[FEAT] Story 2.3 Multi-Timeline Character Consistency

[yeomin4242]

한줄 설명

Implement character trait preservation system that maintains core personality traits across different scenarios while adapting to scenario-specific changes.

문제·기회

No response

제안 내용

No response

완료 기준(AC)

• CharacterTraitExtractor service extracts core traits from base_story using Gemini 2.5 Flash (one-time per character)
• Trait database: character_traits table with character_name, base_story, core_traits JSONB, extracted_at
• Core traits categorized: personality (brave, intelligent), skills (magic, sword fighting), relationships (friends, enemies), values (loyalty, ambition)
• Scenario adaptation preserves core traits while modifying scenario-specific ones
• Example: Hermione in Slytherin preserves "intelligent, studious" but adapts "friends" from Harry/Ron to Draco/Pansy
• /api/ai/character-traits/{base_story}/{character} endpoint retrieves or extracts traits
• Prompt injection: "PRESERVE: [core_traits], ADAPT: [scenario_changes]"
• Trait cache with Redis (TTL 7 days, popular characters cached indefinitely)
• Gemini API cost optimization: Batch extract all characters from popular stories, cache aggressively
• VectorDB integration: Store extracted traits in ChromaDB for semantic similarity search
• Unit tests for trait preservation logic >80% coverage

관련 참고자료

No response

관련 이슈·블로커

No response