Dialogue EpiK: Knowledge Graphs are the Essential Path for AI to Understand Human Knowledge

The knowledge graph, as a data format for AI in education to learn human-understood knowledge, is rapidly driving the application and practice of artificial intelligence, and the development space of the knowledge graph industry will gradually expand. EpiK Protocol is known as the world's first distributed storage project for AI data, and with the mainnet launch approaching, it has attracted widespread attention in the industry.

In this issue, EpiK ecological ambassador and founder of an AI application technology company, Principles, shared detailed and exciting insights on topics such as EpiK's development roadmap, the construction and advantages of knowledge graphs, innovative application scenarios, and the incentive plan for the mainnet launch. This is believed to help and inspire you to gain a deeper understanding of knowledge graphs and their prospects.

Chain Catcher: Please introduce EpiK Protocol and its specific practical explorations in the industry. What is the current development progress?

Principles: EpiK Protocol, the Mingzhi Protocol, is the world's first distributed storage protocol for AI data. By integrating IPFS storage technology, a token incentive mechanism, and a DAO governance model, it creates a globally open autonomous community with four core capabilities: trusted storage, trusted incentives, trusted governance, and trusted finance. This allows for the organization of global community users to collaborate with extremely low management costs, continuously generating high-quality AI data that can be co-built, shared, and mutually beneficial, broadening AI's cognition and promoting the arrival of a fully intelligent era.

Currently, EpiK Protocol has completed five rounds of testnet testing over the past year and is preparing for the mainnet launch on August 15. During the testnet phase, rewards for both the labeling system and storage system of EpiK Protocol came from foundation subsidies, with limited incentives.

With the launch of the mainnet, an economic model ten times that of the testnet will officially start, significantly increasing the speed at which EpiK Protocol collects AI data. There will be more ways for the community to participate in EpiK Protocol, continuously creating real value for the AI and blockchain industries.

Chain Catcher: What specific technical solutions does the Mingzhi Protocol EpiK adopt in distributed storage?

Principles: The technical solution of the Mingzhi Protocol is based on the modification of Filecoin's Lotus. We have adopted Filecoin's proof of replication and proof of spacetime schemes to ensure the fairness of input and output for storage devices, eliminating issues such as sybil attacks.

The difference is that we have two systems for labeling and storage. During the process of deploying machines to participate, users can also feel the obvious differences between us and Filecoin. At the same time, we have modified the penalty and incentive mechanisms of the storage system, reducing the cost of participating in EpiK storage machines, enhancing the decentralization and security of the entire network's data, and significantly increasing the network's TPS throughput and node scale.

In addition, we have added an incentive mechanism for the labeling system, integrating data production and data storage into one protocol, ensuring that there are no intermediaries profiting from price differences. Another innovation is that we have introduced the Coinbase mechanism from the Bitcoin technology system to EpiK, allowing for automatic aggregation of earnings and tokens to be staked, which has indirectly led to the emergence of the first open mining pool technology solution for distributed storage within the EpiK ecosystem.

This last point is very important, as it is expected to make the mining of PoC projects "foolproof," allowing EpiK to have a broader grassroots foundation.

Chain Catcher: EpiK Protocol is known as the world's first "AI Data + Distributed Storage" solution. Why combine the two? What specific pain points does it address?

Principles:

• Compared to distributed storage projects represented by Filecoin, EpiK Protocol not only possesses Filecoin's storage capabilities but also adds an AI data processing collaborative network, continuously supplying high-value effective AI data to the distributed storage ecosystem.
• In response to the current state of the distributed storage ecosystem, EpiK Protocol has pioneered a new business model for distributed storage, upgrading from providing paid storage for enterprises to providing lower-cost, continuously updated AI data.
• Compared to traditional AI data labeling, EpiK Protocol is the world's first distributed storage protocol for AI data, transforming the data labeling industry chain into a data-sharing economic model, greatly reducing the cost of data labeling.

Currently, with algorithms not yet optimized, the actual labeling efficiency is already ten times that of traditional labeling efficiency. The team has developed a gamified knowledge crowdsourcing product called Knowledge Continent, which launched its beta version on April 22 this year. Users can collect and classify data through gamified quizzes.

Download to contribute your knowledge and earn EPK

Chain Catcher: Can you explain to readers what a knowledge graph is? Will your final product provide services similar to Baidu Encyclopedia or Wikipedia?

Principles: A knowledge graph, like text, video, images, and voice, is a medium for knowledge transmission. We can teach knowledge to each other through language, text, video, and images, just like teachers educate students through classes, books, and recorded videos.

However, humans cannot educate AI robots in this way because they lack cognitive abilities and cannot understand the meaning of this data. We need to find a data format that AI can understand to educate AI about human-understood knowledge, and this special data format is the knowledge graph. It can be said that the knowledge graph is the necessary path for AI to understand human knowledge and become general AI.

For example, with a medical knowledge graph, we can cultivate an AI medical assistant that can provide intelligent consultation services for everyone; with an insurance knowledge graph, we can train an AI insurance assistant that can provide intelligent underwriting and coverage for every family; with a university knowledge graph, we can train an intelligent recommendation assistant for university majors that serves every graduate.

With this understanding, we can answer that the effective data provided by EpiK Protocol will not be data meant for human viewing like Baidu Encyclopedia and Wikipedia, but rather knowledge graph data labeled from different fields of human knowledge for robots to see. These data will unleash tremendous value through AI's powerful computing capabilities.

Chain Catcher: How will you build knowledge graphs in the future? What are your advantages?

Principles: The Mingzhi Protocol currently has tools that connect the community of domain experts in knowledge graphs (whose main work is to define the data format of knowledge graphs in different fields and establish acceptance rules for knowledge graph data), the bounty data labeling community (whose main work is to help domain experts complete specific knowledge graph data collection and acceptance), and the data storage community (whose main work is to store knowledge graph data on the mainnet and provide paid access externally).

These tools enable us to complete a data-sharing economic operation system for the first time under the premise of high mutual trust among all parties, allowing data contributors to continuously enjoy the monetization benefits of future data. In our practical tests, this mechanism has already easily achieved a cost advantage ten times that of traditional data labeling. The significant reduction in costs will provide a continuous driving force for the future development of AI.

As top AI experts have said, in the process of building AI rockets, neural network algorithms are the rocket engines, and data is the fuel.

Since 2006, a new generation of AI research represented by deep learning has emerged, and the world has cultivated AI talent for a full 15 years. Correspondingly, various deep learning algorithms and frameworks have also been studied for 15 years, yielding fruitful results. The rocket engine does not face bottlenecks; the real bottleneck may shift to data. The cost and collaboration advantages of the Mingzhi Protocol will play a significant role in the future implementation of AI technology.

Chain Catcher: How is EpiK's token economic model designed? Why is it designed this way?

Principles: The principle behind the design of EpiK's token economic model is that while each role in the ecosystem pursues its own maximum benefit, they can form a synergy to jointly promote the overall growth of the project. In the EpiK ecosystem, there are several key roles: domain experts, bounty hunters (users of the Knowledge Continent app), storage miners, and knowledge gateways (AI companies).

Each role performs its duties, forming a complete ecological closed loop. Among them, domain experts and bounty hunters (users of the Knowledge Continent app) are responsible for the production of effective data; storage miners are responsible for storing data; knowledge gateways connect data users and miners.

This ensures the effectiveness and usability of each piece of data. According to the economic white paper, after the mainnet launch, 332,000 EPK will be produced daily. Of this, 75% goes to miners, which is 248,000 daily, 9% to domain expert teams, 1% to voting users, and 15% for bandwidth rewards or knowledge funds.

The token price is calculated based on a reward of 0.35 USD (currently over 1 USD), allowing miners to break even within 90 days, domain experts to have a knowledge fund of 5 million available annually, and voting users to receive a 7% annualized return in token terms. The remaining 15% is interestingly set to be shared between bandwidth rewards and the knowledge fund.

Here’s an analogy: when you run a library with many people and high-quality books, I would prefer to invest more money in tables and chairs to enhance the experience for visitors. This represents that when the quality of AI data is high and more AI companies use it, a larger proportion will be allocated to bandwidth rewards (public IP miners) to keep the bandwidth stable; conversely, when the library has few visitors and poor-quality books, more money should be spent on acquiring books. This represents the knowledge fund, specifically placed on the Knowledge Continent app to attract more users to participate in the co-construction of the AI data knowledge graph.

For more detailed information on the economic model, you can refer to the image below:

Chain Catcher: With the EpiK mainnet about to launch, what incentive plans will be introduced to the community? How will you enhance community consensus?

Principles: First, after the mainnet launch, the incentives for each role will be significantly increased. During the testnet phase, the total daily earnings for storage nodes were 10,000 EPK, while after the mainnet launch, it will be 248,000 EPK; during the testnet phase, the total daily earnings for bounty hunters answering questions on Knowledge Continent were 500 EPK, while after the mainnet launch, it will be 20,000 to 40,000 EPK.

Additionally, after the mainnet launch, we will also initiate the EpiK Protocol Grants ecological support plan, which includes recruiting storage and knowledge ecological ambassadors from various regions around the world, supporting ecological tools, and supporting EPK DeFi application scenarios. We will also introduce more gameplay to Knowledge Continent, such as NFT Buff bonuses and open naming rights for Knowledge Continent buildings.

Chain Catcher: What do you think are the main challenges EpiK will face next? How do you plan to respond?

Principles: I believe the main challenges are the efficiency of collaborative tools, which can be broken down into the usage threshold for domain experts and the efficiency of the task generation engine. Currently, our data collection costs are already far lower than those of traditional labeling companies, but to further maximize the labor of data contributors toward tangible monetization, we need to continuously optimize data collection efficiency.

The lower the usage threshold for domain experts, the more domain experts can participate, and the faster the expansion of knowledge fields will be. To achieve this, we will initially train some personnel to teach domain experts how to accurately define and accept the data they want using our system, and continuously iterate the product based on their feedback.

The goal of the task generation engine is to issue tasks more precisely to meet the data collection needs of domain experts. We have already established partnerships with several laboratories that have strong algorithm reserves in various niche fields.

Chain Catcher: What is EpiK's roadmap for the next year? What is the long-term vision?

Principles: First, this month we will launch EpiK Protocol Mainnet 1.0 "Rosetta." It is named Rosetta because Rosetta was the first time humanity understood ancient Egyptian writing, opening the door to ancient Egyptian civilization; it is a monument that breaks the dimensional wall. EpiK's goal is to enable AI to understand human civilization, opening the door for AI to comprehend human civilization, marking a milestone in breaking the dimensional wall.

In the coming year, the EpiK team will focus on launching EpiK Protocol Mainnet 2.0 "Hammurabi," which will primarily introduce all functions related to EpiK DAO governance, initiating community autonomy. This version is named after the first written legal code in human history, "Hammurabi." After this, the EpiK team will focus on the knowledge ecosystem, expanding more knowledge fields, increasing the number and granularity of domain experts, and deepening the knowledge graphs in various fields.

We will connect with more data demanders and integrate legal fiat exchange services to help data demanders acquire the data they need at lower fiat costs.

On a smaller scale, the EpiK team will provide more DeFi or CeFi services for EPK, such as cross-chain, lending, and options; additionally, the EpiK team will integrate more gameplay into Knowledge Continent to make data labeling less tedious.

In the long term, EpiK's goal is very clear: to continuously transform human knowledge in various fields into knowledge graph data that AI can understand, broadening AI's cognition and promoting the early arrival of the cognitive intelligence era.

Chain Catcher: As the blockchain industry develops rapidly, what other innovative application scenarios and technological innovations do you think are worth paying attention to?

Principles: Besides the distributed storage track, I am also keeping a close eye on DeFi and the Metaverse. EPK, as an asset, naturally involves not only the demand to participate in the EpiK ecosystem but also various derivatives, such as the highly requested lending services.

Additionally, Knowledge Continent, as an educational and entertaining knowledge crowdsourcing product, was designed from the outset as a game world based on land and buildings, which naturally combines with the Metaverse and NFTs. This is not a forced attempt to ride the trend but a natural horizontal expansion.

In the Metaverse and NFT tracks, there are already some well-validated tools and mechanisms in the market. It is reported that the EpiK team is also in close communication with leading players in these tracks to integrate these gameplay elements into Knowledge Continent.