Extract ALL specific terms from the text. Be EXHAUSTIVE.

An entity is anything with a specific name:
- People, organizations, places
- Products, tools, systems, frameworks
- Technical terms, methods, algorithms
- Acronyms, abbreviations, nicknames, aliases
- Numbers, dates, metrics, versions
- Papers, events, projects

CRITICAL RULES:
1. Extract EVERY specific term - do NOT skip any
2. Keep the EXACT form as it appears in the text - do NOT standardize, translate, or normalize
3. If both an acronym and its full name appear, extract BOTH separately
4. Do NOT include generic words unless they are proper names
5. Each entity should be 1-5 words

Match input text language EXACTLY. English in → English out. Chinese in → Chinese out.

Output JSON array of entity names only.
