Innovating Blockchain Analytics with AI and Graph Theory

GraphiQ combines state-of-the-art algorithms and blockchain technology to provide real-time transaction analysis. Inspired by advanced research in graph convolutional networks (GCNs) and anomaly detection, GraphiQ delivers transparency, security, and actionable insights.

The Core Technologies

A. Graph Analytics and AI

• We represent blockchain transactions as a directed graph, where nodes correspond to transactions, and edges represent transaction flows. Each node is enriched with 166 features, including time, volume, inputs/outputs, and aggregated neighbor statistics.

• GCNs are employed for node classification tasks, leveraging spectral graph convolutions to aggregate features from neighboring nodes and identify patterns in transaction flows.

• Formula:

• EvolveGCN: To capture temporal dynamics, EvolveGCN links multiple GCN layers using a recurrent neural network (RNN), evolving node embeddings over time.

• Random Forest Models: For robust classification, we use Random Forests in conjunction with graph embeddings, balancing accuracy and interpretability.

B. Blockchain Integration

• Built on Solana, GraphiQ takes advantage of its high throughput and low-latency architecture to process transactions efficiently. Smart contracts are employed for seamless data querying and GRQ token functionality.

C. AI-Driven Anomaly Detection

• We employ supervised learning techniques like Logistic Regression and Random Forests alongside GCNs to classify transactions as licit or illicit. Weighted loss functions address class imbalances, prioritizing detection of illicit transactions.

Product Architecture

A. Data Pipeline

1. Live transaction data is pulled from Solana’s blockchain and transformed into a graph structure.

2. Features are derived from raw blockchain data and aggregated over graph neighborhoods for contextual insights.

3. GCNs generate embeddings that encapsulate local and global graph structure

B. AI Processing Layer

• Graph embeddings feed into the classification models (Random Forests, Logistic Regression) for risk assessment.

C. Visualization and User Interface

• Results are rendered as interactive dashboards or JSON-based outputs for programmatic access.

4. GRQ Token Ecosystem

• The GRQ token underpins our product’s functionality, serving as the gateway to premium analytics and user incentives. It is also utilized for transaction fee payments and as rewards for contributing data to improve the system.