CONTEXTO.AI

System v2.1 // GloVe-100d Semantic Space Exploration & Neural Navigation System

🌌 Overview

Contexto.AI is an advanced semantic search engine disguised as a game. It challenges users to locate a specific target word within a 40,000-dimensional semantic vector space. Unlike traditional word games that rely on spelling or dictionary definitions, Contexto.AI operates on meaning.

Powered by GloVe (Global Vectors for Word Representation) embeddings, the system calculates the cosine similarity between your guess and the secret target word. A proprietary algorithm ranks your proximity, guiding you through the manifold toward semantic convergence.

This repository features a state-of-the-art AI Solver capable of navigating this high-dimensional space autonomously using centroid pruning strategies.

---

🚀 Key Features

🧠 Semantic Vector Engine

At the core lies ai_solver.py, utilizing NumPy and SciPy to perform lightning-fast vector mathematics. It loads pre-trained GloVe embeddings (100 dimensions) to map the relationships between 40,000 English words.

🔮 3D Manifold Visualization

The application renders a real-time, interactive 3D Scatter Plot of the semantic space using Plotly.js.

• Dynamic Point Cloud: Watch as thousands of semantic nodes are plotted in 3D space.
• Trajectory Tracking: Visual lines connect your guesses, showing your path through the word galaxy.
• Target Highlights: The system reveals the exact coordinates of the hidden word upon victory.

🤖 Autonomous AI Co-Pilot

Includes a fully integrated AI agent that plays alongside you—or for you.

• Auto-Play Mode: Watch the AI autonomously deduce the target word.
• Centroid Pruning Strategy: The AI calculates the centroid of previous best guesses to triangulate the target vector.
• Transparent Reasoning: The AI explains its "thought process" for every guess (e.g., "Fits constraint from 'apple' (Sim: 0.72 vs Exp: 0.75)").

Full Technical Breakdown: How the AI Works ->

🎨 Cyber-Futuristic Interface

A premium, dark-mode UI designed for immersion.

• Neon-accented aesthetics.
• Responsive layout with fixed sidebars.
• Real-time feedback loops.

---

📸 Demo

Figure 1: The 3D Manifold visualizing a cloud of semantic neighbors around the target word.

---

🛠️ Installation & Setup

Prerequisites

• Python 3.9+
• Pip
• Virtual Environment (recommended)

1. Clone the Repository

git clone https://github.com/ryan-rudd/contexto-ai.git
cd contexto-ai

2. Install Dependencies

python3 -m venv venv
source venv/bin/activate
pip install -r requirements.txt

3. Download Model Data

The system requires the GloVe 100d model. (Note: Ensure embeddings.bin and vocab.json are present in the root directory. If utilizing standard GloVe files, run python scripts/setup_model.py to convert them).

4. Launch the System

python app.py

Access the interface at http://localhost:8080.

---

🕹️ Controls

• Input Field: Type any English word to probe the semantic space.
• Execute AI Guess: Trigger the AI to make a single optimal move.
• Auto-Play: Toggle the AI to solve the game automatically.
• Orbit/Pan/Zoom: Interact with the 3D graph to explore word clusters.

---

📐 Technical Architecture

Backend (Flask):

• app.py: RESTful API endpoints for game state and vector calculations.
• ai_solver.py: Core logic for vector math, candidate pruning, and next-best-guess selection.

Frontend (HTML5/JS):

• index.html: Single Page Application (SPA) architecture.
• Plotly.js: WebGL-powered 3D rendering.
• Fetch API: Asynchronous communication with the Python backend.

---

📜 License

Distributed under the MIT License. See LICENSE for more information.