WhoDis

Our project focuses on retrieving and analyzing transaction history data of a NEAR blockchain address. This process begins by fetching the wallet's transaction history, capturing all incoming and outgoing activities, such as transfers, token swaps, and other on-chain events. Once this data is collected, we extract a list of all other wallet addresses involved in the transactions, and fetch metadata about each address, like whether it's a smart contract, if it belongs to a known Dapp, and if it does, then to which category does that project belong (DeFi, NFT, tooling...). After fetching all required data, we pass it to an LLM (in this case: gpt-4o) to systematically analyze the transactions and identify patterns, and behaviors that reflect the user's activities and interests. This enables us to generate a detailed description that accurately characterizes the user or wallet's primary activities, such as whether they are involved in frequent trading, liquidity providing, participating in decentralized finance (DeFi) platforms, or any other blockchain-based activity. By integrating both raw transaction data and cross-checked insights from relevant external sources, we create a more refined and precise understanding of how the wallet is being used. The objective of this project is to provide users or stakeholders with a clear, data-backed representation of a wallet’s transactional footprint.

We built this project using a Python backend (Flask) and a HTML+JS+CSS web interface. When the user inputs a wallet address, a Python script retrieves transaction data via APIs, processes and analyzes it to map user activity based on transaction patterns. External data from the PikesPeak API is integrated to enhance the analysis, and the backend serves the processed information to the frontend, which displays a behavior profile.