Portal AI

A framework for applying Large Language Models (LLMs) to A2F Knowledge Portals, specifically designed to work with PIGEAN results.

Table of Contents

• Overview
• Components
  • KG Ingress
  • Portal AI Frontend
  • Portal Tools
• Getting Started
• Development
• Contributing

Overview

This repository processes PIGEAN results and builds knowledge graphs for AI-powered portals. The project consists of three main components:

1. KG Ingress: Processes PIGEAN results into a knowledge graph
2. Portal AI Frontend: Next.js application for interacting with the knowledge graph
3. Portal Tools: Tool server that exposes portal endpoints as LLM tools

Components

KG Ingress

The knowledge graph ingestion package processes PIGEAN results into a structured knowledge graph using LinkML schemas.

Prerequisites

• Python 3.9+
• LinkML (pip install linkml)

Model Generation

The knowledge graph schema is defined in portal-model.yaml using LinkML. To generate the Python model classes:

pip install linkml
gen-python portal-model.yaml > kg-ingress/kg_ingress/models/portal_model.py

Building the Knowledge Graph

1. Navigate to the kg-ingress directory
2. Follow the README instructions for setting up the ingestion pipeline
3. Process your PIGEAN results through the pipeline to build the knowledge base

Portal AI Frontend

A modern Next.js application built with Tailwind CSS and shadcn/ui components for interacting with the knowledge graph.

Prerequisites

• Node.js 18+
• pnpm

Installation

cd portal-ai
pnpm install

Development

# Start the development server
pnpm dev

# Build for production
pnpm build

# Start production server
pnpm start

Portal Tools

A tool server that wraps various portal endpoints and exposes them as tools for LLMs through the Portal AI frontend.

Prerequisites

• Python 3.9+
• uv (Python package installer and resolver)
• Docker (optional)

Installation

cd portal-tools

# Install dependencies using uv
uv pip install -e .

# Or if you want to create a virtual environment first
uv venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate
uv pip install -e .

Running the Server

# Using Python directly
python -m portal_tools/app.py

# Using Docker
docker build -t portal-tools .
docker run -p 8000:8000 portal-tools

Contributing

1. Fork the repository
2. Create a feature branch
3. Commit your changes
4. Push to the branch
5. Create a Pull Request