Skip to content

LORG - LLM-Powered Search Engine

LORG (Answer AI) is an AI-powered search engine that combines Large Language Models with knowledge graph generation to provide accurate, context-aware search results.

Key Features

  • LLM-Powered Answers: Get AI-generated answers to your queries using OpenAI models
  • Knowledge Graph Generation: Automatically creates structured knowledge graphs from queries
  • Multi-Engine Search: Aggregates results from Google, Yahoo, Bing, and DuckDuckGo via SearxNG
  • Relevance Scoring: Uses AI to score and rank search results by relevance
  • Multiple Interfaces: Use as a CLI tool, library, API server, or web application

How It Works

  1. Query Processing: Your query is sent to an LLM which generates an initial answer and knowledge graph
  2. Keyword Extraction: Keywords are extracted from the knowledge graph for web search
  3. Web Search: Multiple search engines are queried via SearxNG
  4. Content Analysis: Web pages are fetched and analyzed for relevance using AI
  5. Result Ranking: Results are scored and ranked by relevance to your query

Example Usage

lorg "What is quantum computing?"

import LorgSearch from 'lorg';

const searcher = new LorgSearch(process.env.OPENAI_API_KEY);
const results = await searcher.search('What is quantum computing?');
console.log(results.answer);

lorg server -p 3000

# Then make requests:
curl -X POST http://localhost:3000/search \
  -H "Content-Type: application/json" \
  -d '{"query": "What is quantum computing?"}'

Requirements

  • Node.js 18 or higher
  • OpenAI API key
  • Internet connection for web search

License

LORG is released under the MIT License.