Problem Description

Decentralized finance (DeFi) is undoubtedly one of the fastest-growing industries. UniSwap surpassing Coinbase in spot trading volume by 2023 is proof that the growth and adoption of DeFi is serious. With industry leaders like BlackRock and Fidelity showing interest in DeFi, it is no longer a blockchain experiment anymore.

DeFi inherently offers automated, self-service products and services, where users interact directly with the application, with third-party intermediaries such as banks, corporations, agents, and government. It's an open borderless financial system available anywhere and anytime to anyone. DeFi is important because it makes financial services more transplant, accountable, and accessible to people who have been excluded from the traditional financial system In short, more innovation, and more growth in the DeFi space = good for the industry = good for everyone.

Despite the growing popularity of the DeFi industry, a non-sophisticated user can face difficulties navigating and understanding. It can be confusing and misleading. There are many data and technical moving parts, and some tools aren’t that advanced to provide a coherent view of the industry. There is immense complexity associated with understanding and utilizing the data generated from DEXs, liquidity pools, and financial transactions occurring on public blockchains, DEXs & DEX aggregators. If you aren't deep in the industry, it's very difficult to wrap your head around.

The nature of the complexity, coupled with the lack of structured and user-friendly interfaces, poses a significant barrier to entry for many general users and innovators who aren’t deep in the industry. Moreover, existing DEX management methodologies and smart contracts lack the capability to handle advanced operations, limiting the range of products and services that can be built on them.

Solution Description

The project DeFiGraph is an initiative to build a Knowledge Graph (KG) foundation for the DeFi sector and develop a user-friendly front (dApp) end and APIs for developers to interact with the KG. Non-technical users can use natural language to query complex data instead of having to learn a query language. The project is being executed with the collaboration of key stakeholders, including L3A, Openmesh, Deeplink, MIT researchers, and Neo4j.

The project scope includes data collection, KG design, semantic mapping, graph database implementation, and the creation of an interactive front end and APIs. The entire project will be open source and all the code will be available to the public.

The goal is to build a Knowledge Graph (KG) that maps relationships between various DeFi entities, such as liquidity protocols, DEXs, DEX aggregators, meta aggregators, and public blockchains. This KG will combine with Language Learning Models (LLMs) that can understand user queries and provide context-aware responses. The user interface acts as a bridge, allowing users to query the KG through natural language and receive tailored insights.

Project Benefits for SNET AI platform

Can serve as an inspiration for creating user-friendly interfaces within the DeFi Knowledge Graph and Language Learning Model project. DeFiGraph core product can be available at SingularityNET Marketplace and tailored services such as AGIX token-related queries can be provided

Milestone & Budget

Milestone 1: Data Gathering, Preprocessing & infrastructure setup

Responsible Stakeholders: Openmesh/L3A | Budget allocation: $0

Acquire relevant data from various DeFi protocols and blockchain sources. Clean, preprocess, and standardize the data to ensure consistency and reliability.

Universal Data Collector (UDC)

The UDC is one of the core components of Openmesh. It facilitates the collection, enhancement, and standardization of a variety of data sources, extracting atomic events and packaging them into indexed data sets that can be used for live-streamed and historical data analytics.

After the raw data is collected, sorted, and enhanced, it’s produced to an Apache Kafka cluster, where a network of stream processors pulls the data and processes the raw data into standardized events. For DEXs, the raw data is filtered into different categories – first by DEX, then by event type, e.g. liquidity pool data, trading pairs, trading volume e.t.c. After the events are filtered, their schemas are transformed into a standardized format, which is then produced back to Kafka to their appropriate topics. Blockchain data is processed in a similar manner, with smart contract events being pre-processed to decipher which DeFi protocol they belong to before being standardized and sorted.

Here is an example of real-time data collection:

Data Storage: Store the processed data, knowledge graph, and LLM checkpoints in appropriate databases and file systems for easy retrieval and access.

Deliverables: Cleaned and preprocessed dataset containing historical and live data from DeFi protocols and blockchains.

Here are the details;

Data Sources Identification and Acquisition:

Objective: Identify and acquire relevant data sources essential for a comprehensive understanding of the DeFi sector.

• Unified APIs Usage: An integrated approach to collect data across various platforms and protocols.
• On-chain Data: Capture transactional data, smart contract interactions, and other on-chain events from public blockchains including Ethereum and Binance Smart Chain.
• DeFi Platforms Integration: Incorporate data from APIs provided by DeFi protocols, DEXs, and other DeFi projects. Here is the
  current coverage
• Detailed Information:
  Supported Feeds and Symbols Documentation

Blockchain and Network Metrics:

• Block sizes
• Block timestamps
• Blockchain confirmations
• Blockchain network health metrics
• Mempool data
• Uncle block data
• Hash rates
• Network difficulty
• Ethereum Improvement Proposals (EIPs) data
• Gas prices

Transactional Data:

• Transaction details
• Transaction volumes
• Transaction data (buys, sells, swaps, adds, and removes)
• SORs, swap routes
• Transaction types
• Smart contract interactions
• Asset transfers
• Slippage data (records of price slippage occurring on the platform)

Smart Contracts and Protocols:

• Smart contract addresses
• Smart contract events
• Smart contract creation data
• Governance data (voting records, proposal details, governance token distribution, if any)
• Additional AMM-specific data as relevant
• Yield farming data (current APY, assets involved, amount of liquidity provided)
• Staking data
• Fees data (protocol fees, liquidity provider fees)
• Impermanent loss data (data relating to the impermanent loss experienced by liquidity providers)

Asset and Token Metrics:

• Asset prices and historical price data
• Token data (information about native tokens, including price, distribution, and usage within the platform)
• Asset types
• Asset transfer data
• Asset custody data

Liquidity and Exchange Metrics:

• Liquidity pool data (Total Value Locked (TVL), individual asset contributions, etc.)
• Trading pairs data
• Trading volume
• Wallet addresses of liquidity providers
• Total Value Locked (TVL) in protocols
• Individual asset liquidity
• Liquidity provider data
• Yield rates

API Integration and Real-time Data Ingestion:

Objective: Establish seamless integrations with relevant APIs to access both real-time and historical DeFi data.

• API Endpoints: Utilize APIs from blockchain explorers, DeFi protocols, and data aggregators.
• Data Types: Fetch real-time and historical data including transaction details, token prices, liquidity pools, yield farming strategies, and more.
• Integration Details:
  WebSocket API Documentation

Data Normalization:

Objective: Convert and structure data from varied sources into a consistent and compatible format for further processing and analysis.

• Consistent Format: Ensure that data from different DeFi platforms, blockchain networks, and protocols is normalized into a unified schema.
• Normalization Details:
  Schema Reference Documentation

Infrastructure and Real-time Data Integration:

Objective: Develop and maintain a robust infrastructure tool to ensure seamless data acquisition, storage, and distribution.

xNode Deployment: Utilize the xNode framework, a multipurpose infrastructure tool developed by the L3A Protocol & Openmesh.

Here is an example of real-time data collection:

Infrastructure Overview:

• xNode is designed for clusters of bare metal servers and acts as a central hub for real-time data access.
• Comprehensive support for raw blockchain data and data from Ethereum blockchains and DEXes.

Automation and Extensibility:

• Automated setup processes encompassing server provisioning, Kubernetes configuration, networking setup, and essential services (e.g., Apache Kafka, PostgreSQL).
• Leverage Apache Kafka's capabilities to integrate diverse data feeds.

Customization and Tailoring:

Objective: Ensure that the infrastructure aligns with the specific requirements of the DeFi KG system project.

Customization by Openmesh Team:

• Personalization of the xNode framework to suit project needs.
• Extend the setup of cloud services, middleware components, API configurations, and integration of xNode with existing systems.

Timeline: Week 3 - Week 6

Milestone 2: Knowledge Graph and Model Development (10 weeks)

Responsible Stakeholders: MIT/Neo4j | Budget allocation: $20,000

Design and construct a comprehensive knowledge graph that represents the interconnectedness of various DeFi concepts, protocols, assets, and interactions.

The main KG GTF, is where entities (nodes) are connected through relationships (edges). In the context of DeFi, entities could be protocols, tokens, transactions, or liquidity pools, and the relationships could be transactions, ownerships, dependencies, etc.

Entity Identification

• Protocols: Such as Uniswap, Compound, Hashflow.
• Tokens: e.g., ETH, DAI, USDC.
• Liquidity Pools: Specific pools where liquidity providers add assets.
• Transactions: On-chain actions like swaps, minting, burning (see data table for more details)

Relationship Mapping

• Ownership: Who owns which tokens or participates in which pool, LP and trades relationships, token swaps vs. pools LTV ratios, token transfers between wallets
• Dependency: If one protocol relies on another, e.g., a Dapp depending on a particular stablecoin or its own native token.
• Transaction: Token A swapped for Token B on Protocol X.

Property Definition

• Entities can have properties. For example, a token might have properties like its current price, total supply, or contract address.

KG Database Selection

• Neo4j: A popular choice for graph databases, suitable for defining, querying, and visualizing knowledge graphs.

Semantic Modeling

• Define a consistent vocabulary or ontology for the graph, ensuring interoperability and understanding across platforms. RDF (Resource Description Framework) and OWL (Web Ontology Language) could be used.

Data Integration

• Data Ingestion
• Normalization: Ensure data from various sources is made consistent.
• Cleaning: Removing erroneous or irrelevant data.
• Transformation: Converting data into a format suitable for the KG.

KG Population

• Using tools and connectors specific to the chosen graph database (like Neo4j's Cypher query language) to populate the KG with entities, relationships, and properties.

Query Design:

• Develop a set of common queries that users might want to perform on the KG, such as "Find all DEXs where a specific token is traded" or "Show all transactions involving Token A above a certain value."

Optimization:

• Performance: Ensure fast query times even with massive datasets.
• Scalability: As DeFi grows, the KG should handle an increasing number of entities and relationships without performance degradation.

Deliverables: A well-defined knowledge graph architecture incorporating semantic relationships and essential data points.

Timeline: Week 7 - Week 12

Milestone 3: Language Learning Model Implementation

Responsible Stakeholders: MIT/Neo4j | Budget allocation: $10,000

Develop a specialized Language Learning Model (LLM) tailored to DeFi-related language understanding. Train the model to comprehend and generate contextually relevant DeFi queries and responses.

Deliverables: A functional prototype of the DeFi-specific LLM capable of understanding and generating DeFi-related text.

Timeline: Week 10 - Week 16

Milestone 4: User Interface Development

Responsible Stakeholders: Openmesh/L3A | Budget allocation: $0

Design and build an intuitive web interface that enables users to interact with the Knowledge Graph and LLM seamlessly. Ensure user-friendly navigation and visually appealing design.

We have designed something similar:

In terms of queries, on-chain data works better, specifically raw data like blocks, transactions, e.t.c.

Try:

• - "100 most recent Ethereum blocks"
• - "Most valuable transaction on Ethereum in July, 2023"
• - "Details on the highest-priced ADA/USD trade on Coinbase on the 5th of June, 2023"
   

Deliverables: A user interface prototype with interactive elements, allowing users to query the Knowledge Graph and receive informative responses.

Timeline: Week 13 - Week 20

Milestone 5: API Development

Responsible Stakeholders: Openmesh/L3A | Budget allocation: $0

Develop Application Programming Interfaces (APIs) that offer programmatic access to the functionalities of the Knowledge Graph and LLM. Ensure the APIs are well-documented and easily integrated into third-party applications.

Deliverables: A set of well-documented APIs providing access to Knowledge Graph and LLM capabilities for developers and applications.

Timeline: Week 17 - Week 24

Milestone 6: Testing and Deployment (6 weeks)

Responsible Stakeholders: Openmesh/L3A, MIT & Neo4j | Budget allocation: $5,000

Rigorously test the developed systems for accuracy, functionality, and performance. Identify and rectify any bugs or inconsistencies.

Deliverables: Thoroughly tested and debugged Knowledge Graph, LLM, web interface, and APIs.

Timeline: Week 21 - Week 26

Milestone 7: Deployment and Documentation

Responsible Stakeholders: Openmesh/L3A, MIT & Neo4j | Budget allocation: $3,000

Deploy the finalized systems to production servers and prepare comprehensive user documentation and guides for seamless adoption.

Deliverables: Deployed Knowledge Graph, functional LLM, user-friendly web interface, and well-documented APIs.

Timeline: Week 27 - Week 32

Milestone 8: Outreach and Community Engagement, Continuous Improvement and Iteration.

Responsible Stakeholders: Openmesh/L3A, MIT & Neo4j | Budget allocation: $2,000

Engage with the DeFi community, raise awareness about the platform's capabilities, gather valuable user feedback, and foster collaboration.

Deliverables: Increased user engagement, community feedback, and strengthened collaboration within the DeFi ecosystem.

Continuously enhance the systems based on user feedback, refine features, and optimize performance for sustained growth and relevance.

Deliverables: Improved systems with enhanced features and functionalities, addressing user needs and industry advancements.

Timeline: Ongoing after milestone 7

Related Links

Long Description

The problem

Decentralized finance (DeFi) is undoubtedly one of the fastest-growing industries. UniSwap surpassing Coinbase in spot trading volume by 2023 is proof that the growth and adoption of DeFi is serious. With industry leaders like BlackRock and Fidelity showing interest in DeFi, it is no longer a blockchain experiment anymore.

DeFi inherently offers automated, self-service products and services, where users interact directly with the application, with third-party intermediaries such as banks, corporations, agents, and government. It's an open borderless financial system available anywhere and anytime to anyone. DeFi is important because it makes financial services more transplant, accountable, and accessible to people who have been excluded from the traditional financial system In short, more innovation, and more growth in the DeFi space = good for the industry = good for everyone.

Despite the growing popularity of the DeFi industry, a non-sophisticated user can face difficulties navigating and understanding. It can be confusing and misleading. There are many data and technical moving parts, and some tools aren’t that advanced to provide a coherent view of the industry. There is immense complexity associated with understanding and utilizing the data generated from DEXs, liquidity pools, and financial transactions occurring on public blockchains, DEXs & DEX aggregators. If you aren't deep in the industry, it's very difficult to wrap your head around.

The nature of the complexity, coupled with the lack of structured and user-friendly interfaces, poses a significant barrier to entry for many general users and innovators who aren’t deep in the industry. Moreover, existing DEX management methodologies and smart contracts lack the capability to handle advanced operations, limiting the range of products and services that can be built on them.

The project DeFiGraph is an initiative to build a Knowledge Graph (KG) foundation for the DeFi sector and develop a user-friendly front (dApp) end and APIs for developers to interact with the KG. Non-technical users can use natural language to query complex data instead of having to learn a query language. The project is being executed with the collaboration of key stakeholders, including L3A, Openmesh, Deeplink, MIT researchers, and Neo4j.

The project scope includes data collection, KG design, semantic mapping, graph database implementation, and the creation of an interactive front end and APIs. The entire project will be open source and all the code will be available to the public.

The goal is to build a Knowledge Graph (KG) that maps relationships between various DeFi entities, such as liquidity protocols, DEXs, DEX aggregators, meta aggregators, and public blockchains. This KG will combine with Language Learning Models (LLMs) that can understand user queries and provide context-aware responses. The user interface acts as a bridge, allowing users to query the KG through natural language and receive tailored insights.

Innovation and Industry Impact:

The core innovation lies in combining the power of a Knowledge Graph with an intuitive user interface, like ChatGPT. This innovation can help the sector by enabling non-technical users to seamlessly interact with the vast amount of DeFi data, protocols, and applications. The project's potential to provide instant insights, data-driven decisions, and simplified experiences can be extremely innovative.

The Bigger Picture:

In the evolving landscape of decentralized finance, user-friendly and data-enriched interfaces are essential to drive mainstream adoption especially when the industry itself moves very fast. By providing a comprehensive understanding of DeFi applications, and its utilities, our project aligns with the broader movement of democratizing financial services and granting users control over their assets and financial decisions. Further, establishing an open framework for Knowledge Graph for DeFi is critical for innovation, and adoption.

This project would resonate with the core mission of democratizing and decentralizing AI. It empowers the DeFi industry and shares the common goal of supporting open, democratic, and inclusive technologies that benefit everyone.

The Core Product

Our solutions will be available via web-based dApp and APIs (web2) for the v1 and web3 interfaces and can be introduced as the project scales in the future.

Main Products and Capabilities:

DeFi Knowledge-Graph-as-a-service (KGaaS):

• Enables users to access structured and interconnected DeFi data.
• Provides instant answers, insights, and recommendations in natural language.
• Supports queries about protocols, tokens, liquidity pools, yield farming, and more.
• Enhances user experience by bridging the gap between raw data and comprehension.

You can ask questions like;

• Which protocols had the highest volume last week?
• List popular DeFi projects on Ethereum.
• Show me the price history of [specific token].
• I have 10 ETH, What's the best DEX to trade with the least amount of gas fee/price slippage
• How do I earn rewards in a pool?
• What is the highest yield pool on Ethereum & Solana?

The APIs will allow developers to run the queries, instead of using the web-based UI.

We have developed something similar and our past experience & mistakes will help us to build the DeFiGraph project better

You can try the main application.

In terms of queries, on-chain data works better, specifically raw data like blocks, transactions

Try:

• - "100 most recent Ethereum blocks"
• - "Most valuable transaction on Ethereum in July, 2023"
• - "Details on the highest-priced ADA/USD trade on Coinbase on the 5th of June, 2023"
   

SingularityNET Marketplace Integration:

Can serve as an inspiration for creating user-friendly interfaces within the DeFi Knowledge Graph and Language Learning Model project. DeFiGraph core product can be available at SingularityNET Marketplace and tailored services such as AGIX token-related queries can be provided

Use cases, Innovation, and Enabling

An array of intriguing applications can be developed leveraging the foundational R&D of this project.

General

• Personalized DeFi Bots: The system could be used to create personalized DeFi bots. 

• DeFi Market Insights: Get real-time insights into DeFi market trends, token prices, and liquidity pool data to make informed investment decisions.

• Automated Yield Farming: Automate yield farming strategies based on dynamic market conditions and user preferences.

• Portfolio Rebalancing: Automatically rebalance DeFi portfolios based on market conditions and risk appetite.

• Token Swap Analysis: Evaluate various DEXs for token swaps based on slippage, gas fees, and liquidity.

• Smart Order Routing & DEX comparison: Look for DEXs with the best liquidity and lowest slippage using advanced order routing.

• Advanced Analytics: Generate advanced analytics and reports on DeFi protocols, yield farming strategies and market behavior.

• Automated Trading Strategies: Develop sophisticated trading algorithms that execute orders based on LLM insights, market trends, and risk assessments.

For DeFi Protocols & LPs:

• Automated Market Makers (AMM) Innovation: Leveraging Knowledge of Graph-enhanced LLMs could lead to the design of more advanced and efficient AMM models. For instance, these models could dynamically adjust fees or pool weights based on real-time market conditions, enhancing liquidity provision and reducing slippage.

• On-Chain Yield Products: The system could be used to develop innovative on-chain yield products. By analyzing a vast range of DeFi data, these products could dynamically optimize yield strategies, switching between different lending platforms, liquidity pools, and yield farming opportunities to maximize returns.

• Real-time On-Chain Credit Scoring System: The Knowledge Graph-enhanced LLMs could be used to create a real-time, on-chain credit scoring system. Unlike traditional credit scoring, this system could incorporate a wider variety of factors, such as a user's on-chain transaction history, DeFi portfolio performance, and even social signals from DAO participation.

• Liquidity Provision Strategies: Optimize liquidity provisioning strategies by analyzing pool data, historical trends, and user preferences.

• Liquidity Pool Analysis: Analyze historical liquidity pool data to identify trends and make informed decisions about liquidity provisioning.

• Token Pairing Selection: Choose the best token pairs for liquidity provision based on historical trading volumes and market trends.

• ​​DeFi Regulatory Compliance: The system could be used to build tools to ensure regulatory compliance within the DeFi space. These tools could help DeFi platforms stay updated with changing regulations, monitor transactions for suspicious activity, and maintain detailed records for audit purposes, thereby instilling trust and confidence among users and regulators alike.

Developers and DApp Builders:

• Data Enrichment: Access enriched data from the knowledge graph and LLM via API to enhance applications and DApps.

• Interactive Learning: Offer API-powered interactive learning tools that allow users to query and explore DeFi concepts.

Project Progress and Experience

Our project has moved beyond conceptualization and has firmly transitioned into an action-oriented phase. We've made substantial progress in shaping the direction of the project; defining key deliverables, establishing the high-level technical roadmap, identifying the crucial stakeholders and setting up the data and infrastructure layer for this project. A comprehensive groundwork has been laid down, and resources have been effectively allocated. 

• L3A, Deeplink & Openmesh have been in the web3 & cloud industry for over 7 years now and we've accumulated a substantial amount of experience in dealing with complex projects, data, public blockchains, and decentralized finance. Our team has been directly involved in many innovative projects such as Fantom, Elrond, Ocean Protocol, Yearn, Cosmos, IOTX, and REN for the past years. Launched in late 2020, Openmesh managed to cover 80% of crypto web3 data in a short span of 2 years while building its core technology on bare metal data centers around the world. Openmesh collects over 300+GB worth of complex web3 data every day, making it the largest open web3 data lake in the world.

• Eta X

  - an open-source initiative to build an agnostic price discovery engine, smart order router (SOR) framework for decentralized exchanges (DEXs)

• Pythia

  -  Open analytics & query engine/layer for web3. This dApp allows you to ask questions like “What is the last Ethereum transaction” or “Display 100 most recent transactions on Ethereum”

• UnfiedAPIs

  - A single API for all web3 and crypto data 

• xNode allows you to build your own decentralized data cloud with data connectivity and data API query & analytics engines in minutes instead of weeks/months, and pay only for computing and storage, instead of licensing and setup fees.

  Data Collection and Infrastructure Setup (Customization of xNode):

  Here is an example of real-time data collection:

  L3 Atom Infrastructure Dashboard
   

  A critical phase of our progress is the seamless integration of diverse data sources into the xNode framework. Data categorization includes On-chain DeFi data, DEX native data, and Ethereum consensus layer data. Rigorous data cleaning and preprocessing techniques are applied to ensure that the data's integrity and reliability are upheld.

   

Data Collection:

• Data Sources Identification

  : Identify relevant data sources for DeFi, including blockchain networks (e.g., Ethereum, Binance Smart Chain), DeFi protocols (e.g., Uniswap, Compound), and other relevant platforms.

• API Integration

  : Utilize APIs provided by blockchain explorers, DeFi protocols, and data aggregators to fetch real-time and historical data, such as transaction details, token prices, liquidity pools, yield farming strategies, and more.

• Data Normalizatio

  n: Normalize data from different sources into a consistent format, ensuring uniformity and compatibility for further processing.

• xNode is a multipurpose infrastructure tool, a one-click data infrastructure developed by the L3A Protocol & Openmesh. Designed to run on a cluster of bare metal servers, the xNode acts as a central hub, providing access to a vast array of data sources in real-time. It achieves this by integrating L3A's core open-source data collection technology, which has comprehensive support for raw blockchain data, and data from the most popular Centralized Exchanges (CEXes) and Decentralized Exchanges (DEXes).

• One of the key features of the xNode is its automated setup process. This setup process includes server provisioning, configuration of Kubernetes, networking setup, and essential services like Apache Kafka and PostgreSQL. The xNode's inherent extensibility is another notable feature. Leveraging the capabilities of Apache Kafka, users of the SuperNode can tap into diverse data feeds and integrate them seamlessly with their existing applications.

• Our technical team has personalized the xNode framework, tailoring it to the project's specific requirements. This customization extends to the setup of cloud services, middleware components, API configurations, and the seamless integration of the xNode with existing systems. While this is an ongoing process, we've allocated the initial essentials to support the project's initial stages. 

  User Interface and API Considerations: The project aims to create a user-friendly front end similar to ChatGPT for users to interact with the Knowledge Graph. APIs for DeFi Knowledge-Graph-as-a-service are being considered as a potential addition to the project.

   

  

   

• Success Metrics and Future Considerations: Initial success metrics have been defined, although we recognize the need to assess efficiency, cost, and scalability in-depth. We're working towards refining these metrics to ensure accurate evaluation.

Stakeholder Involvement:

The main stakeholders driving this initiative have been identified:

• L3A/Openmesh: Responsible for providing essential data and managing data infrastructure. Data collection and integration efforts are under their purview.
• MIT: Comprising experts in KG design, semantic mapping, integrations, and SPARQL. They bring in-depth technical expertise to the project.
• Neo4j Sydney Team: Consulting from experts experienced in building commercial-grade graph databases and KGs. Their guidance strengthens our efforts.

Funding Amount

We estimate USD $40,000 would cover part expenses for the project. We are genuinely excited about the potential of this initiative to revolutionize the DeFi landscape. As strong believers in its transformative power, we are committed to its success. The budget & expenses will be shared among L3 & Openmesh. 100% of the funding will be allocated to MIT and Neo4j consultants, not to L3A/Deeplink/Openmesh team or infrastructure. Any additional costs will be supported by L3A & Openmesh.

Any additional funds will be allocated through our

. We are waiting for the final audits so we can allocate extra funds for this project. OpenR&D is powered by smart contracts that hold funds in a public, transparent escrow system. Anyone can verify the funds.

KPIs to closely monitor for the project:

• Milestone Completion: Timely achievement of project milestones and deliverables, ensuring progress according to the defined timeline.
• Data Quality and Diversity: Ensuring accurate and diverse data collection, leading to a comprehensive and reliable knowledge graph.
• User Engagement and Experience: Measuring user interaction, feedback, and query effectiveness to ensure a seamless and valuable user experience.
• Technical Accuracy and Validations: Continuously monitoring and reducing AI bias, ensuring data integrity, and maintaining consistency with real-world data.
• Innovation and Industry Impact: Assessing market adoption, product differentiation, and industry recognition to gauge the project's innovation and influence within the DeFi community.

Our Project Milestones

Milestone 1: Data Gathering, Preprocessing & infrastructure setup

Responsible Stakeholders: Openmesh/L3A | Budget allocation: $0

Acquire relevant data from various DeFi protocols and blockchain sources. Clean, preprocess, and standardize the data to ensure consistency and reliability.

Universal Data Collector (UDC)

The UDC is one of the core components of Openmesh. It facilitates the collection, enhancement, and standardization of a variety of data sources, extracting atomic events and packaging them into indexed data sets that can be used for live-streamed and historical data analytics.

After the raw data is collected, sorted, and enhanced, it’s produced to an Apache Kafka cluster, where a network of stream processors pulls the data and processes the raw data into standardized events. For DEXs, the raw data is filtered into different categories – first by DEX, then by event type, e.g. liquidity pool data, trading pairs, trading volume e.t.c. After the events are filtered, their schemas are transformed into a standardized format, which is then produced back to Kafka to their appropriate topics. Blockchain data is processed in a similar manner, with smart contract events being pre-processed to decipher which DeFi protocol they belong to before being standardized and sorted.

Here is an example of real-time data collection:

Data Storage: Store the processed data, knowledge graph, and LLM checkpoints in appropriate databases and file systems for easy retrieval and access.

Deliverables: Cleaned and preprocessed dataset containing historical and live data from DeFi protocols and blockchains.

Here are the details;

Data Sources Identification and Acquisition:

Objective: Identify and acquire relevant data sources essential for a comprehensive understanding of the DeFi sector.

• Unified APIs Usage: An integrated approach to collect data across various platforms and protocols.
• On-chain Data: Capture transactional data, smart contract interactions, and other on-chain events from public blockchains including Ethereum and Binance Smart Chain.
• DeFi Platforms Integration: Incorporate data from APIs provided by DeFi protocols, DEXs, and other DeFi projects. Here is the
  current coverage
• Detailed Information:
  Supported Feeds and Symbols Documentation

Blockchain and Network Metrics:

• Block sizes
• Block timestamps
• Blockchain confirmations
• Blockchain network health metrics
• Mempool data
• Uncle block data
• Hash rates
• Network difficulty
• Ethereum Improvement Proposals (EIPs) data
• Gas prices

Transactional Data:

• Transaction details
• Transaction volumes
• Transaction data (buys, sells, swaps, adds, and removes)
• SORs, swap routes
• Transaction types
• Smart contract interactions
• Asset transfers
• Slippage data (records of price slippage occurring on the platform)

Smart Contracts and Protocols:

• Smart contract addresses
• Smart contract events
• Smart contract creation data
• Governance data (voting records, proposal details, governance token distribution, if any)
• Additional AMM-specific data as relevant
• Yield farming data (current APY, assets involved, amount of liquidity provided)
• Staking data
• Fees data (protocol fees, liquidity provider fees)
• Impermanent loss data (data relating to the impermanent loss experienced by liquidity providers)

Asset and Token Metrics:

• Asset prices and historical price data
• Token data (information about native tokens, including price, distribution, and usage within the platform)
• Asset types
• Asset transfer data
• Asset custody data

Liquidity and Exchange Metrics:

• Liquidity pool data (Total Value Locked (TVL), individual asset contributions, etc.)
• Trading pairs data
• Trading volume
• Wallet addresses of liquidity providers
• Total Value Locked (TVL) in protocols
• Individual asset liquidity
• Liquidity provider data
• Yield rates

API Integration and Real-time Data Ingestion:

Objective: Establish seamless integrations with relevant APIs to access both real-time and historical DeFi data.

• API Endpoints: Utilize APIs from blockchain explorers, DeFi protocols, and data aggregators.
• Data Types: Fetch real-time and historical data including transaction details, token prices, liquidity pools, yield farming strategies, and more.
• Integration Details:
  WebSocket API Documentation

Data Normalization:

Objective: Convert and structure data from varied sources into a consistent and compatible format for further processing and analysis.

• Consistent Format: Ensure that data from different DeFi platforms, blockchain networks, and protocols is normalized into a unified schema.
• Normalization Details:
  Schema Reference Documentation

Infrastructure and Real-time Data Integration:

Objective: Develop and maintain a robust infrastructure tool to ensure seamless data acquisition, storage, and distribution.

xNode Deployment: Utilize the xNode framework, a multipurpose infrastructure tool developed by the L3A Protocol & Openmesh.

Here is an example of real-time data collection:

Infrastructure Overview:

• xNode is designed for clusters of bare metal servers and acts as a central hub for real-time data access.
• Comprehensive support for raw blockchain data and data from Ethereum blockchains and DEXes.

Automation and Extensibility:

• Automated setup processes encompassing server provisioning, Kubernetes configuration, networking setup, and essential services (e.g., Apache Kafka, PostgreSQL).
• Leverage Apache Kafka's capabilities to integrate diverse data feeds.

Customization and Tailoring:

Objective: Ensure that the infrastructure aligns with the specific requirements of the DeFi KG system project.

Customization by Openmesh Team:

• Personalization of the xNode framework to suit project needs.
• Extend the setup of cloud services, middleware components, API configurations, and integration of xNode with existing systems.

Timeline: Week 3 - Week 6

Milestone 2: Knowledge Graph and Model Development (10 weeks)

Responsible Stakeholders: MIT/Neo4j | Budget allocation: $20,000

Design and construct a comprehensive knowledge graph that represents the interconnectedness of various DeFi concepts, protocols, assets, and interactions.

The main KG GTF, is where entities (nodes) are connected through relationships (edges). In the context of DeFi, entities could be protocols, tokens, transactions, or liquidity pools, and the relationships could be transactions, ownerships, dependencies, etc.

Entity Identification

• Protocols: Such as Uniswap, Compound, Hashflow.
• Tokens: e.g., ETH, DAI, USDC.
• Liquidity Pools: Specific pools where liquidity providers add assets.
• Transactions: On-chain actions like swaps, minting, burning (see data table for more details)

Relationship Mapping

• Ownership: Who owns which tokens or participates in which pool, LP and trades relationships, token swaps vs. pools LTV ratios, token transfers between wallets
• Dependency: If one protocol relies on another, e.g., a Dapp depending on a particular stablecoin or its own native token.
• Transaction: Token A swapped for Token B on Protocol X.

Property Definition

• Entities can have properties. For example, a token might have properties like its current price, total supply, or contract address.

KG Database Selection

• Neo4j: A popular choice for graph databases, suitable for defining, querying, and visualizing knowledge graphs.

Semantic Modeling

• Define a consistent vocabulary or ontology for the graph, ensuring interoperability and understanding across platforms. RDF (Resource Description Framework) and OWL (Web Ontology Language) could be used.

Data Integration

• Data Ingestion
• Normalization: Ensure data from various sources is made consistent.
• Cleaning: Removing erroneous or irrelevant data.
• Transformation: Converting data into a format suitable for the KG.

KG Population

• Using tools and connectors specific to the chosen graph database (like Neo4j's Cypher query language) to populate the KG with entities, relationships, and properties.

Query Design:

• Develop a set of common queries that users might want to perform on the KG, such as "Find all DEXs where a specific token is traded" or "Show all transactions involving Token A above a certain value."

Optimization:

• Performance: Ensure fast query times even with massive datasets.
• Scalability: As DeFi grows, the KG should handle an increasing number of entities and relationships without performance degradation.

Deliverables: A well-defined knowledge graph architecture incorporating semantic relationships and essential data points.

Timeline: Week 7 - Week 12

Milestone 3: Language Learning Model Implementation

Responsible Stakeholders: MIT/Neo4j | Budget allocation: $10,000

Develop a specialized Language Learning Model (LLM) tailored to DeFi-related language understanding. Train the model to comprehend and generate contextually relevant DeFi queries and responses.

Deliverables: A functional prototype of the DeFi-specific LLM capable of understanding and generating DeFi-related text.

Timeline: Week 10 - Week 16

Milestone 4: User Interface Development

Responsible Stakeholders: Openmesh/L3A | Budget allocation: $0

Design and build an intuitive web interface that enables users to interact with the Knowledge Graph and LLM seamlessly. Ensure user-friendly navigation and visually appealing design.

We have designed something similar:

In terms of queries, on-chain data works better, specifically raw data like blocks, transactions, e.t.c.

Try:

• - "100 most recent Ethereum blocks"
• - "Most valuable transaction on Ethereum in July, 2023"
• - "Details on the highest-priced ADA/USD trade on Coinbase on the 5th of June, 2023"
   

Deliverables: A user interface prototype with interactive elements, allowing users to query the Knowledge Graph and receive informative responses.

Timeline: Week 13 - Week 20

Milestone 5: API Development

Responsible Stakeholders: Openmesh/L3A | Budget allocation: $0

Develop Application Programming Interfaces (APIs) that offer programmatic access to the functionalities of the Knowledge Graph and LLM. Ensure the APIs are well-documented and easily integrated into third-party applications.

Deliverables: A set of well-documented APIs providing access to Knowledge Graph and LLM capabilities for developers and applications.

Timeline: Week 17 - Week 24

Milestone 6: Testing and Deployment (6 weeks)

Responsible Stakeholders: Openmesh/L3A, MIT & Neo4j | Budget allocation: $5,000

Rigorously test the developed systems for accuracy, functionality, and performance. Identify and rectify any bugs or inconsistencies.

Deliverables: Thoroughly tested and debugged Knowledge Graph, LLM, web interface, and APIs.

Timeline: Week 21 - Week 26

Milestone 7: Deployment and Documentation

Responsible Stakeholders: Openmesh/L3A, MIT & Neo4j | Budget allocation: $3,000

Deploy the finalized systems to production servers and prepare comprehensive user documentation and guides for seamless adoption.

Deliverables: Deployed Knowledge Graph, functional LLM, user-friendly web interface, and well-documented APIs.

Timeline: Week 27 - Week 32

Milestone 8: Outreach and Community Engagement, Continuous Improvement and Iteration.

Responsible Stakeholders: Openmesh/L3A, MIT & Neo4j | Budget allocation: $2,000

Engage with the DeFi community, raise awareness about the platform's capabilities, gather valuable user feedback, and foster collaboration.

Deliverables: Increased user engagement, community feedback, and strengthened collaboration within the DeFi ecosystem.

Continuously enhance the systems based on user feedback, refine features, and optimize performance for sustained growth and relevance.

Deliverables: Improved systems with enhanced features and functionalities, addressing user needs and industry advancements.

Timeline: Ongoing after milestone 7

Risk and Mitigation

1. AI Bias: Solution: Rigorous data preprocessing and diverse datasets to minimize bias. Ongoing monitoring and fine-tuning for unbiased outputs.

Confidence Level: 5 (Subject to change as the project matures)

2. Data Quality: Solution: Stringent data validation and collaboration with domain experts for accuracy and relevance.

Confidence Level: 6 (Subject to change as the project matures)

3. Milestone Dependencies: Solution: Adaptive approach to prioritize tasks, allocate resources effectively, and maintain project timeline.

Confidence Level: 7 (Subject to change as the project matures)

4. Data Refresh and Updates: Solution: Automated data pipelines for regular updates, ensuring AI models rely on the latest information.

Confidence Level: 4 (Subject to change as the project matures)

5. Evaluation Metrics: Solution: Custom evaluation metrics tailored to our project's uniqueness. User surveys, real-world testing, and continuous feedback loops.

Confidence Level: 3 (Subject to change as the project matures).

Project initiatives of this magnitude require awareness, and active community participation, as they hold the potential to transform entire industries. Recognizing this, Openmesh is taking proactive steps by orchestrating a significant tech summit—a 3-day event gathering over 50 experts in fields such as data, cloud computing, infrastructure, smart contracts, blockchain, fintech, gaming, AI, and security.

This summit will be a valuable opportunity to gain insights from top minds, forge connections, and engage in panels, AMA office hours, rapid recruitment sessions, and project explorations. Through this event, we aim to rally support and excitement for our undertaking, fostering an environment conducive to industry-wide advancement. The summit will serve as a platform to announce and garner backing for this pivotal project, ensuring that it receives the necessary momentum and enthusiasm to thrive.

The team

This initiative is not driven by commercial motives; it's fueled by our collective commitment to drive change and enable a more inclusive DeFi ecosystem. Our team, L3A, is deeply passionate about the transformative potential of this project. Our goal is to make DeFi accessible to a wider audience, contributing to a more democratized financial landscape.

, also known as

, is at the forefront of building a decentralized data infrastructure that empowers individuals with access to global data without a middleman. Our team has been directly involved in many innovative projects such as Fantom, Elrond, Ocean Protocol, Yearn, Cosmos, IOTX, and REN for the past years. With a strong track record in handling Web3 data, Openmesh has laid a solid foundation to take on this project. Our commitment to data sovereignty, open-source principles, and decentralization align perfectly with the project's goals. Openmesh's innovative Unified API and xNode technologies provide a unique advantage, streamlining data access and storage. Our passion for making data open and accessible, combined with their technical expertise, makes Openmesh the ideal collaborator for this initiative.

MIT: Comprising a dynamic team of 3 members, including a Ph.D. and 2 skilled developers, MIT plays a pivotal role in shaping the project's technical aspects. MIT's renowned expertise in knowledge graph design, semantic mapping, and graph databases brings unparalleled depth to the project. Their expertise spans KG design, semantic mapping, integrations, graph databases, and SPARQL. Their contributions are vital in bringing the technical vision of the initiative to life.

Sydney Team: We have successfully engaged the Neo4j Sydney team as consultants for our initiative. Their experience in building commercial-grade graph databases and KGs is a valuable asset. While their guidance is instrumental in guiding our project, any additional costs associated with their involvement beyond the initial budget are covered by L3A. This underscores our commitment to the initiative's outcomes and its transformative potential.

Together, Openmesh, MIT, and the Neo4j Sydney team form a powerful alliance that brings innovation, technical proficiency, and tangible solutions to this initiative. Their collective strengths make them well-suited to drive the integration of Knowledge Graphs and Language Learning Models with DeFi data, shaping the future of decentralized data accessibility and insights in the financial sector.

Related Links