web_interface/README.md

ConGr Visualizer Web Interface

A web-based interface for exploring and visualizing ConGrs (Consensus Graphs) from research datasets.

Features

Browse Existing Graphs

• Dataset Selection: Choose from available datasets (BIO, FP, HIST, REFS, MATH, AIME)
• Entity Selection: Browse entities within each dataset
• Model Information: See which language models were used for each graph
• Graph Visualization: Interactive network visualization using vis.js
• Metadata Display: View graph statistics and consensus text

Create New Graphs

• Text Input: Enter multiple text sequences to create new ConGrs
• Real-time Visualization: See the graph structure as it's created
• Save Functionality: Save created graphs to pickle files

Available Datasets

• BIO: Biography datasets with various public figures
• FP: False Presupposition datasets
• HIST: Historical events datasets
• REFS: Reference datasets
• MATH: Mathematical problem datasets
• AIME: American Invitational Mathematics Examination datasets

Models

The graphs are generated using various language models:

• olmo7b
• qwen72b
• llama70b
• llama8b

Installation

1. Install dependencies:

pip install -r requirements.txt

2. Start the server:

python server.py

3. Open your browser and navigate to:

http://localhost:8080

Usage

Browsing Existing Graphs

1. Select Dataset: Choose a dataset from the dropdown menu
2. Select Entity: Choose an entity from the available options
3. View Graph: The graph will be automatically loaded and displayed
4. View Information: Graph metadata and consensus text will be shown

Creating New Graphs

1. Enter Text: Input multiple text sequences (one per line)
2. Create Graph: Click "Create Graph" to generate a new ConGr
3. Save Graph: Optionally save the graph to a pickle file

API Endpoints

• GET /api/datasets - Get available datasets
• GET /api/entities?dataset=<dataset> - Get entities for a dataset
• POST /api/load_existing_graph - Load an existing graph
• POST /api/create_graph - Create a new graph from text sequences
• POST /api/save_graph - Save a graph to file

Testing

Run the test script to verify the server is working correctly:

python test_server.py

File Structure

web_interface/
├── server.py          # Flask server
├── index.html         # Web interface
├── requirements.txt   # Python dependencies
├── test_server.py     # Test script
└── README.md         # This file

Graph Information

When viewing a graph, you can see:

• Dataset: The source dataset
• Entity: The specific entity or topic
• Model: The language model used
• Sequences: Number of input sequences
• Nodes: Number of nodes in the graph
• Edges: Number of edges in the graph
• Consensus: The consensus text generated from the graph

Visualization Features

• Hierarchical Layout: Graphs are displayed in a hierarchical structure
• Color Coding: Consensus nodes are highlighted in green
• Interactive: Zoom, pan, and hover for more information
• Responsive: Works on desktop and mobile devices