Exclusive Interview with Thomas Emmerich, Director of Epik Knowledge Ecosystem “Building an Everlasting Knowledge Vault”


 1,060 total views

Blockcast.cc: Dear all, BTC has gone back up to $56,000 and Eth is at its all-time high too. The crypto industry is alive and moving fast. Blockcast.cc has captured all these moments since 2017.

My name is Keith Phua, Chief Commercial Officer at Blockcast.cc

Blockcast.cc is a broadcasting news source for the blockchain community. Founded in Singapore. We have a network of partners in China, Hong Kong, South Korea, Singapore and Malaysia.

We have also helped global blockchain and cryptocurrency companies reach their marketing, public relations and advertising goals since 2017.

We have been interviewing interesting personnel in the blockchain space. Today we have Thomas Emmerich, Director of Epik Knowledge Ecosystem.

Blockcast.cc: Hi Thomas, can you, would you please briefly introduce the Epik project and team background?

Thomas: EpiK is a compound word, and its full name is Epigraphy Knowledge, meaning the knowledge engraved on the stone, which is also the original intention of the project, that is, through the form of sharing economy, processing human’s comprehensible knowledge into knowledge comprehensible for the machine, and stored on the blockchain eternally for the machine to read safely and credibly and apply to various artificial intelligence applications.
For friends not engaged in the artificial intelligence industry, they may not understand the differences between knowledge understandable to humans and knowledge understandable to machines, here let me explain this a little! Current human knowledge whose carrier are mostly articles, pictures, video, audio and encyclopedias, are collectively called unstructured data by us. Human brain is very magical, after reading the book Three Kingdoms, they can conclude that Cao Cao likes widows . Such reasoning represents a cognitive ability, and if we want to deduce such a conclusion, we must first know who Cao Cao is? What does widow mean? In what ways does one person like another? (Here you can imagine the plot of the Three Kingdoms)? The one by one answers to these front questions is called background knowledge. How do machines acquire background knowledge? The current industry-wide accepted solution is the knowledge graph. So what EpiK does is extracting the knowledge from the carrier of human knowledge (such as articles, pictures, videos, audio, etc.) and processing it into a special data format that the machine can understand, for the machine to use for building its own background knowledge, and then the machine can provide more intelligent services based on such background knowledge. For example, combing human insurance knowledge, after learning this knowledge machines can not only provide us with an intelligent recommendation of insurance strategy but also guide us to underwrite more efficiently. The knowledge graph is like a neuron, the denser it is, the smarter it is. We can work out a super-powerful background knowledge base and turn Siri around us into Jarvis, the intelligent assistant of Iron Man.

EpiK was first born at the National University of Singapore. During the outbreak of the epidemic in March last year, James had a preliminary idea, and quick found Leo and Sigrid. They thought that the construction of knowledge graph should not only storing hash value of data on the chain, but also storing all data on the chain. After seeing IPFS, Filecoin, Arweave and Swarm, they found it impossible to embed the economic model of knowledge graph collaborative network seamlessly into them, so they finally decided to build a two-layer network for it on Filecoin based on Filecoin’s replication proof and space-time proof strategy. After the outbreak was alleviated, the founding team went to Shenzhen in June last year to learn about the current state of decentralized storage ecology. Feeling the time is ripe, they officially launched the EpiK Protocol. Later, more excellent partners joined us, after nearly a year of hard work, we finally had a chance to come to blockcast.cc and introduce our progress to you.


Blockcast.cc: We understand that EpiK’s goal is to lead all community participants to share domain knowledge graph data, to expand AI awareness, and to go on a different track of storage. What are the differences between Epik and other storage projects such as IPFS, and what are their characteristics and advantages?

Thomas: IPFS is an unstimulated distributed data transfer protocol while EpiK has an excitation layer, I will talk more about the difference between EpiK and the Filecoin which also has a motivation layer.

First of all, from the technical level, current incentive mechanism of Filecoin emphasizes the stability of single file stored in one copy, and the storage incentive is far greater than the bandwidth incentive, which is more suitable for the cold storage of large files on the whole; while EpiK’s incentive mechanism emphasizes multiple backup storage of a single file, and the storage incentive and bandwidth incentive are closely related, which is more suitable for the hot storage of small files.

Secondly, from the application level, Filecoin focuses on meeting anyone’s any storage needs, and has no requirement for data quality, as long as the storage cost is paid, it can be used to store either encyclopedias or dirty data; EpiK focuses on meeting the orderly storage requirements in specific fields, and has clear requirements for data quality, storage is completely free, but to qualify for storage requires support by voting in community.
EpiK and Filecoin complement each other. EpiK collects ordered knowledge graph type small log files in each field, giving full play to the advantages of EpiK small file hot storage; and then periodically restore the small log files into small snapshot files to uploaded to Filecoin for snapshot backup, giving full play to the advantage of Filecoin large file cold storage, and then continue to provide effective data for Filecoin. So we say that EpiK is the knowledge graph data folder on Filecoin large file system, and is the layer2 network of Filecoin in the knowledge graph vertical domain.


Blockcast.cc: Knowing that EpiK has a game-based knowledge graph data acquisition product called knowledge mainland, can you introduce what is knowledge mainland? What’s its working principle?

Thomas: It is complex to process human knowledge into the forms of knowledge graph which can be understood by machine. Many processes, such as middle collection, integration and disambiguation, can’t be fully automated and need a lot of manual annotation. Moreover, the annotation of knowledge graph data is more difficult than the traffic lights we commonly use in verification codes, for example, insurance knowledge graph data, people without insurance knowledge may not be suitable to participate in annotating such data. So EpiK designed a bounty hunter role that allows everyone to complete a series of certifications to unlock the role of a bounty hunter, where each person can complete a series of certifications to unlock markings tasks in different areas and participate in knowledge graph together. Knowledge mainland is the first app in the EpiK ecosystem for bounty hunter to complete the task of annotating missions, considering the marking work itself is a very boring job, in order to let everyone easily participate in annotating, knowledge mainland adopted a game-like way, the story background is that on the knowledge mainland, the human civilization were destroyed due to some external forces, need to depend on everyone to re-collect human civilization puzzles, and those explorers who participate in the restoration of human civilization can be awarded EPK tokens.

At present, knowledge mainland has started internal testing, in order to allow everyone to participate in the initial annotation with a threshold of zero, knowledge mainland chooses to release the first question bank of voice question bank that can be participated in without any professional knowledge, the questions for Chinese users are mainly ancient poems and numbers, while the questions for an overseas user are mainly film lines and numbers. These question banks are from our experts in the voice fields of openSLR. The EpiK Foundation is recruiting the first batch of domain experts, some of whom have been identified, covering the fields of digital economy, video games, health care, rural industrial structure and etc. They are also combing question bank in their fields, and these questions will be posted to the knowledge mainland to meet with you.

On the first day of the internal test of knowledge mainland, the Chinese and English communities have answered 11661 questions and collected thousands of voice data. The invitation function is under development and will start fission after the internal test is stable.


Blockcast.cc: What position does Knowledge mainland have in EpiK’s ecosystem and how does it help the project vision come to fruition?

Thomas: Knowledge mainland is the first annotation application in EpiK ecosystem, which has the significance of the milestone. But knowledge mainland will not be the only annotation App in the EpiK ecosystem. Different areas may have different requirements for annotation, some can be completed on the mobile phone, some may not. We welcome the introduction of more annotating tools in the EpiK ecology, and to help fields experts better distribute tasks, but also help bounty hunters complete tasks more efficiently.


Blockcast.cc: Can you tell us a little about the economic model of EpiK tokens and how the value of EPK token is represented?

Thomas: EpiK’s economic model is centred around four main roles within the ecosystem, and they are miners, domain experts, bounty hunters and knowledge gateways, respectively. The domain expert is responsible for organizing the production and acceptance of knowledge graph data, the bounty hunter is responsible for helping the domain expert to collect and process the knowledge graph data, the miner is responsible for storing and accessing the knowledge graph data, and the knowledge gateway is responsible for providing API services of the knowledge graph data.

75% of EPK daily mined by the whole network is awarded to the miner group, each mining machine needs 1000 EPK foundation pledge to obtain the qualification to participate in the block output, the more data stored, the higher the calculation power, and the higher the rate of block output, you can obtain reward once you output a block, the block output reward will begin to release after 7 days and will be fully released in the following 7 days. Because only domain experts can upload data, miners can’t scan data themselves, only get data they do not have from other miners in the network, which requires access mortgage to obtain access traffic. A mortgage of 1 EPK can read 10Mb data from the network every day, the more mortgage, the more traffic can be read daily. For any new file, the first 100 miners who complete the storage can get double the amount of hash rate.

In addition to 75% of the EPK allocated to miners, 9% of the EPK mined every day will be given to the domain experts group, each domain experts divvy up the rewards in real-time according to their own contributions to the amount of data. In order to avoid a large gap in the scale of data in different fields, the system will calculate the weight after taking a lot of the data amount contributed by all domain experts. In addition, in order to avoid the evil of domain experts, each domain expert can only be nominated by someone who is already a domain expert, and need to get one hundred thousand EPK votes from the community after nomination. Nominees who have already been nominated enough by voting can be qualified as a domain expert, and can upload data to share income, or can nominate others. The daily earnings of the domain expert can’t be withdrawn until 7 days later. The rogue domain expert can be blacklisted by voting of a group of domain experts, the domain expert who is blacklisted will not be able to withdraw the rewarded, and the domain expert who nominates the domain expert who was penalized will also be penalized, and the number of votes he needs to get the qualification of domain expert will be increased by 25thousand for each blacklisted person he nominated.

In order to encourage people to vote, 1% of EPK mined by the whole network every day will be given to all the people who participate in the voting, the more they vote the more they share the income, as long as they vote, they can share the income and has nothing to do with whether the person voted is a domain expert.

Finally, there is 15% of EPK left in the whole network. This part will introduce an index of access demand, which is the proportion of EPK pledged by the whole network for access to data to the total circulation of EPK. The higher the proportion is, which means that the better the whole network data quality is, the remaining 15% of EPK will be allocated more to miners to enhance bandwidth and optimize access performance; on the contrary, if this proportion is not high, which means that data quality needs to be strengthened, the remaining 15% of EPK will be allocated more to bounty hunters to do more annotation tasks to optimize data quality. It’s like having more visitors in a library, which requires more seats, which means paying the miners to increase bandwidth. When visitors are less, which need more money to buy better quality books to attract visitors, which means giving bounty hunters and domain experts to generate more quality knowledge graph data.


Blockcast.cc: EpiK was listed on three exchanges MEXC, BitMart, Hobbit recently. Is it in the planning? 

At present, EpiK team focuses more focus on the preparation of launch on the mainnet. Friends in the community spontaneously helped us contact the exchange, which is also welcome, Before we launch on the mainnet, miners need a mortgage, improving EPK liquidity can make it miners can more convenient for the miners to get EPK, which are conducive to ecological development. EpiK is doing something that is not limited to the blockchain industry, we are working with DataFunTalk, DataUnion and OpenKG to better achieve EpiK’s goal of building a decentralized knowledge graph collaborative network. So in the selection of investors, in addition to institutions such as FBG and 1475 which have a profound layout in the field of decentralized storage, there is also JackDaw with a license for legal access, as well as network data which has been deep plowing hardware, cluster, IDC computer room for many years, and CLIM Fund with huge data resources, which escort for the project development. We are very grateful to their support.


Blockcast.cc: How can users participate in Eipk’s ecosystem? Can you share the mining tutorials? How can they benefit more?

You can now choose to be a miner and participate in our testnet pre-mining activities (https://github.com/EpiK-Protocol/go-epik/wiki/How-to-join-Testnet-v3.0),
there are 1w EPK rewards every day to help us better test the code;

You can also choose to become a bounty hunter and download the knowledge mainland (https://www.epikg.com/) to answer the question, there are 500 EPK rewards distributed to each hour on average every day;

You can also apply to become one of our domain expert (https://docs.google.com/forms/d/e/1FAIpQLSdvyInslFIZ9vuS-O0kKX_CkCi4mLZZPc2klksdTiBOHq0f2w/viewform?usp=sf_link) and organize human knowledge into a knowledge graph in your field of expertise.


Blockcast.cc: Could you tell me about Epik’s follow-up development plan and future goals?

In the short term, EpiK’s goal is to serve the miners well and find more domain experts, build showcase in 1-2 fields, and perfect the tool construction for decentralizing knowledge graph collaboration closed-loop.


Blockcast.cc: Thank you Thomas for your time and your insights. For the rest of you, stay safe and good night.