feat: Add experiments framework and novelty-driven agent loop

- Add complete experiments directory with pilot study infrastructure
  - 5 experimental conditions (direct, expert-only, attribute-only, full-pipeline, random-perspective)
  - Human assessment tool with React frontend and FastAPI backend
  - AUT flexibility analysis with jump signal detection
  - Result visualization and metrics computation

- Add novelty-driven agent loop module (experiments/novelty_loop/)
  - NoveltyDrivenTaskAgent with expert perspective perturbation
  - Three termination strategies: breakthrough, exhaust, coverage
  - Interactive CLI demo with colored output
  - Embedding-based novelty scoring

- Add DDC knowledge domain classification data (en/zh)
- Add CLAUDE.md project documentation
- Update research report with experiment findings

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

experiments/conditions/c1_direct.py

@@ -0,0 +1,111 @@
+"""
+Condition 1: Direct Generation (Baseline)
+
+Single LLM call asking for creative ideas directly.
+No attribute decomposition, no expert perspectives.
+"""
+
+import sys
+from pathlib import Path
+
+# Add backend to path for imports
+sys.path.insert(0, str(Path(__file__).parent.parent.parent / "backend"))
+
+from typing import List, Dict, Any
+from app.services.llm_service import ollama_provider, extract_json_from_response
+from experiments.config import MODEL, TEMPERATURE, IDEAS_DIRECT, PROMPT_LANGUAGE
+
+
+def get_direct_generation_prompt(query: str, idea_count: int, lang: str = "en") -> str:
+    """Generate prompt for direct idea generation."""
+    if lang == "en":
+        return f"""/no_think
+Generate {idea_count} creative and innovative ideas for "{query}".
+
+Requirements:
+1. Each idea should be specific and actionable
+2. Ideas should be diverse, covering different aspects and applications
+3. Include both practical improvements and creative innovations
+4. Ideas should be 15-30 words each
+
+Return JSON only:
+{{"ideas": ["idea 1", "idea 2", "idea 3", ...]}}
+
+Generate exactly {idea_count} ideas."""
+    else:
+        return f"""/no_think
+為「{query}」生成 {idea_count} 個創意點子。
+
+要求：
+1. 每個點子要具體可行
+2. 點子要多元，涵蓋不同面向和應用
+3. 包含實用改進和創意創新
+4. 每個點子 15-30 字
+
+只回傳 JSON：
+{{"ideas": ["點子1", "點子2", "點子3", ...]}}
+
+生成正好 {idea_count} 個點子。"""
+
+
+async def generate_ideas(
+    query: str,
+    model: str = None,
+    temperature: float = None,
+    idea_count: int = None,
+    lang: str = None
+) -> Dict[str, Any]:
+    """
+    Generate ideas using direct LLM generation (C1 baseline).
+
+    Args:
+        query: The object/concept to generate ideas for
+        model: LLM model to use (default from config)
+        temperature: Generation temperature (default from config)
+        idea_count: Number of ideas to generate (default from config)
+        lang: Language for prompts (default from config)
+
+    Returns:
+        Dict with ideas and metadata
+    """
+    model = model or MODEL
+    temperature = temperature or TEMPERATURE
+    idea_count = idea_count or IDEAS_DIRECT
+    lang = lang or PROMPT_LANGUAGE
+
+    prompt = get_direct_generation_prompt(query, idea_count, lang)
+
+    response = await ollama_provider.generate(
+        prompt=prompt,
+        model=model,
+        temperature=temperature
+    )
+
+    result = extract_json_from_response(response)
+    ideas = result.get("ideas", [])
+
+    return {
+        "condition": "c1_direct",
+        "query": query,
+        "ideas": ideas,
+        "idea_count": len(ideas),
+        "metadata": {
+            "model": model,
+            "temperature": temperature,
+            "prompt_language": lang,
+            "mechanism": "direct_llm_generation"
+        }
+    }
+
+
+# For testing
+if __name__ == "__main__":
+    import asyncio
+
+    async def test():
+        result = await generate_ideas("Chair")
+        print(f"Generated {result['idea_count']} ideas:")
+        for i, idea in enumerate(result['ideas'], 1):
+            print(f"  {i}. {idea}")
+
+    asyncio.run(test())