The Universal NFT Vector Database: A Scaleable Vector Database for NFT Similarity Matching

TL;DR

This paper introduces a scalable, off-chain vector database system for NFTs that enables similarity matching and analysis across multiple blockchains, addressing duplication issues and aiding non-technical users.

Contribution

The authors developed a modular, extendable, cloud-based NFT vector database with a visualization GUI, facilitating cross-chain NFT similarity analysis and data aggregation.

Findings

Created a vector representation of NFTs based on ERC-721 standards.

Built a cloud-centered, extendable NFT processing system.

Developed a user-friendly visualization dashboard.

Abstract

Non-Fungible Tokens (NFTs) are a type of digital asset that represents a proof of ownership over a particular digital item such as art, music, or real estate. Due to the non-fungible nature of NFTs, duplicate tokens should not possess the same value. However, with the surge of new blockchains and a massive influx of NFTs being created, a wealth of NFT data is being generated without a method of tracking similarity. This enables people to create almost identical NFTs by changing one pixel or one byte of data. Despite the similarity among NFTs, each NFT is assigned a completely different token ID. To address the NFT duplication issue, we developed a modular, easily-extendable, hardware-agnostic, cloud-centered NFT processing system that represents NFTs as vectors. We established a database containing a vector representation of the NFTs in accordance with the Ethereum Request for Comment 721 (ERC-721) token standards to initiate the process of aggregating NFT data from various blockchains. Finally, we developed an NFT visualization dashboard application with a user-friendly graphical user interface (GUI) to provide non-technical users access to the aggregated NFT data. The Universal NFT Vector Database is an off-chain framework for NFT data aggregation based on similarity, which provides an organized way to query and analyze NFT data that was previously unavailable through on-chain solutions.