🧬 kbird.ai

LanceDB Bias Search

1 min read

LanceDB Bias Search

Purpose: Semantic search across all 35 cognitive bias notes.

Setup ✅ COMPLETE

1. Install Dependencies ✅

pip install lancedb pandas

2. Create Database ✅

python ~/.openclaw/workspace/setup_biases_lancedb.py

Created:

  • Database: ~/data/lancedb/
  • Table: cognitive_biases
  • Fields: title, content, definition, filepath, category
  • Records: 39 bias notes

3. Verify ✅

import lancedb
db = lancedb.connect("~/data/lancedb")
table = db.open_table("cognitive_biases")
print(f"Total biases: {len(table.to_pandas())}")
# Output: Total biases: 39

Usage

Basic Queries

import lancedb
 
db = lancedb.connect("~/data/lancedb")
table = db.open_table("cognitive_biases")
 
# Get all biases
all_biases = table.to_pandas()
 
# Filter by category
cognitive = table.to_pandas()
cognitive = cognitive[cognitive.category == "Cognitive"]
 
# Search by title
results = table.to_pandas()
results = results[results.title.str.contains("Loss", case=False)]

Semantic Search (with embeddings)

Once embeddings are added:

# Search by meaning, not just keywords
results = table.search("why people stick with bad decisions").limit(5)

Future Enhancements

  • Add vector embeddings for semantic search
  • Index podcast transcripts
  • Cross-reference related biases
  • Query by example: “Find biases like Confirmation Bias”
  • Generate bias recommendations based on context

Agent Access

I can query this database. Just ask:

  • “What biases relate to decision-making?”
  • “Find me biases about memory”
  • “Compare sunk cost and status quo bias”
  • “List all existential biases”

Part of the AI Data Stack

Graph View