections(self, context_groups: List[List[ContextElement]]) -> List[Dict[str, Any]]:
"""Create connections between elements from different groups."""
connections = []
for i, group1 in enumerate(context_groups):
for j, group2 in enumerate(context_groups[i+1:], i+1):
for elem1 in group1:
for elem2 in group2:
similarity = self._calculate_element_similarity(elem1, elem2)
if similarity > 0.5:
connections.append({
"source": elem1.id,
"target": elem2.id,
"strength": similarity,
"groups": (i, j)
})
return connections
def _calculate_element_similarity(self, elem1: ContextElement, elem2: ContextElement) -> float:
"""Calculate similarity between two elements."""
# Simple similarity based on metadata overlap
common_metadata = set(elem1.metadata.keys()) & set(elem2.metadata.keys())
total_metadata = set(elem1.metadata.keys()) | set(elem2.metadata.keys())
if not total_metadata:
return 0.0
metadata_similarity = len(common_metadata) / len(total_metadata)
# Content similarity (simplified)
content1 = str(elem1.content).lower()
content2 = str(elem2.content).lower()
content_words1 = set(content1.split())
content_words2 = set(content2.split())
if not content_words1 and not content_words2:
content_similarity = 1.0
elif not content_words1 or not content_words2:
content_similarity = 0.0
else:
content_similarity = len(content_words1 & content_words2) / len(content_words1 | content_words2)
return (metadata_similarity + content_similarity) / 2
def _calculate_clustering_coefficient(self, edges: List[Dict[str, Any]], nodes: List[Dict[str, Any]]) -> float:
"""Calculate clustering coefficient of the mesh."""
if len(nodes) < 3:
return 0.0
# Create adjacency list
adjacency = defaultdict(set)
for edge in edges:
adjacency[edge["source"]].add(edge["target"])
adjacency[edge["target"]].add(edge["source"])
# Calculate local clustering coefficients
clustering_coeffs = []
for node in nodes:
neighbors = adjacency[node["id"]]
if len(neighbors) < 2:
continue
possible_edges = len(neighbors) * (len(neighbors) - 1) / 2
actual_edges = 0
for neighbor1 in neighbors:
for neighbor2 in neighbors:
if neighbor1 != neighbor2 and neighbor2 in adjacency[neighbor1]:
actual_edges += 1
if possible_edges > 0:
clustering_coeffs.append(actual_edges / possible_edges)
return np.mean(clustering_coeffs) if clustering_coeffs else 0.0
def _analyze_cross_group_presence(self, context_groups: List[List[ContextElement]]) -> Dict[str, int]:
"""Analyze which elements appear in multiple groups."""
element_presence = defaultdict(int)
for group in context_groups:
group_elements = set(elem.id for elem in group)
for elem_id in group_elements:
element_presence[elem_id] += 1
return dict(element_presence)
def _calculate_relationship_strength(self, node1: Dict[str, Any], node2: Dict[str, Any]) -> float:
"""Calculate relationship strength between two nodes."""
# Simple relationship strength based on relevance scores and metadata overlap
relevance_factor = min(node1["relevance_score"], node2["relevance_score"])
# Metadata similarity
metadata1 = node1.get("metadata", {})
metadata2 = node2.get("metadata", {})
if not metadata1 and not metadata2:
metadata_similarity = 1.0
elif not metadata1 or not metadata2:
metadata_similarity = 0.0
else:
common_keys = set(metadata1.keys()) & set(metadata2.keys())
total_keys = set(metadata1.keys()) | set(metadata2.keys())
metadata_similarity = len(common_keys) / max(len(total_keys), 1)
return (relevance_factor + metadata_similarity) / 2
def _calculate_graph_metrics(self, nodes: List[Dict[str, Any]], edges: List[Dict[str, Any]]) -> Dict[str, float]:
"""Calculate metrics of the knowledge graph."""
if not nodes:
return {}
num_nodes = len(nodes)
num_edges = len(edges)
# Basic metrics
density = num_edges / max(num_nodes * (num_nodes - 1) / 2, 1)
# Average degree
degrees = defaultdict(int)
for edge in edges:
degrees[edge["source"]] += 1
degrees[edge["target"]] += 1
avg_degree = np.mean(list(degrees.values())) if degrees else 0
return {
"nodes": num_nodes,
"edges": num_edges,
"density": density,
"average_degree": avg_degree,
"connected_components": self._count_connected_components(nodes, edges)
}
def _count_connected_components(self, nodes: List[Dict[str, Any]], edges: List[Dict[str, Any]]) -> int:
"""Count connected components in the graph."""
if not nodes:
return 0
# Build adjacency list
adjacency = defaultdict(set)
for edge in edges:
adjacency[edge["source"]].add(edge["target"])
adjacency[edge["target"]].add(edge["source"])
# Find connected components using DFS
visited = set()
components = 0
for node in nodes:
if node["id"] not in visited:
self._dfs(node["id"], adjacency, visited)
components += 1
return components
def _dfs(self, node: str, adjacency: Dict[str, set], visited: set):
"""Depth-first search for connected components."""
visited.add(node)
for neighbor in adjacency[node]:
if neighbor not in visited:
self._dfs(neighbor, adjacency, visited)
def _detect_communities(self, nodes: List[Dict[str, Any]], edges: List[Dict[str, Any]]) -> List[List[str]]:
"""Detect communities in the graph (simplified algorithm)."""
# Simple community detection based on edge density
if not nodes or not edges:
return [[node["id"]] for node in nodes]
# Group nodes by their connectivity patterns
communities = []
remaining_nodes = [node["id"] for node in nodes]
while remaining_nodes:
# Start a new community
community = [remaining_nodes.pop(0)]
# Add connected nodes
to_add = True
while to_add:
to_add = False
for node_id in remaining_nodes[:]:
# Check if node is well-connected to community
connections = sum(1 for edge in edges
if (edge["source"] == node_id and edge["target"] in community) or
(edge["target"] == node_id and edge["source"] in community))
if connections >= len(community) * 0.3: # At least 30% connection rate
community.append(node_id)
remaining_nodes.remove(node_id)
to_add = True
break
communities.append(community)
return communities
def _calculate_recency_weight(self, timestamp: datetime) -> float:
"""Calculate recency weight based on timestamp."""
age_seconds = (datetime.utcnow() - timestamp).total_seconds()
return max(0, 1 - age_seconds / 86400) # Decay over 24 hours
def _calculate_diversity_weight(self, element: ContextElement, context_groups: List[List[ContextElement]]) -> float:
"""Calculate diversity weight based on element uniqueness."""
# Simple diversity calculation based on content uniqueness
content = str(element.content).lower()
word_count = len(content.split())
# Normalize by typical word count (assuming 20 words is average)
diversity_score = min(1.0, word_count / 20.0)
return diversity_score
def _calculate_attention_distribution(self, attention_weights: Dict[str, float]) -> Dict[str, float]:
"""Calculate attention weight distribution."""
if not attention_weights:
return {}
weights = list(attention_weights.values())
return {
"mean": np.mean(weights),
"std": np.std(weights),
"min": np.min(weights),
"max": np.max(weights),
"top_10_percent": np.percentile(weights, 90)
}
def _calculate_attention_entropy(self, attention_weights: Dict[str, float]) -> float:
"""Calculate entropy of attention distribution."""
if not attention_weights:
return 0.0
weights = list(attention_weights.values())
total = sum(weights)
if total == 0:
return 0.0
# Normalize to probabilities
probs = [w / total for w in weights]
# Calculate entropy
entropy = -sum(p * np.log2(p) for p in probs if p > 0)
return entropy
def _combine_synthesis_results(self, synthesis_results: Dict[str, Any]) -> Dict[str, Any]:
"""Combine results from multiple synthesis methods."""
combined_elements = {}
# Combine elements from all methods
for method_name, result in synthesis_results.items():
if "elements" in result:
for element in result["elements"]:
element_id = element["id"]
if element_id not in combined_elements:
combined_elements[element_id] = {
"id": element_id,
"content": element["content"],
"methods": [method_name],
"confidence_scores": [element.get("confidence", element.get("relevance_score", 0.5))]
}
else:
combined_elements[element_id]["methods"].append(method_name)
if "confidence" in element:
combined_elements[element_id]["confidence_scores"].append(element["confidence"])
# Calculate combined confidence
for element_id, element_data in combined_elements.items():
scores = element_data["confidence_scores"]
element_data["combined_confidence"] = np.mean(scores) if scores else 0.5
element_data["method_agreement"] = len(element_data["methods"])
return {
"combined_elements": list(combined_elements.values()),
"total_unique_elements": len(combined_elements),
"average_method_agreement": np.mean([elem["method_agreement"] for elem in combined_elements.values()])
}
def _calculate_method_agreement(self, synthesis_results: Dict[str, Any]) -> float:
"""Calculate agreement between synthesis methods."""
if len(synthesis_results) < 2:
return 1.0
element_sets = []
for method_name, result in synthesis_results.items():
if "elements" in result:
element_ids = set(elem["id"] for elem in result["elements"])
element_sets.append(element_ids)
if not element_sets:
return 0.0
# Calculate Jaccard similarity between all pairs
similarities = []
for i in range(len(element_sets)):
for j in range(i+1, len(element_sets)):
intersection = len(element_sets[i] & element_sets[j])
union = len(element_sets[i] | element_sets[j])
similarity = intersection / max(union, 1)
similarities.append(similarity)
return np.mean(similarities) if similarities else 0.0
def _calculate_synthesis_quality(self, result: Dict[str, Any]) -> float:
"""Calculate overall quality of synthesis result."""
if "elements" in result:
elements = result["elements"]
if not elements:
return 0.0
avg_relevance = np.mean([elem.get("relevance_score", 0.5) for elem in elements])
avg_confidence = np.mean([elem.get("confidence", 0.5) for elem in elements])
return (avg_relevance + avg_confidence) / 2
return 0.5 # Default quality score
def _calculate_synthesis_confidence(self, result: Dict[str, Any]) -> float:
"""Calculate confidence in synthesis result."""
# Base confidence on method and coverage
method_confidence = {
"fusion": 0.8,
"mesh": 0.7,
"hierarchical": 0.9,
"graph_based": 0.6,
"attention_based": 0.8,
"ensemble": 0.95
}
base_confidence = method_confidence.get(result.get("method", ""), 0.7)
# Adjust based on result coverage
if "elements" in result:
coverage_bonus = min(0.2, len(result["elements"]) / 100)
return min(1.0, base_confidence + coverage_bonus)
return base_confidence
def _calculate_temporal_scope(self, context_groups: List[List[ContextElement]]) -> Tuple[datetime, datetime]:
"""Calculate temporal scope of synthesized context."""
all_timestamps = []
for group in context_groups:
for element in group:
all_timestamps.append(element.timestamp)
if all_timestamps:
return (min(all_timestamps), max(all_timestamps))
else:
current_time = datetime.utcnow()
return (current_time, current_time)
def _identify_semantic_clusters(self, result: Dict[str, Any]) -> List[Dict[str, Any]]:
"""Identify semantic clusters in the synthesis result."""
clusters = []
if "elements" in result:
elements = result["elements"]
# Simple clustering based on content similarity
if len(elements) > 1:
# Group elements with similar themes
themes = defaultdict(list)
for element in elements:
# Extract theme from content
content = str(element.get("content", "")).lower()
if "urgent" in content or "asap" in content:
themes["urgency"].append(element)
elif "team" in content or "collaborate" in content:
themes["collaboration"].append(element)
elif "data" in content or "analysis" in content:
themes["data_analysis"].append(element)
else:
themes["general"].append(element)
for theme, theme_elements in themes.items():
if len(theme_elements) > 1:
clusters.append({
"theme": theme,
"elements": [elem["id"] for elem in theme_elements],
"size": len(theme_elements),
"coherence": len(theme_elements) / len(elements)
})
return clusters
# Integration with main context engineering system
class ContextCompressionAndSynthesis:
"""Integrated context compression and synthesis system."""
def __init__(self):
self.compression_engine = ContextCompressionEngine()
self.synthesis_engine = ContextSynthesisEngine()
async def process_context_pipeline(
self,
input_context: List[ContextElement],
compression_strategy: CompressionStrategy = CompressionStrategy.ADAPTIVE,
synthesis_method: SynthesisMethod = SynthesisMethod.FUSION,
quality_threshold: float = 0.6
) -> Dict[str, Any]:
"""Process context through compression and synthesis pipeline."""
# Step 1: Compress context
compression_result = await self.compression_engine.compress_context(
input_context, compression_strategy, quality_threshold
)
# Step 2: Prepare compressed elements for synthesis
compressed_elements = self._extract_compressed_elements(compression_result)
# Step 3: Synthesize contexts if multiple groups
# For demo, treat all compressed elements as one group
synthesis_result = await self.synthesis_engine.synthesize_contexts(
[compressed_elements], synthesis_method, quality_threshold
)
return {
"compression": compression_result,
"synthesis": {
"id": synthesis_result.id,
"method": synthesis_result.synthesis_method.value,
"quality_score": synthesis_result.quality_score,
"confidence": synthesis_result.confidence,
"temporal_scope": {
"start": synthesis_result.temporal_scope[0].isoformat(),
"end": synthesis_result.temporal_scope[1].isoformat()
},
"semantic_clusters": synthesis_result.semantic_clusters
},
"pipeline_metrics": self._calculate_pipeline_metrics(compression_result, synthesis_result)
}
def _extract_compressed_elements(self, compression_result: Dict[str, Any]) -> List[ContextElement]:
"""Extract context elements from compression result."""
elements = []
if compression_result.get("strategy") == "hierarchical":
for level_name, level_data in compression_result.get("levels", {}).items():
for compressed_elem in level_data:
if isinstance(compressed_elem, dict):
# Convert back to ContextElement-like object
elements.append(type('ContextElement', (), {
'id': compressed_elem.get("id", f"compressed_{level_name}_{len(elements)}"),
'content': compressed_elem.get("content", compressed_elem.get("summary", "")),
'relevance_score': compressed_elem.get("relevance_score", 0.5),
'confidence': compressed_elem.get("confidence", 0.5),
'timestamp': datetime.utcnow(),
'metadata': compressed_elem.get("metadata", {})
})())
return elements
def _calculate_pipeline_metrics(
self,
compression_result: Dict[str, Any],
synthesis_result: SynthesizedContext
) -> Dict[str, Any]:
"""Calculate metrics for the compression-synthesis pipeline."""
return {
"compression_ratio": compression_result.get("compression_ratio", 0),
"synthesis_quality": synthesis_result.quality_score,
"synthesis_confidence": synthesis_result.confidence,
"semantic_clusters": len(synthesis_result.semantic_clusters),
"temporal_span_hours": (synthesis_result.temporal_scope[1] - synthesis_result.temporal_scope[0]).total_seconds() / 3600,
"processing_efficiency": synthesis_result.quality_score / max(compression_result.get("compression_ratio", 1), 0.1)
}
if __name__ == "__main__":
print("Context Compression and Synthesis System Initialized")
print("=" * 60)
system = ContextCompressionAndSynthesis()
print("Ready for advanced context compression and multi-source synthesis!") |