Search Capabilities

Learn how to search your memory effectively.

Key-Based Lookup

Fast, exact matching using unique keys:

researcher = memory.get("Yann LeCun")
if researcher:
    print(f"Found: {researcher.name}")

Performance: O(1) - Constant time, blazingly fast.

---

Semantic Search

Find entities by meaning, not exact keys:

memory.build_index()
results = memory.search("deep learning research papers", top_k=5)

for result in results:
    print(result.name, result.research_focus)

Performance: O(n) similarity computation + FAISS optimization.

Building the Index

# Build from scratch
memory.build_index()

# Force rebuild (useful after many updates)
memory.build_index(force=True)

Search Parameters

results = memory.search(
    query="artificial intelligence",  # Natural language query
    top_k=10                              # Return top 10 results
)

• query: Natural language string describing what you're looking for
• top_k: Number of top results to return (default: 5)

---

Combining Search Methods

# First, narrow with semantic search
candidates = memory.search("machine learning", top_k=20)

# Then, verify with exact lookup
specific = memory.get("Yann LeCun")

# Filter results
active_researchers = [r for r in candidates if r.is_active]

---

Search Best Practices

1. Use semantic search for discovery - Find conceptually related items
2. Use key lookup for known items - Direct access is faster
3. Refine query terms - Better queries = better results
4. Adjust k based on needs - Start with k=5, increase if needed

See more in Advanced Usage.