How EpiK Builds the Web3 Knowledge Graph
In 2012, Google introduced the knowledge graph concept to make AI more intelligent. But how is this achieved? The key lies in transforming our existing knowledge into a format that machines can comprehend, known as the knowledge graph. In machine learning, content such as books, speech, and videos are challenging for machines to understand. However, when we convert them into structured triple data within a knowledge graph, machines gain the ability to comprehend and even perform reasoning based on that knowledge.
EpiK Protocol is an AI-first blockchain that strives to create an open and collaborative AI data network. EpiK has amassed over 2 billion AI data points throughout its three-year establishment and developed over 30 AI bots in more than 28 specialized domains. The knowledge graph plays a central role in our data accumulation efforts.
While constructing the knowledge graph, we discovered a significant gap in the web3 industry’s knowledge graph. The web3 sector, given its financial and transactional nature, presents numerous AI-driven use cases. An AI system with a deep understanding of web3, including up-to-date project information, pricing data, and trading indicators, can serve as an invaluable trading assistant, enabling users to make better market analyses. Achieving this necessitates a web3 knowledge graph that allows AI to comprehend and interpret the information directly.
The AI Data Construction Capability of EpiK Protocol
EpiK has built a comprehensive supply chain for collecting AI data. So far, more than 130,000 users have participated in AI data collection through our platform. Here’s an overview of the process:
Step 1: Domain Experts — EpiK Protocol has a team of contracted domain experts. For instance, if we require a web3 knowledge graph, we engage recognized web3 domain experts to define the necessary “entities” format within the EpiK backend.
Step 2: Knowledge Mainland — This is an AI data labeling game developed by EpiK Protocol where tasks assigned by domain experts are completed. Currently, over 130,000 users worldwide actively label AI data on this platform, accumulating 2 billion records of AI data across various specialized domains.
Step 3: Storage Network — The generated AI data is securely stored in a decentralized network. EpiK currently operates with over 267,734 storage nodes. Data miners can earn profits by storing AI data, while AI developers can access it anytime.
Currently, the web3 knowledge graph is still under construction. However, EpiK has already developed the most web3-savvy chatbot, ChatGPT, available in the market. We have gathered nearly 20 GB of web3 data, fed it to ChatGPT, and trained it for six months. This ChatGPT model is live on the AI BOT STORE (https://bot.epik-protocol.io/).
In 2024, EpiK will continue to build the Web3 knowledge graph. Just imagine a future where you have an investment assistant to whom you can directly ask questions like, “Please help me analyze the BTC 4H candlestick chart for the past week using indicators like MACD and Bollinger Bands. How do you predict BTC’s trend in the next 4 hours?” or “I’m considering investing in a new project called XXX. Can you analyze its market capitalization, codebase, and other factors to determine if it’s a high-quality project?”
Isn’t this future something worth looking forward to? Let’s go all-in on AI in 2024.
