File size: 28,661 Bytes
0a47069
e80a3bc
 
6b7a4d5
9df6a7e
e80a3bc
5b8b298
ef82d45
 
c08ff6f
 
 
 
f302085
1a3c349
c08ff6f
 
1a6fb12
033af6f
146c098
c08ff6f
e80a3bc
2fe1b55
c08ff6f
d57920f
 
c08ff6f
e80a3bc
64fce5c
9df6a7e
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
d505821
 
 
 
9df6a7e
 
d505821
9df6a7e
d505821
 
 
 
 
9df6a7e
d505821
 
 
 
 
 
9df6a7e
d505821
9df6a7e
 
d505821
9df6a7e
93a860a
 
0a47069
 
 
 
d57920f
 
93a860a
d57920f
93a860a
146c098
93a860a
0a47069
 
9df6a7e
7b26e52
9df6a7e
e80a3bc
d57920f
9df6a7e
a147abc
ef82d45
 
ce3d575
aabeb2d
ef82d45
cf1ebab
96692cf
 
 
 
 
 
cf1ebab
 
 
 
 
96692cf
 
 
 
 
 
 
 
cf1ebab
 
5b8b298
24c2c08
 
f302085
3eb99bc
 
24c2c08
 
 
3eb99bc
 
 
cf1ebab
93a860a
d28f7eb
 
f302085
 
 
 
 
04f338f
8603187
1a6fb12
8603187
 
 
 
 
2fe1b55
8603187
9df6a7e
8603187
0b572d3
8603187
 
1a6fb12
04f338f
 
0b572d3
 
8603187
0b572d3
 
 
 
 
 
 
 
 
6be7d0b
 
146c098
785b76a
 
 
6be7d0b
 
 
 
 
a88e1d8
6be7d0b
0a47069
146c098
 
 
 
e80a3bc
ff9aa0c
ec95760
146c098
ec95760
 
 
146c098
d57920f
e80a3bc
0a47069
ef82d45
785b76a
 
 
ef82d45
 
 
dde6bd9
785b76a
 
 
dde6bd9
ef82d45
aabeb2d
cf1ebab
785b76a
 
 
ef82d45
ff9aa0c
cf1ebab
785b76a
cf1ebab
785b76a
ef82d45
 
cf1ebab
ff9aa0c
 
cf1ebab
785b76a
 
 
9df6a7e
 
cf1ebab
 
 
1a3c349
e80a3bc
785b76a
cf1ebab
785b76a
e80a3bc
64fce5c
 
96692cf
 
 
 
069b24e
 
3eb99bc
 
cf1ebab
24c2c08
 
abe9cba
 
 
3eb99bc
ba4c3a1
 
3eb99bc
abe9cba
ba4c3a1
 
 
 
abe9cba
 
 
ba4c3a1
 
abe9cba
ba4c3a1
 
 
abe9cba
ba4c3a1
 
 
 
 
abe9cba
 
 
 
 
ba4c3a1
 
 
 
 
 
 
abe9cba
ba4c3a1
abe9cba
 
ba4c3a1
24c2c08
abe9cba
 
 
 
3eb99bc
d28f7eb
abe9cba
24c2c08
dde6bd9
24c2c08
 
e80a3bc
24c2c08
e80a3bc
f302085
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
cf1ebab
f302085
 
 
 
 
 
 
 
 
 
 
71fb1a6
9df6a7e
71fb1a6
 
 
cf1ebab
9df6a7e
cf1ebab
 
d57920f
 
 
 
71fb1a6
e80a3bc
146c098
e80a3bc
 
 
 
a147abc
 
 
 
798df17
e80a3bc
798df17
 
 
 
2fe1b55
798df17
2fe1b55
 
 
c08ff6f
 
a147abc
 
 
 
 
 
 
 
 
c08ff6f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
2fe1b55
c08ff6f
2fe1b55
 
c08ff6f
2fe1b55
 
 
 
 
 
a147abc
e80a3bc
 
ec95760
 
 
 
 
 
a147abc
ec95760
 
 
 
 
d57920f
ec95760
2fe1b55
 
ec95760
2fe1b55
 
b5ff045
2fe1b55
 
a147abc
2fe1b55
 
 
 
 
 
ec95760
e80a3bc
6b7a4d5
9009e60
 
798df17
1a6fb12
798df17
8750468
 
 
 
 
 
 
 
 
 
2fe1b55
8750468
798df17
0b572d3
798df17
9009e60
 
033af6f
9009e60
033af6f
 
 
 
 
9009e60
 
033af6f
 
 
 
 
 
9009e60
033af6f
 
 
 
9009e60
033af6f
 
 
 
9009e60
033af6f
 
 
9009e60
033af6f
 
9009e60
033af6f
 
9009e60
033af6f
 
 
9009e60
033af6f
9009e60
033af6f
 
9009e60
033af6f
9009e60
033af6f
3b5daf4
146c098
3b5daf4
 
 
 
 
d57920f
146c098
d57920f
 
 
 
 
146c098
d57920f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
146c098
 
d57920f
 
 
146c098
d57920f
146c098
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
# portfolio/npc_social_network/npc/npc_base.py
import random
import copy
import re
from typing import Optional, List
from datetime import datetime
from .npc_memory import Memory, MemoryStore
from .npc_emotion import EmotionManager
from .npc_behavior import BehaviorManager
from .emotion_config import (EMOTION_LIST, EMOTION_CATEGORY_MAP, EMOTION_DECAY_RATE
                            , PERSONALITY_TEMPLATE, EMOTION_RELATION_IMPACT
                            , POSITIVE_RELATION_EMOTIONS, NEGATIVE_RELATION_EMOTIONS
                            , COGNITIVE_RELATION_EMOTIONS)
from .personality_config import AGE_PROFILE, PERSONALITY_PROFILE
from .npc_relationship import RelationshipManager
from ..models.llm_helper import (query_llm_with_prompt, query_llm_for_emotion, summarize_memories
                                , analyze_gossip, classify_and_extract_plan_details
                                , evaluate_social_action_outcome
                                , generate_reflection_topic)
from .npc_memory_embedder import search_similar_memories, add_memory_to_index
from .npc_manager import get_korean_postposition
from .npc_planner import PlannerManager
from .. import simulation_core
from typing import TYPE_CHECKING, Tuple, List
if TYPE_CHECKING:
    from .npc_manager import NPCManager
    from ..manager.conversation_manager import Conversation


# Personality ๋ฐ์ดํ„ฐ๋ฅผ ๊ด€๋ฆฌํ•˜๊ณ  ์š”์•ฝ ๊ธฐ๋Šฅ์„ ์ œ๊ณตํ•˜๋Š” ํด๋ž˜์Šค ์ •์˜
class PersonalityManager:
    """
    NPC์˜ ์„ฑ๊ฒฉ ๋ฐ์ดํ„ฐ๋ฅผ ๊ด€๋ฆฌํ•˜๊ณ  ๊ด€๋ จ ๊ธฐ๋Šฅ์„ ์ œ๊ณตํ•˜๋Š” ํด๋ž˜์Šค
    """
    def __init__(self, initial_traits=None):
        self.traits = copy.deepcopy(initial_traits) if initial_traits else copy.deepcopy(PERSONALITY_TEMPLATE)
    
    def get(self, key, default=None):
        return self.traits.get(key, default)
    
    def __getitem__(self, key):
        return self.traits[key]
    
    def __setitem__(self, key, value):
        self.traits[key] = value
    
    def keys(self):
        return self.traits.keys()
    
    def get_narrative_summary(self) -> str:
        """ํ˜„์žฌ ์„ฑ๊ฒฉ ์ˆ˜์น˜๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์„œ์ˆ ์ ์ธ ๋ฌ˜์‚ฌ๋ฅผ ๋™์ ์œผ๋กœ ์ƒ์„ฑ"""
        narratives = []

        sensitive = self.traits.get("sensitive", 0.5)
        stoic = self.traits.get("stoic", 0.5)
        congitive_bias = self.traits.get("cognitive_bias", 0.5)

        # ๊ฐ ์„ฑ๊ฒฉ ์ถ•์— ๋Œ€ํ•œ ์„œ์ˆ ์  ๋ฌ˜์‚ฌ ์ƒ์„ฑ
        if sensitive > 0.7 and stoic < 0.3:
            narratives.append("๊ฐ์ˆ˜์„ฑ์ด ๋งค์šฐ ํ’๋ถ€ํ•˜๊ณ  ๊ฐ์ •์˜ ์˜ํ–ฅ์„ ์‰ฝ๊ฒŒ ๋ฐ›๋Š” ํŽธ์ž…๋‹ˆ๋‹ค.")
        elif stoic > 0.7 and sensitive < 0.3:
            narratives.append("๊ฐ์ •์„ ๊ฑฐ์˜ ๋“œ๋Ÿฌ๋‚ด์ง€ ์•Š๋Š” ๊ณผ๋ฌตํ•˜๊ณ  ๋ฌด๋˜ํ•œ ์„ฑํ–ฅ์ž…๋‹ˆ๋‹ค.")
        else:
            narratives.append("์ƒํ™ฉ์— ๋”ฐ๋ผ ๊ฐ์ •์„ ํ‘œํ˜„ํ•  ์ค„ ์•„๋Š” ๊ท ํ˜• ์žกํžŒ ๋ชจ์Šต์„ ๋ณด์ž…๋‹ˆ๋‹ค.")

        if congitive_bias > 0.7:
            narratives.append("๋Œ€ํ™”๋ฅผ ํ•  ๋•Œ ๋…ผ๋ฆฌ์ ์ด๊ณ  ์‚ฌ์‹ค์— ๊ธฐ๋ฐ˜ํ•˜์—ฌ ํŒ๋‹จํ•˜๋ ค๋Š” ๊ฒฝํ–ฅ์ด ๊ฐ•ํ•ฉ๋‹ˆ๋‹ค.")
        elif congitive_bias < 0.3:
            narratives.append("์ง๊ด€๊ณผ ๊ฐ์ •์— ๋”ฐ๋ผ ์ƒํ™ฉ์„ ํŒ๋‹จํ•˜๋Š” ๊ฒฝํ–ฅ์ด ์žˆ์Šต๋‹ˆ๋‹ค.")

        if not narratives:
            return "ํ‰๋ฒ”ํ•˜๊ณ  ๊ท ํ˜• ์žกํžŒ ์„ฑ๊ฒฉ์ž…๋‹ˆ๋‹ค."
        
        return "".join(narratives)

# NPC ํด๋ž˜์Šค ์ •์˜
class NPC:
    """
    NPC ํด๋ž˜์Šค
    NPC์— ๋Œ€ํ•œ ๊ธฐ๋ณธ ์ •๋ณด๋ฅผ ๊ฐ์ฒดํ™”
    """
    def __init__(self, name: str, korean_name: str, job: str,
                 personality: Optional[dict]=None):
        self.name = name
        self.korean_name = korean_name
        self.job = job
        self.manager: Optional['NPCManager'] = None

        # npc ๊ธฐ์–ต
        self.memory_store = MemoryStore()

        # ๊ฐ์ • ์ƒํƒœ ๋“ฑ ๊ฐ์ • ๊ด€๋ฆฌ์ž ์ดˆ๊ธฐํ™”
        self.personality = PersonalityManager(personality)
        self.relationships = RelationshipManager(self)
        self.reputation = {}  # ๋‹ค๋ฅธ NPC์— ๋Œ€ํ•œ ํ‰ํŒ ์ €์žฅ
        self.emotion = EmotionManager(EMOTION_DECAY_RATE, self.personality)
        self.knowledge: dict[str, dict] = {}    # ๋‹ค๋ฅธ NPC์— ๋Œ€ํ•œ ์‚ฌ์‹ค ์ •๋ณด๋ฅผ ์ €์žฅํ•˜๋Š” '์ง€์‹ ๋ฒ ์ด์Šค'
        # ํ–‰๋™ ๊ด€๋ฆฌ์ž
        self.behavior = BehaviorManager()
        self.planner = PlannerManager(self)
        # ๋‚ด๋ถ€: float ๊ฐ์ • ์ˆ˜์น˜ ๊ด€๋ฆฌ์šฉ ๋ฒ„ํผ
        self._emotion_buffer = self.emotion.get_buffer()

        # baseline ๊ฐ์ • ๊ฐ’ ๋ถ€์—ฌ
        baseline_buffer = {}
        for emo in self._emotion_buffer:
            base_min = 0.05
            base_max = 0.3
            category = EMOTION_CATEGORY_MAP.get(emo, None)
            bias_map = {
                "core": "affect_bias", "social": "social_bias",
                "cognitive": "cognitive_bias", "complex": "complex_bias"
            }
            bias = self.personality.get(bias_map.get(category, "sensitive"), 1.0)

            baseline_value = base_min + (base_max - base_min) * bias
            baseline_value *= self.personality.get("sensitive", 1.0)    # ์ „์ฒด ๋ฏผ๊ฐ๋„ ์ ์šฉ

            baseline_buffer[emo] = round(baseline_value, 3)

        # baseline buffer๋ฅผ EmotionManager์— ๋“ฑ๋ก
        self.emotion.set_baseline(baseline_buffer)
        # emotion_buffer ์ดˆ๊ธฐํ™”
        self._emotion_buffer = self.emotion.get_buffer()

        # ํ”„๋กœํ•„ ์ง€์ • โ†’ ๊ธฐ๋ณธ์€ "stable", ๋‚˜์ค‘์— NPC๋งˆ๋‹ค ๋‹ค๋ฅด๊ฒŒ ๋ถ€์—ฌ
        profile_type = "stable"
        profile = PERSONALITY_PROFILE[profile_type]
        # Personality ๋ณ€ํ™” ์†๋„ ๊ฐœ๋ณ„ ์„ค์ •
        self.personality_change_rate = {
            "sensitive": random.uniform(*profile["sensitive"]),
            "stoic": random.uniform(*profile["stoic"]),
            "cognitive_bias": random.uniform(*profile["cognitive_bias"]),
        }

        # Personality baseline ์ €์žฅ (์ดˆ๊ธฐ๊ฐ’ ๋ณต์‚ฌ)
        self.personality_baseline = copy.deepcopy(self.personality.traits)

        # ๋‚˜์ด ์ถ”๊ฐ€ (18 ~ 50์„ธ ๊ธฐ๋ณธ ๋žœ๋ค ์ดˆ๊ธฐํ™” โ†’ ๋‚˜์ค‘์— ์ƒ์„ฑ ์‹œ age ์ง€์ • ๊ฐ€๋Šฅ)
        self.age = random.randint(18, 50)

        # ์ดˆ๊ธฐ personality_stage๋Š” update_personality_stage()์—์„œ ์ž๋™ ์„ค์ •
        self.personality_stage = None
        self.update_personality_stage()

    def generate_dialogue_turn(self, conversation: "Conversation", is_final_turn: bool=False) -> Tuple[str, str]:
        """๋Œ€ํ™”์˜ ํ˜„์žฌ ํ„ด์— ๋Œ€ํ•œ ์‘๋‹ต๊ณผ ํ–‰๋™์„ ์ƒ์„ฑ (๊ธฐ์–ต๊ณผ ๊ด€๊ณ„๋ฅผ ์ด ๋™์›)"""
        from ..models.llm_helper import generate_dialogue_action     
        # 1. ์ƒ๊ฐํ•˜๊ธฐ: '์ฒซ ๋งˆ๋””'์™€ '๋Œ€๋‹ต'์— ๋”ฐ๋ผ ๋‹ค๋ฅธ ์ •๋ณด๋กœ ๊ธฐ์–ต์„ ํƒ์ƒ‰.
        if not conversation.conversation_history:   # ๋Œ€ํ™” ๊ธฐ๋ก์ด ์—†์œผ๋ฉด '์ฒซ ๋งˆ๋””' ์ฐจ๋ก€
            query_text = conversation.topic
        else:   # ๋Œ€ํ™” ๊ธฐ๋ก์ด ์žˆ์œผ๋ฉด '๋Œ€ํ™”' ์ฐจ๋ก€
            query_text = "\n".join(conversation.conversation_history[-2:])  # ์ตœ๊ทผ 2ํ„ด์˜ ๋Œ€ํ™”๋กœ ๊ธฐ์–ต ํƒ์ƒ‰

        _, _, relevant_memories = search_similar_memories(self, query_text, top_k=3)

        # 2. ๋Œ€์‚ฌ ์ƒ์„ฑ: ๋ชจ๋“  ์ •๋ณด๋ฅผ ์ข…ํ•ฉํ•˜์—ฌ llm_helper์— ์š”์ฒญ
        result = generate_dialogue_action(
            speaker = self,
            target = conversation.participants[1-conversation.turn_index],
            conversation=conversation,
            memories = relevant_memories,
            is_final_turn = is_final_turn
        )
        dialogue = result.get("dialogue", "...")
        action = result.get("action", "END").strip().upper()

        # ํ–‰๋™ํ•˜๊ธฐ (์ž ๊ธˆ๊ณผ ํ•จ๊ป˜)
        with simulation_core.simulation_lock:
            emotion = query_llm_for_emotion(dialogue)
            if emotion and emotion in EMOTION_LIST:
                self.update_emotion(emotion, 1.0)

        return dialogue, action

    def remember(self, content: str, importance: int = 5, emotion: str = None,
                 strength: float = 1.0, memory_type:str = "Event",
                 context_tags: Optional[List[str]]=None) -> Memory:
        """
        NPC๊ฐ€ ์ƒˆ๋กœ์šด ๊ธฐ์–ต์„ ์ €์žฅํ•˜๊ณ  ๊ฐ์ • ์ƒํƒœ์— ๋ฐ˜์˜
        """
        memory = Memory(
            content=content,
            importance = importance,
            emotion=emotion or "neutral",
            memory_type=memory_type,
            context_tags=context_tags
            )
        self.memory_store.add_memory(memory)
        # ์ค‘์š”๋„๊ฐ€ 3 ์ด์ƒ์ธ ๊ธฐ์–ต๋งŒ ๋ฒกํ„ฐ ์ธ๋ฑ์Šค์— ์ถ”๊ฐ€ํ•˜์—ฌ ํšจ์œจ์„ฑ ํ™•๋ณด
        if memory.importance >= 3:
            add_memory_to_index(self, memory)

        if emotion and emotion in EMOTION_LIST:
            self.update_emotion(emotion, strength)  # strength = importance / 5.0์œผ๋กœ๋„ ๊ฐ€๋Šฅ
        # ๋งŒ์•ฝ ๊ธฐ์–ต์ด '์†Œ๋ฌธ'์ด๋ผ๋ฉด, ํ‰ํŒ์„ ์—…๋ฐ์ดํŠธ
        if memory_type == "Gossip":
            # (๊ฐ„๋‹จํ•œ ์˜ˆ์‹œ: LLM์œผ๋กœ ์†Œ๋ฌธ์˜ ๋Œ€์ƒ๊ณผ ๊ธ/๋ถ€์ •์„ ํŒŒ์•…ํ•ด์•ผ ๋” ์ •ํ™•ํ•จ)
            # ๊ธ์ • ์†Œ๋ฌธ ์˜ˆ์‹œ: "๋ฐฅ์ด ์ฐฐ๋ฆฌ๋ฅผ ๋„์™”๋‹ค"
            # ๋ถ€์ • ์†Œ๋ฌธ ์˜ˆ์‹œ: "์—˜๋ฆฐ์ด ์•จ๋ฆฌ์Šค์™€ ๋‹คํ‰œ๋‹ค"
            self.process_gossip_and_update_reputation(memory)

        return memory   # ์ƒ์„ฑ๋œ ๊ธฐ์–ต ๊ฐ์ฒด ๋ฐ˜ํ™˜

    def update_emotion(self, emotion:str, strength: float = 1.0):
        """
        ํŠน์ • ๊ฐ์ •์„ ๊ฐ์ • ์ƒํƒœ์— ๋ฐ˜์˜
        """
        self.emotion.update_emotion(emotion, strength)
        self._emotion_buffer = self.emotion.get_buffer()

    def decay_emotions(self):
        """
        ์‹œ๊ฐ„์ด ์ง€๋‚จ์— ๋”ฐ๋ผ ๋ชจ๋“  ๊ฐ์ •์ด ์„œ์„œํžˆ ๊ฐ์†Œ
        """
        self.emotion.decay_emotions()
        self._emotion_buffer = self.emotion.get_buffer()

    def recall_all_memories(self) -> List[str]:
        """
        ์ €์žฅ๋œ ๋ชจ๋“  ๊ธฐ์–ต(๋‹จ๊ธฐ + ์žฅ๊ธฐ)์„ ๋ฆฌ์ŠคํŠธ๋กœ ๋ฐ˜ํ™˜
        """
        return [m.content for m in self.memory_store.get_all_memories()]

    def get_composite_emotion_state(self, top_n: int = 3) -> List[tuple]:
        """
        ํ˜„์žฌ ๊ฐ์ • ์ƒํƒœ ์ค‘ ์ƒ์œ„ N๊ฐœ๋ฅผ ๋ฐ˜ํ™˜
        """
        buffer = self.emotion.get_buffer()  # EmotionManager์—์„œ ์ตœ์‹  ๊ฐ์ • ์ƒํƒœ ๋ณต์‚ฌ๋ณธ ๊ฐ€์ ธ์˜ค๊ธฐ
        sorted_emotions = sorted(buffer.items(), key=lambda x: x[1], reverse=True)
        composite = [(emo,round(score, 2)) for emo, score in sorted_emotions[:top_n] if score > 0.1]
        return composite # ์˜ˆ [('fear', 2.0), ('anger'), 1.5]
    
    def summarize_emotional_state(self) -> str:
        """
        ๊ฐ์ • ์ƒํƒœ ์š”์•ฝ ํ…์ŠคํŠธ ์ƒ์„ฑ
        """
        composite = self.get_composite_emotion_state()
        dominant = self.emotion.get_dominant_emotion()
        if not dominant:
            return "ํ˜„์žฌ ํŠน๋ณ„ํ•œ ๊ฐ์ • ์—†์ด ํ‰์˜จํ•œ ์ƒํƒœ์ž…๋‹ˆ๋‹ค."
        return f"ํ˜„์žฌ {dominant} ๋“ฑ์˜ ๊ฐ์ •์„ ๋А๋ผ๊ณ  ์žˆ์Šต๋‹ˆ๋‹ค.\n(์ฃผ์š” ๊ฐ์ •: {composite if composite else '์—†์Œ'})"
    
    def get_relationship_description(self, target_name: str) -> str:
        """
        ํŠน์ • ๋Œ€์ƒ๊ณผ์˜ ๊ด€๊ณ„๋ฅผ ์ž์—ฐ์–ด๋กœ ๋ฐ˜ํ™˜
        """
        return self.relationships.get_relationship_summary(target_name)


    def get_top_emotions(self, top_n=6):
        """
        ํ˜„์žฌ NPC ๊ฐ์ • ์ƒํƒœ ์ƒ์œ„ N๊ฐœ ๋ฐ˜ํ™˜
        """
        return self.emotion.get_top_emotions(top_n=top_n)
    
    def update_personality(self):
        """
        ๊ฒฝํ—˜์— ๋”ฐ๋ผ ์„ฑ๊ฒฉ์„ ์ ์ง„์ ์œผ๋กœ ๋ณ€ํ™”์‹œํ‚ด
        - MemoryType ๋ณ„ ๊ฐ€์ค‘์น˜ + EmotionInfluence ๋ณ„ ๊ฐ€์ค‘์น˜ ๋ชจ๋‘ ๋ฐ˜์˜
        """
        recent_memories = [m for m in self.memory_store.get_recent_memories(limit=10) if m.importance >= 5]
        if not recent_memories:
            return

        total_influence = {"sensitive": 0.0, "stoic": 0.0, "cognitive_bias": 0.0}

        # Personality ๋ณ€ํ™” ๋ฐ˜์˜
        for mem in recent_memories:
            if not mem.emotion or mem.emotion not in EMOTION_RELATION_IMPACT: continue

            # ๊ฐ ๊ฐ์ •์˜ ๊ณ ์œ ํ•œ ์˜ํ–ฅ๋ ฅ(impact)์„ ๊ฐ€์ ธ์˜ด
            emotion_impact = EMOTION_RELATION_IMPACT.get(mem.emotion, 0.0)

            # 1. ๊ฐ์ •์˜ ๊ธ์ •/๋ถ€์ •/์ธ์ง€์  ์†์„ฑ์— ๋”ฐ๋ฅธ ์˜ํ–ฅ ๋ฐฉํ–ฅ ๊ฒฐ์ •
            influence_vector = {"sensitive": 0.0, "stoic": 0.0, "cognitive_bias": 0.0}

            # ๊ฐ์ • ์˜ํ–ฅ๋ ฅ์ด ์–‘์ˆ˜์ด๋ฉด(๊ธ์ •์  ๊ฐ์ •), ๋ฏผ๊ฐ๋„์™€ ๋‚ด์„ฑ์„ ๋‚ฎ์ถค
            if mem.emotion in POSITIVE_RELATION_EMOTIONS:
                influence_vector["sensitive"] -= abs(emotion_impact) / 2.0    # ๊ธ์ •์  ๊ฒฝํ—˜์€ ๋ฏผ๊ฐ๋„๋ฅผ ๋‚ฎ์ถค
                influence_vector["stoic"] -= abs(emotion_impact) / 4.0        # ๊ธ์ •์ ์ธ ๊ฒฝํ—˜์€ ๋‚ด์„ฑ์ ์ผ ํ•„์š”๋ฅผ ์ค„์ž„
            # ๊ฐ์ • ์˜ํ–ฅ๋ ฅ์ด ์Œ์ˆ˜์ด๋ฉด(๋ถ€์ •์  ๊ฐ์ •), ๋ฏผ๊ฐ๋„์™€ ๋‚ด์„ฑ์„ ๋†’์ž„
            elif mem.emotion in NEGATIVE_RELATION_EMOTIONS:
                influence_vector["sensitive"] += abs(emotion_impact) / 2.0    # ๋ถ€์ •์  ๊ฒฝํ—˜์€ ๋ฏผ๊ฐ๋„๋ฅผ ๋†’์ž„
                influence_vector["stoic"] += abs(emotion_impact) / 4.0        # ๋ถ€์ •์  ๊ฒฝํ—˜์€ ๋‚ด์„ฑ์ ์œผ๋กœ ๋งŒ๋“ฆ
            # ์ธ์ง€์  ๊ฐ์ •์€ cognitive_bias์— ์˜ํ–ฅ
            if mem.emotion in COGNITIVE_RELATION_EMOTIONS:
                influence_vector["cognitive_bias"] += 0.4   # ์ธ์ง€์  ๊ฒฝํ—˜์€ ๋…ผ๋ฆฌ์  ์‚ฌ๊ณ ๋ฅผ ๊ฐ•ํ™”

            # 2. ๋ณ€ํ™” ๊ฐ•๋„ ๊ณ„์‚ฐ (๊ธฐ์–ต์˜ ์ค‘์š”๋„์™€ ๋‚˜์ด์— ๋”ฐ๋ผ)
            age_factor = AGE_PROFILE.get(self.personality_stage, 1.0)
            change_strength = (mem.importance / 10.0) * age_factor

            # 3. ์ด๋ฒˆ ๊ธฐ์–ต์œผ๋กœ ์ธํ•œ ์ตœ์ข… ๋ณ€ํ™”๋Ÿ‰ ๋ˆ„์ 
            for trait in total_influence.keys():
                total_influence[trait] += influence_vector[trait] * change_strength

        # 4. ๋ˆ„์ ๋œ ๋ณ€ํ™”๋Ÿ‰์„ ์„ฑ๊ฒฉ์— ์ ์šฉ
        for trait, total_change in total_influence.items():
            if abs(total_change) > 0.001: # ์˜๋ฏธ์žˆ๋Š” ๋ณ€ํ™”๋งŒ ์ ์šฉ
                rate = self.personality_change_rate.get(trait, 0.003)
                delta = total_change * rate
                self.personality[trait] = max(0.0, min(1.0, self.personality[trait] + delta))

        # 5. ๊ธฐ์ค€์„ ์œผ๋กœ ํšŒ๊ท€ (ํ•ญ์ƒ์„ฑ)
        base_decay = 0.001
        for trait in self.personality.keys():
            decay_rate = base_decay * (1.0 + self.personality.get("stoic", 0.0) - self.personality.get("sensitive", 0.0))
            decay_rate = max(0.0005, min(decay_rate, 0.005))
            self.personality[trait] += (self.personality_baseline[trait] - self.personality[trait]) * decay_rate

        print(f"[Personality Update] {self.korean_name} โ†’ sensitive: {self.personality['sensitive']:.3f}, "
              f"stoic: {self.personality['stoic']:.3f}, cognitive_bias: {self.personality['cognitive_bias']:.3f}", flush=True)

    def decay_memories(self):
        """
        ์‹œ๊ฐ„์ด ์ง€๋‚จ์— ๋”ฐ๋ผ ๊ธฐ์–ต importance ๊ฐ์†Œ
        - MemoryStore์˜ decay_memories์— self(NPC ๊ฐ์ฒด)๋ฅผ ์ „๋‹ฌ
        """
        self.memory_store.decay_memories(npc=self)

    def update_personality_stage(self):
        """
        ํ˜„์žฌ ๋‚˜์ด์— ๋”ฐ๋ผ personality_stage ์ž๋™ ์—…๋ฐ์ดํŠธ
        """
        if self.age < 3:
            self.personality_stage = "infancy_and_toddlerhood"
        elif self.age < 6:
            self.personality_stage = "early_childhood"
        elif self.age < 12:
            self.personality_stage = "middle_childhood"
        elif self.age < 18:
            self.personality_stage = "adolescence"
        elif self.age < 30:
            self.personality_stage = "young_adulthood"
        elif self.age < 65:
            self.personality_stage = "middle_adulthood"
        else:
            self.personality_stage = "older_adulthood"

    def get_memory_importance_scaling(self, memory_age_in_days: float) -> float:
        """
        NPC์˜ ๋‚˜์ด์— ๋”ฐ๋ผ memory_age_in_days ๊ธฐ๋ฐ˜ scaling factor ๋ฐ˜ํ™˜
        - ์ Š์€ NPC โ†’ ์ตœ๊ทผ ๊ธฐ์–ต ๋” ๊ฐ•์กฐ
        - ๋…ธ๋…„ NPC โ†’ ์˜ค๋ž˜๋œ ๊ธฐ์–ต ๋” ๊ฐ•์กฐ
        """
        if self.personality_stage in [
            "infancy_and_toddlerhood", "early_childhood", "middle_childhood",
            "adolescence", "young_adulthood"
            ]:
            # ์ตœ๊ทผ ๊ธฐ์–ต ๊ฐ•์กฐ โ†’ ์˜ค๋ž˜๋œ ๊ธฐ์–ต ์˜ํ–ฅ ์•ฝํ™”
            return max(0.5, 1.5 - 0.05 * memory_age_in_days)    # ์ตœ๋Œ€ 1.5, ์ ์  ๊ฐ์†Œ
        elif self.personality_stage in [
            "middle_adulthood", "older_adulthood"
        ]:
            return min(1.5, 0.5 + 0.05 * memory_age_in_days)    # ์ตœ์†Œ 0.5, ์ ์  ์ฆ๊ฐ€
        else:
            # fallback โ†’ ์ •์ƒํ™”
            return 1.0

    def summarize_and_store_memories(self, memories: List[Memory]):
        """
        ๊ธฐ์–ต ๋ฆฌ์ŠคํŠธ๋ฅผ ์š”์•ฝํ•˜๊ณ , NPC ์žฅ๊ธฐ ๊ธฐ์–ต์œผ๋กœ ์ €์žฅ
        """
        if not memories: return
        summary = summarize_memories(memories)
        if "[LLM Error]" in summary: return
        self.remember(
            content=f"[์š”์•ฝ๋œ ๊ธฐ์–ต] {summary}",
            importance=9,
            emotion=self.emotion.get_dominant_emotion(),
            memory_type="Summary"
        )

    def update_autonomous_behavior(self, time_context: str):
        """
        NPC๊ฐ€ ์Šค์Šค๋กœ ํŒ๋‹จํ•˜์—ฌ ํ–‰๋™์„ ๊ฒฐ์ •ํ•˜๋Š” ์ž์œจ ํ–‰๋™ ๋กœ์ง
        ์ฃผ๊ธฐ์ ์œผ๋กœ ํ˜ธ์ถœ๋˜์–ด ๋ชฉํ‘œ ์„ค์ •, ๊ณ„ํš ์‹คํ–‰ ๋“ฑ์„ ๋‹ด๋‹น.
        """
        # ํ”Œ๋ ˆ์ด์–ด๊ฐ€ ๋น„ํ™œ์„ฑํ™” ์ƒํƒœ์ผ ๊ฒฝ์šฐ, ์ž์œจ ํ–‰๋™์„ ์‹คํ–‰ํ•˜์ง€ ์•Š๋„๋ก ํ•˜๋Š” ์•ˆ์ „์žฅ์น˜
        if self.name == "player" and not self.manager.player_is_active:
            return

        # 1. ๊ณ„ํš ์‹คํ–‰ ๋กœ์ง์„ NPC๊ฐ€ ์ง์ ‘ ์ฒ˜๋ฆฌ
        if self.planner.has_active_plan():
            step = self.planner.get_current_step()
            if not step:
                self.planner.clear_plan()
                return
            print(f"[{self.korean_name}์˜ ํ–‰๋™ ๊ณ„ํš ์‹คํ–‰] {self.planner.current_plan.current_step_index + 1}/{len(self.planner.current_plan.steps)}: {step}")
            
            # LLM์œผ๋กœ ๊ณ„ํš์„ ๋ถ„์„ํ•˜์—ฌ ํ–‰๋™์„ ๊ฒฐ์ •
            plan_details = classify_and_extract_plan_details(step)
            action_type = plan_details.get("action_type")
            target_korean_name = plan_details.get("target")

            # ํ–‰๋™ ์‹คํ–‰ ์ „ '์กฐ๊ฑด ํ™•์ธ'๋‹จ๊ณ„
            can_execute = True
            if target_korean_name == "ํ”Œ๋ ˆ์ด์–ด" and not self.manager.player_is_active:
                # ๋ชฉํ‘œ ๋Œ€์ƒ์ด 'ํ”Œ๋ ˆ์ด์–ด'์ธ๋ฐ, ํ”Œ๋ ˆ์ด์–ด๊ฐ€ ๋น„ํ™œ์„ฑํ™” ์ƒํƒœ์ผ ๊ฒฝ์šฐ
                can_execute = False
                simulation_core.add_log(f"[{self.korean_name}์˜ ์ƒ๊ฐ] ์ง€๊ธˆ์€ ํ”Œ๋ ˆ์ด์–ด๊ฐ€ ์—†์œผ๋‹ˆ, '{step}' ๊ณ„ํš์€ ๋‚˜์ค‘์— ๋‹ค์‹œ ์‹œ๋„ํ•ด์•ผ๊ฒ ๋‹ค.")
                return

            # ํ–‰๋™ ์‹คํ–‰
            if action_type == "TALK" and target_korean_name:
                target_npc = self.manager.get_npc_by_korean_name(target_korean_name)
                if target_npc and target_npc != self:
                    # ์ฃผ์ œ๊ฐ€ ์žˆ๋Š” ๋Œ€ํ™” ์‹œ์ž‘
                    simulation_core.conversation_manager.start_conversation(self, target_npc, topic=plan_details.get("topic"))

            elif action_type == "HELP" and target_korean_name:
                target_npc = self.manager.get_npc_by_korean_name(target_korean_name)
                target_postposition = get_korean_postposition(target_korean_name, "์ด", "๊ฐ€")
                target_postposition_Eul = get_korean_postposition(target_korean_name, "์„", "๋ฅผ")

                if target_npc and target_npc != self:
                    # '๋•๊ธฐ' ํ–‰๋™์˜ ๊ฒฐ๊ณผ๋ฅผ ์‹œ๋ฎฌ๋ ˆ์ด์…˜
                    outcome = evaluate_social_action_outcome(self, target_npc, "HELP")
                    summary = outcome.get('outcome_summary')

                    # ๊ฒฐ๊ณผ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์–‘์ชฝ ๋ชจ๋‘์˜ ์ƒํƒœ๋ฅผ ์—…๋ฐ์ดํŠธ
                    with simulation_core.simulation_lock:
                        # ํ–‰๋™์ž(๋‚˜)์˜ ๊ธฐ์–ต๊ณผ ๊ด€๊ณ„ ์—…๋ฐ์ดํŠธ
                        self.remember(content=f"{target_korean_name}{target_postposition_Eul} ๋„์™”๋‹ค. ({summary})", importance=8, emotion=outcome.get('initiator_emotion'))
                        self.relationships.update_relationship(target_npc.name, outcome.get('initiator_emotion'), strength=3.0)

                        # ๋Œ€์ƒ์ž(์ƒ๋Œ€๋ฐฉ)์˜ ๊ธฐ์–ต๊ณผ ๊ด€๊ณ„ ์—…๋ฐ์ดํŠธ
                        target_npc.remember(content=f"{self.korean_name}{target_postposition} ๋‚˜๋ฅผ ๋„์™€์ฃผ์—ˆ๋‹ค. ({summary})")
                        target_npc.relationships.update_relationship(self.name, outcome.get('target_emotion'), strength=3.0)
                
            else:   # SOLOT_ACTION ๋˜๋Š” ๊ธฐํƒ€
                # ๋Œ€ํ™”๊ฐ€ ์•„๋‹Œ ๋‹ค๋ฅธ ํ–‰๋™์€ ๊ทธ๋Œ€๋กœ ๊ธฐ์–ต์—๋งŒ ๊ธฐ๋ก
                self.remember(
                    content=f"'{self.planner.current_goal.description}' ๋ชฉํ‘œ๋ฅผ ์œ„ํ•ด ํ˜ผ์ž ํ–‰๋™ ํ–ˆ๋‹ค: {step}",
                    importance = 6,
                    emotion="engagement",
                    memory_type="Behavior"
                )
            # ํ˜„์žฌ ๋‹จ๊ณ„ ์™„๋ฃŒ ์ฒ˜๋ฆฌ
            self.planner.complete_step()
            
        # 2. ๊ณ„ํš์ด ์—†์œผ๋ฉด ์ƒˆ๋กœ์šด ๋ชฉํ‘œ๋ฅผ ์ƒ์„ฑํ•˜๋„๋ก ์‹œ๋„
        else:
            # ์†Œ๋ฌธ์„ ํผ๋œจ๋ฆฌ๋ ค๋Š” ์‹œ๋„
            gossip_memories = [m for m in self.memory_store.get_all_memories() if m.memory_type == "Gossip" and not m.is_shared]
            if gossip_memories and random.random() < 0.2: # 20% ํ™•๋ฅ ๋กœ ์†Œ๋ฌธ ์ „ํŒŒ ์‹œ๋„
                gossip_to_spread = random.choice(gossip_memories)
                
                # ์•„์ง ๋Œ€ํ™”ํ•œ ์  ์—†๋Š” NPC๋ฅผ ํƒ€๊ฒŸ์œผ๋กœ ์„ ์ •
                potential_targets = [npc for npc in self.manager.get_interactive_npcs() if npc.name != self.name]
                if potential_targets:
                    target_npc = random.choice(potential_targets)
                    
                    # LLM์„ ํ™œ์šฉํ•˜์—ฌ ์†Œ๋ฌธ์„ ํผ๋œจ๋ฆฌ๋Š” ๋Œ€ํ™” ์ƒ์„ฑ
                    gossip_dialogue = self.generate_dialogue(
                        user_input=f"'{target_npc.korean_name}'์—๊ฒŒ '{gossip_to_spread.content}'์— ๋Œ€ํ•ด ๋„Œ์ง€์‹œ ์ด์•ผ๊ธฐํ•œ๋‹ค.",
                        time_context=time_context,
                        target_npc=target_npc,
                        create_memory=False
                    )

                    if gossip_dialogue and "[LLM Error]" not in gossip_dialogue:
                        simulation_core.add_log(f"[์†Œ๋ฌธ ์ „ํŒŒ] {self.korean_name} -> {target_npc.korean_name}: {gossip_dialogue}")
                        gossip_to_spread.is_shared = True

                        # 2. '์†Œ๋ฌธ ์ „ํŒŒ'๋ผ๋Š” ํ–‰๋™์— ๋Œ€ํ•œ ๋ช…ํ™•ํ•œ ๊ธฐ์–ต์„ ๋ณ„๋„๋กœ ์ƒ์„ฑ
                        self.remember(
                            content=f"{target_npc.korean_name}์—๊ฒŒ '{gossip_to_spread.content}'๋ผ๋Š” ์†Œ๋ฌธ์— ๋Œ€ํ•ด ์ด์•ผ๊ธฐํ–ˆ๋‹ค.",
                            importance=5,
                            emotion="engagement", # '๋‹ดํ•ฉ'๊ณผ ๊ฐ™์€ ์‚ฌํšŒ์  ํ–‰๋™์— ๋Œ€ํ•œ ๊ฐ์ •
                            memory_type="Gossip_Spreading" # '์†Œ๋ฌธ ์ „ํŒŒ'๋ผ๋Š” ๋ณ„๋„์˜ ํƒ€์ž… ์ง€์ •
                        )

            # ์ผ์ • ํ™•๋ฅ ๋กœ ์ƒˆ๋กœ์šด ๋ชฉํ‘œ๋ฅผ ์ƒ์„ฑ (๋งค๋ฒˆ ์ƒ์„ฑํ•˜์ง€ ์•Š๋„๋ก)
            if random.random() < 0.05: # 5% ํ™•๋ฅ 
                self.planner.generate_goal()

            elif random.random() < 0.1: # 10% ํ™•๋ฅ ๋กœ ๋‹ค๋ฅธ NPC์—๊ฒŒ ๊ฐ€๋ฒผ์šด ์ธ์‚ฌ ์‹œ๋„
                potential_targets = [npc for npc in self.manager.get_interactive_npcs() if npc != self]
                if potential_targets:
                    weights = []
                    # ๊ฐ€์ค‘์น˜ ๊ณ„์‚ฐ ๋กœ์ง
                    for target in potential_targets:
                        # ๊ด€๊ณ„ ์ ์ˆ˜๋ฅผ -100 ~ 100์—์„œ 0~2 ๋ฒ”์œ„๋กœ ๋ณ€ํ™˜ํ•˜์—ฌ ๊ธฐ๋ณธ ๊ฐ€์ค‘์น˜ 1์— ๋”ํ•จ
                        # score 100 -> weight 6 / score 0 -> weight 1 / score -100 -> weight 0.1 (์ตœ์†Œ ํ™•๋ฅ )
                        score = self.relationships.get_relationship_score(target.name)
                        weight = 1 + (score / 20.0)
                        weights.append(max(0.1, weight)) # ๊ฐ€์ค‘์น˜๊ฐ€ 0์ด ๋˜์ง€ ์•Š๋„๋ก ์ตœ์†Œ๊ฐ’ ๋ณด์žฅ
                    
                    # ๊ฐ€์ค‘์น˜์— ๋”ฐ๋ผ ๋Œ€ํ™” ์ƒ๋Œ€ 1๋ช… ์„ ํƒ
                    target_npc = random.choices(potential_targets, weights=weights, k=1)[0]
                     
                    # ์ฃผ์ œ๊ฐ€ ์—†๋Š” ๊ฐ€๋ฒผ์šด ๋Œ€ํ™” ์‹œ์ž‘
                    simulation_core.conversation_manager.start_conversation(self, target_npc)

    def create_symbolic_memory(self):
        """
        ์ตœ๊ทผ ๊ฒฝํ—˜๊ณผ ์„ฑ๊ฒฉ์„ ๋ฐ”ํƒ•์œผ๋กœ ๋™์ ์œผ๋กœ ์„ฑ์ฐฐ ์ฃผ์ œ๋ฅผ ์ •ํ•˜๊ณ , ๊ฐ€์น˜๊ด€์„ ํ˜•์„ฑ
        """
        simulation_core.add_log(f"[{self.korean_name}: ๊นŠ์€ ์ƒ๊ฐ์— ์ž ๊น๋‹ˆ๋‹ค...]")

        # 1. ์ตœ๊ทผ ๊ธฐ์–ต 10๊ฐœ๋ฅผ ๊ฐ€์ ธ์˜ต๋‹ˆ๋‹ค.
        recent_memories = self.memory_store.get_all_memories()[-10:]
        if len(recent_memories) < 8:    # ์ถฉ๋ถ„ํ•œ ๊ธฐ์–ต์ด ์Œ“์—ฟ์„ ๋•Œ๋งŒ ์ƒ์„ฑ
            return
        
        # 2. ์„ฑ๊ฒฉ ์ •๋ณด๋ฅผ ํ•จ๊ป˜ ์ „๋‹ฌํ•˜์—ฌ ์„ฑ์ฐฐ ์ฃผ์ œ ์ƒ์„ฑ
        reflection_topic = generate_reflection_topic(
            recent_memories,
            self.personality.get_narrative_summary()
        )
        simulation_core.add_log(f"[{self.korean_name}์˜ ์„ฑ์ฐฐ ์ฃผ์ œ] {reflection_topic}")

        # 3. ์ƒ์„ฑ๋œ ์ฃผ์ œ์™€ ๊ฐ€์žฅ ๊ด€๋ จ ์žˆ๋Š” ๊ธฐ์–ต๋“ค์„ ๋‹ค์‹œ ๊ฒ€์ƒ‰ํ•ฉ๋‹ˆ๋‹ค.
        _, _, relevant_memories = search_similar_memories(self, reflection_topic, top_k=10)
        if not relevant_memories:
            relevant_memories = recent_memories # ๊ด€๋ จ ๊ธฐ์–ต์ด ์—†์œผ๋ฉด ์ตœ๊ทผ ๊ธฐ์–ต์„ ๊ทธ๋Œ€๋กœ ์‚ฌ์šฉ

        # 4. ๊ด€๋ จ ๊ธฐ์–ต๋“ค์„ ์š”์•ฝ
        memory_summary = summarize_memories(relevant_memories)

        # 5. ์š”์•ฝ๋œ ๊ธฐ์–ต๊ณผ ์„ฑ์ฐฐ ์ฃผ์ œ๋ฅผ ๋ฐ”ํƒ•์œผ๋กœ ์ตœ์ข… ๊นจ๋‹ฌ์Œ์„ ์–ป์Šต๋‹ˆ๋‹ค.
        prompt = f"""
# Instructions
- ๋‹น์‹ ์€ '{self.korean_name}'์ž…๋‹ˆ๋‹ค.
- ์•„๋ž˜์˜ ์ •๋ณด๋“ค์„ ๋ฐ”ํƒ•์œผ๋กœ, ๋‹น์‹ ์˜ ์ธ์ƒ์— ๋Œ€ํ•œ ๊นŠ์€ ๊นจ๋‹ฌ์Œ์ด๋‚˜ ์‹ ๋…์„ ํ•œ ๋ฌธ์žฅ์˜ '์ƒ์ง•์ ์ธ ๊ธฐ์–ต'์œผ๋กœ ๋งŒ๋“ค์–ด์ฃผ์„ธ์š”.

# Reflection Topic
"{reflection_topic}"

# Summary of Related Memories
{memory_summary}

# My Realization (a one-sentence lesson or core belief):
"""
        symbolic_memory_content = query_llm_with_prompt(prompt).strip()

        if symbolic_memory_content and "[LLM Error]" not in symbolic_memory_content:
            self.remember(
                content=symbolic_memory_content,
                importance=10,
                memory_type="Symbolic",
                emotion="realization"   # ๊นจ๋‹ฌ์Œ์ด๋ผ๋Š” ์ƒˆ๋กœ์šด ๊ฐ์ •
            )
            simulation_core.add_log(f"[{self.korean_name}์˜ ์ƒˆ๋กœ์šด ๊นจ๋‹ฌ์Œ] {symbolic_memory_content}")

    def update_reputation(self, target_name: str, score_change: int):
        """ํŠน์ • ๋Œ€์ƒ์— ๋Œ€ํ•œ ํ‰ํŒ ์ ์ˆ˜๋ฅผ ๋ณ€๊ฒฝ"""
        if self.name == target_name: return  # ์ž๊ธฐ ์ž์‹ ์— ๋Œ€ํ•œ ํ‰ํŒx (์ž๊ธฐ ์ž์‹ ์˜ ํ‰ํŒ์— ๋Œ€ํ•œ ๋‚ด์šฉ๋„ ์„ฑ๊ฒฉ ํ˜•์„ฑ์— ์ค‘์š”ํ•  ์ˆ˜ ์žˆ์ง€ ์•Š๋‚˜?)

        current_score = self.reputation.get(target_name, 0)
        self.reputation[target_name] = current_score + score_change

        target_npc = self.manager.get_npc_by_name(target_name)
        target_korean_name = target_npc.korean_name if target_npc else target_name

        print(f"[{self.korean_name}]์˜ {target_korean_name}์— ๋Œ€ํ•œ ํ‰ํŒ: {self.reputation[target_name]} ({score_change:+.0f})")


    def process_gossip_and_update_reputation(self, gossip_memory: Memory):
        """LLM์„ ์ด์šฉํ•ด ์†Œ๋ฌธ์„ ๋ถ„์„ํ•˜๊ณ  ๊ด€๋ จ ์ธ๋ฌผ๋“ค์˜ ํ‰ํŒ์„ ์—…๋ฐ์ดํŠธ"""

        analysis = analyze_gossip(gossip_memory.content)
        if not analysis:
            return
        
        person1_korean_name = analysis.get("person1")
        person2_korean_name = analysis.get("person2")
        sentiment = analysis.get("sentiment")

        if not all([person1_korean_name, person2_korean_name, sentiment]):
            return
        
        score_change = 0

        if sentiment == "positive":
            score_change = 2
        elif sentiment == "negative":
            score_change = -2

        if score_change != 0:
            # LLM์ด ๋ฐ˜ํ™˜ํ•œ ํ•œ๊ธ€ ์ด๋ฆ„์œผ๋กœ NPC ๊ฐ์ฒด๋ฅผ ์ฐพ์•„ ์˜์–ด ID๋ฅผ ๊ฐ€์ ธ์˜จ๋‹ค.
            p1_npc = self.manager.get_npc_by_korean_name(person1_korean_name)
            p2_npc = self.manager.get_npc_by_korean_name(person2_korean_name)

            # ์†Œ๋ฌธ์˜ ๋Œ€์ƒ์ด ๋œ ๋‘ ์‚ฌ๋žŒ์— ๋Œ€ํ•œ ๋‚˜์˜ ํ‰ํŒ์„ ๋ณ€๊ฒฝ
            if p1_npc:
                self.update_reputation(p1_npc.name, score_change)
            if p2_npc:
                self.update_reputation(p2_npc.name, score_change)