Viveka1zha/neuro-symbolic-search

Neuro-Symbolic Search Model

A hybrid search system combining neural embeddings with symbolic reasoning.

Features

• 🧠 Neural semantic search using sentence transformers
• 🔧 Symbolic rule-based filtering
• 🚀 Fast and lightweight
• 📊 Supports metadata filtering

Quick Start

from model import SimpleNeuroSymbolicSearch

# Initialize
search = SimpleNeuroSymbolicSearch()

# Index documents
documents = ["Document 1", "Document 2", "Document 3"]
metadata = [
    {'topic': 'ai', 'level': 'beginner'},
    {'topic': 'ml', 'level': 'advanced'},
    {'topic': 'ai', 'level': 'intermediate'}
]

search.index_documents(documents, metadata)

# Search with filters
results = search.search("artificial intelligence topic:ai", top_k=5)

for r in results:
    print(f"{r['rank']}. {r['document']} (score: {r['score']:.3f})")

Search Syntax

• Simple search: machine learning
• With filters: machine learning topic:ml level:beginner
• Boolean operators: data science topic:ml AND level:advanced

Model Details

• Base Model: all-MiniLM-L6-v2
• Embedding Dimension: 384
• Search Method: Cosine Similarity + Rule Filtering

Requirements

torch>=2.0.0
sentence-transformers>=2.2.0
transformers>=4.30.0

Citation

@misc{neuro-symbolic-search-2025,
  title={Simple Neuro-Symbolic Search},
  author={Your Name},
  year={2025},
  url={https://huggingface.co/YOUR_USERNAME/neuro-symbolic-search}
}