How to use AI to gain insight into the Ethereum market

In this article, we explore the huge potential of systems AI based on the Large Language Model (LLM) in its application to the dynamic analysis of the Ethereum market. For ease of understanding, we use the AI ​​analysis of the Ethereum Non-Fungible Token (NFT) market as an example.

All in all, using AI to get information about Ethereum requires three steps. First of all, vital Ethereum data to be collected metadata on the network and related off the network.

Next, you must build a specialized database designed for LLM. The final step is to apply the LLM-based augmented generation (RAG) recovery approach to Analyze data and get information.

Collecting NFT data online and offline

In the context of NFT, data collection requires deepening both on and off the network.

OpenSea, one of the leading NFT marketplaces, provides fertile ground for on-chain data miningsuch as NFT transaction details and metadata.

The data collection process can be done using the OpenSea API documentation, a simple approach to accessing data on-chain. At the same time, offline data such as NFT images or videos, always stored in the Interplanetary File System (IPFS)decentralized storage network.

The first step is to identify the IPFS hash directed to the content., a unique identifier that points to the NFT image in IPFS. This hash is usually found in the NFT metadata, or is part of the transaction details.

The next step is to create the HTTP Gateway URL. Equipped with an IPFS hash, a URL can be generated which then allows an HTTP request to be sent. Tools like Axios or the built-in fetch function are perfect helpers for sending an HTTP GET request to a generated URL. thus restoring the NFT image data.

Create an LLM Knowledge Database

Armed LLM with the right data to work effectively, it is important to create a comprehensive knowledge base. This knowledge database will serve as a trusted resource for semantic search, allowing the most relevant data to be identified and retrieved. Hence the LLM is provided with the correct context, thus making it easy to generate accurate output for your request.

After the data is received in and out of the chain, it is necessary to systematically clean and systematize the accumulated data.

This process includes identifying relevant properties and attributes that are integral to the NFT analysis, such as categories, design studios, intellectual property owners, and sales history.

There are many methods for extracting these properties from images., for example, using NFT metadata or free image recognition programs. Once the data features have been successfully extracted and encoded, we can quickly create a custom LLM database.

Used an LLM-based AI model to get an idea of ​​NFT.

Relying solely on LLM to generate the actual text, or even matching the model to their database, is not necessarily a factually accurate answer.

That’s why, We propose a RAG approach to analysis data by NFT. RAG introduces a methodology that separates the knowledge database from the linguistic model.

This methodology involves asking or sending a request to an AI agentsuch as queries related to new NFT trends, individual properties or correlations, and relationships between property attributes and NFT market performance.

A search algorithm, such as Azure Cognitive Search, then checks the most relevant text in the knowledge database that is likely to contain the required answer. At the last stage, a brief message is sent to the LLM in the form of instructions for the LLM along with the text of the corresponding document.

then the model uses this information to formulate a response that matches the original query.guaranteeing a fact-based and contextually meaningful result.

AI will become a powerful tool in the blockchain industry

In this article, the Ethereum NFT analysis is used as a case study to demonstrate the effectiveness of using an LLM-based artificial intelligence methodology to gain blockchain insights.

Due to its versatility, this method has the potential for widespread adoption in the blockchain arena. We expect that in the future there will be many LLM tools to improve and facilitate various aspects of blockchain analysis and operation.

*According to information from Forbes USA