Metaverse Identity & Credit Protocol Based on the Blockchain Data
MetaVisa Team [email protected]
www.metavisa.com October 2021
Abstract
Traditional finance is largely backed by reputation and credit, not just assets. A credit score is a value that represents the creditworthiness of an individual; the higher the score, the better a borrower looks to potential lenders. As individuals, people are assessed on their ability to repay a mortgage based on their credit history, not purely on the fact that people are already outright owners of real estate. Similarly, there are ways to assess the capital strength of corporate and institutional borrowers. These “reputation economies” make up the majority of the traditional financial system.
In the current DeFi landscape, over-collateralization is necessary partly because of the
pseudonymous nature of blockchain transactions. A lender rarely knows a borrower’s identity, which introduces an unacceptable degree of risk, as there’s no way to guarantee repayment.
Even on a pseudonymous basis, DeFi also lacks adequate credit scoring or borrower risk assessment mechanisms. So, making sure someone has sufficient “skin in the game” is the only way to ensure they’ll make good on their repayment obligations. In case of default, over- collateralized lenders can simply liquidate the borrower’s collateral.
The solution to bridging the gap between requiring assets and managing uncollateralized loan risk is simple. Ideally, the credit model is robust enough to support active lending rather than purely serving as a theoretical framework.
MetaVisa Protocol aggregates and analyzes all transactions of blockchain objects on multiple networks. MetaVisa Protocol then establishes links among them and evaluates their credit score in the most objective and transparent manner.
By analyzing blockchain data such as on-chain addresses behavioral, MetaVisa Protocol helps users establish and display reliable on-chain identity & credit records and makes it easier for DeFi, NFT, games, DAO, and other DApps to serve their users better through our credit system.
MetaVisa Protocol Credit Score (MCS)
Based on the blockchain data, using cloud computing, machine learning technologies, and model algorithms such as logistic regression, decision trees, random forests, etc., MetaVisa Protocol conducts comprehensive processing and evaluation of data in various dimensions such as Credit history, On-chain behavior preference, Address activity level, Asset holdings &
Portfolio, Address correlation.
The MCS system will award the users with ranked badges based on their MCSs. Users with higher MCSs will be rewarded with high-ranking badges and therefore have privileges in services from various DApps.
Credit history
On-chain behavior preference Address activity level
Asset holdings & Portfolio Address correlation
The credit score of a Token is assessed based on 4 parameters, i.e., (i) the token price, (ii) the market cap (MC) of the token, (iii) the token trading volume, and (iv) the number of
transactions related to the token. High-credit-score tokens have a great advantage over low- credit-score tokens. Once being used as collaterals, they offer high loan-to-value ratios to borrowers; lenders also get favorable lending interest rates with their high-credit-score tokens.
Users can improve their credit scores once they possess those high-valuable digital assets.
MetaVisa Protocol also performs credit ratings for other digital assets such as NFT. The credit scores for NFTs are estimated based on the number of times that they have been auctioned on MetaVisa Protocol or other trusted platforms. NFTs’ scores should not depend on their price because their price is typically determined in an unreliable manner, based on the emotion and the preference of customers.
To evaluate credit scores for a wallet address, MetaVisa Protocol credit scoring model will take into account the following factors:
Total asset. The total asset represents the financial strength of the wallet address; it is the most intuitive and easiest-to-evaluate parameter. It is also the second most important parameter that affects credit scores. The total asset is the sum of balance and total
investment minus total liabilities. The credit score of the address is proportional to its total asset. As this parameter is a volatility index, the credit score is calculated based on both the average total asset in a certain period and the total current asset.
Transaction history. Transaction history is the most important parameter; it has a major impact on credit scores. MetaVisa Protocol collects and aggregates all exchanges, loans, deposits, and other transactions of the wallet address across multiple DApps to calculate various intermediate parameters such as (i) the age of the address, (ii) its transaction amount, (iii) its frequency of transaction, (iv) its number of liquidations and (v) its total value of liquidations.
Loan ratios. Loan ratios (debt ratios) represent the debt position of the account. It is divided into two sub-parameters, i.e., loan-to-balance ratio and loan-to-investment ratio.
Circulating assets. This parameter represents how active users are in the crypto market.
The more money they invest in the crypto market, the more creditworthiness they gain from others. This parameter is evaluated based on 2 sub-parameters, (i) investment-to- total-asset ratio and (ii) ROE (Return on equity).
Trustworthiness of possessing assets. The last parameter for credit rating is the
trustworthiness of the assets owned by the wallet address. The wallet can improve its credit score by possessing high-value and trustworthy digital assets. At present, MetaVisa Protocol evaluates two types of digital assets only (i.e, Token and NFT).
Blockchain Data mining
Data mining is the digging for or ‘‘mining’’ of prior unknown useful information from data.
Defined as ‘‘the exploration and analysis, by automatic or semiautomatic means of large quantities of data in order to discover meaningful patterns and rules’’ (Berry and Linoff, 2004), data mining applies diverse algorithms for finding patterns in data. Data mining predictive models are non-parametric in nature. Therefore, unlike classical statistical techniques, most of the data mining has minimal prior assumptions for model building. Data mining is useful for the following purposes:
Exploratory data analysis – examining the dataset with graphical pictures and basic descriptive statistics;
Descriptive modeling – partitioning the data into groups;
Predictive modeling – building statistical models to predict the target variable;
Discovering patterns and rules – discover items that occur frequently in databases; and Retrieval by content – finding patterns in a new dataset using the procedures from a prior
analysis (Hand et al., 2001).
Data mining models
There are numerous methods and procedures available for exploring factors associated with a binary outcome, but of course, the model choices depend on the research aims. Because the research objectives were to accurately predict pressure ulcer prevalence and identify clinically relevant factors that are associated with pressure ulcers, we cautiously selected four predictive modeling methods: logistic regression, decision trees, random forests (RF), and multivariate adaptive regression splines (MARS). These four data mining models were selected because of their high predictive performance and meaningful interpretations they supply.
Logistic regression
Logistic regression (see, e.g., Hosmer and Lemeshow, 2000) is the standard statistical
Generalized Linear Model (GLM) approach for modeling binary outcomes, i.e., whether or not
a pressure ulcer is present. In this approach, the logit of the conditional probability of having a pressure ulcer is formulated as a linear function of covariates.
The slope parameters in a logistic model can be interpreted as log of odds ratio. The
advantages of logistic regression include simple linear structure, widely available fitting software, some flexibility to deal with categorical variables and model interaction terms. Its disadvantages mainly stem from linearity as well. The linear functional form may not provide satisfactory fit when strong nonlinearity and complex interactions are present. The idea is to approximate nonlinear curve with broken lines (first-order spline functions) with thresholds. These terms are data driven and found by automated greedy search procedures.
Decision trees
Decision trees fit piecewise constant models by recursively partitioning the predictor spaces.
They are helpful in identifying sub-populations with high/low pressure ulcer incidence rates via easily interpreted grouping rules. A rule is induced by a binary split on covariates with questions such as ‘‘Is age less than 40?’’ or ‘‘Is subject male or female?’’ According to some criterion, the algorithm searches for the best split among all possible splits and the data is partitioned
accordingly. The procedure is repeated till the data set is split into a number of mutually exclusive groups. To address the tree model selection and other issues, Breiman et al. (1984) proposed the Classification and Regression Trees (CART) procedure, which has made tree models widely popular in various application fields. Su et al. (2011) introduced different decision tree methods to nursing research. Advantages of decision trees include efficiency in handling categorical variables, invariance to monotone transformations on predictors, ease of understanding, handling missing data, and ability to deal with complex interactions. One of the drawbacks is that the tree models are highly data-adaptive and unstable, meaning that minor alterations in the sample data may cause dramatic changes in the tree model structure. Also predictions from a single tree analysis are often unsatisfactory.
Random forests
The random forests (RF) are among the techniques that help to address the weakness of a single decision tree, by borrowing strength from the instability of tree models. Trees are instable in the sense that predictions from each single tree tend to have small bias but large variance. As a model ensemble method, random forests reduce variance by averaging the predicted values from a number of tree models. The main idea is model ensemble: build up a large number of tree models by perturbation (e.g., bootstrapping) and then combine the predictive power from all tree models. RF achieves high prediction accuracy and can handle a large number of predictors of different types and missing data. The Gini index is used to calculate the
importance rank of predictors (Brieman, 2001). The drawback of random forests is that it does not supply an explicit functional form (i.e. an equation) for the predictive model and the model interpretation is not so easy compared to a single tree model. To remedy this, random forests implements two ways of extracting interpretation: variable importance ranking helps sort variables in terms of their predictive power and partial dependence plot depicts the functional relationship between each predictor and the response after adjusting for other predictors.
Multivariate adaptive regression splines
Multivariate adaptive regression splines (Friedman, 1991) adaptively fits piecewise linear models with truncated power (often of first order for the sake of feasibility) spline basis
functions. The final multivariate adaptive regression splines (MARS) model can be written as a model form. MARS is similar to logistic regression in retaining a model form, however adds more flexibility in handling categorical variables and nonlinear patterns and interactions and little requirement on data preparation and variable selection. Its drawback is that it does not provide as good a fit as random forest technique.
Usage Scenario
The Symbol of Honor
MetaVisa Protocol issues a badge of Honor by evaluating the credit rating of an address. Each badge will be an exclusive NFT owned by the address. The badge will be an important symbol for the user in the Metaverse. Users can also display the badges through traditional social platforms such as Twitter, Facebook, LinkedIn, and Instagram.
Login Credentials
MetaVisa Protocol credit system is associated with the public key, making the MetaVisa Protocol credit system highly identifiable. Based on the W3C WebAuthn, the MetaVisa Protocol credit system can be used as a credential to log in to third-party applications in Metaverse. The login method will be changed from entering the account password to using the private key associated with the MetaVisa Protocol credit system to sign the login operation, greatly improving security and user experience.
Integrate with Game
Games in Metaverse can be combined with the MetaVisa Protocol credit system to achieve special rewards distribution, game asset credit transactions, etc. In addition, when game players form a guild in a game, they can effectively manage the guild members according to the
different MetaVisa Protocol credit levels.
Integrate with DeFi
The decentralized lending platform can combine with the MetaVisa Protocol credit system to improve product experience, such as adding credit lending services, etc. In addition, the IDO platform can also have different credit levels and carry out more effective and reasonable quota allocation.
Integrate with DAO
Most DAO governance is determined by the number of tokens held to determine the voting weight, but this method can easily lead to the problem of centralized domination by a small number of users. DAO can be combined with the MetaVisa Protocol credit system, such as addresses with higher credit scores will have more voting weights.
User growth
In order to attract users in the early stage, Devs usually use airdrops for marketing, but airdrop hunters with thousands of addresses can easily ruin the development of early projects. So instead, devs can use the MetaVisa Protocol credit system to screen out trusted users, thereby improving marketing efficacy quickly.
Customer Management
The devs or the community can formulate a corresponding membership system or point redemption system based on the MetaVisa Protocol credit system to motivate users, maintain user relationships, and thereby enhance user loyalty.
Privacy and Information Security
MetaVisa Protocol will not invade your privacy and collect personal data other than the address behavioral acts on the chain.
MetaVisa Protocol has formulated a special "Credit Data Security and Privacy Protection Policy," dedicated to protecting user privacy.
To ensure the security of your private information, when we output information externally with your authorization, in principle, we will not output your original detailed information but information that has been processed by desensitization and obfuscation.
Established a complete data security management and control process and system and adopted technical means to ensure that the corresponding systems and processes are fully implemented.
We review the data security capabilities of all cooperative organizations. Only those organizations that meet the corresponding standards will cooperate. At the same time, we
will export information security protection standards to these organizations to enhance the information security protection capabilities of the cooperative ecosystem.
Economic Model and Issuance Plan
VISA is a native token issued by MetaVisa Protocol.
VISA Function
MetaVisa Protocol charges for the operation and storage of identity, credit systems, and smart contracts, thereby realizing economic incentives for nodes and preventing resource abuse.
MetaVisa Protocol also encourages users to contribute personal data and address data so as to complete the MetaVisa Protocol identity system and credit evaluation system.
Used to pay for the service fee for invoking MetaVisa identity system and credit system services, and the tokens are burned when calling. The specific burning method will be detailed in the whitepaper;
Used to realize the governance of MetaVisa Protocol. Governance includes voting for node elections, changes to the MetaVisa Protocol identity and credit evaluation model, etc.;
Used to encourage users to provide personal data and address data actively, and to improve the user‘s personal identity system in Metaverse;
Used to upgrade the MID visual presentation by consuming VISA;
Staking VISA can get rewards, and 30% of the net fees will be distributed as staking rewards;
36% of the net fees on MetaVisa Protocol go to a VISA buy/burn.
VISA Issuance Plan
Total supply: 1,000,000,000
Distribution Allocation Details
Community
Rewards 26.00% 260,000,000
Distributed to the community through transaction mining and pledge, etc. Emitted across 36 months, with halvenings every 6 months.
Team and
Consultants 20.00% 200,000,000
6 months cliff since TGE, 15% will be unlocked in the first two years, and the remaining released quarterly within two years.
Treasury 12.00% 120,000,000
These parts of the token will be retained as a contingency. Unlocked 20% by the third month from TGE, and the remaining released quarterly within two years.
Strategic Investor Phase I /Seed
8.00% 80,000,000
Unlocked 16% from TGE, and the
remaining released linearly within 6 months.
Strategic Investor Phase II /Private
10.00% 100,000,000
Unlocked 16% from TGE, and the
remaining released linearly within 6 months.
Partnership &
Ecosystem Incentive
8.00% 80,000,000 Locked for one year after issuance, and then released linearly within two years.
Marketing 15.00% 150,000,000 Used for marketing activities, including but not limited to airdrops, etc.
IDO 1.00% 10,000,000 100% unlocked.
Roadmap
Q4 2021
Initial blockchain integrations: Ethereum Data cleaning & Build algorithm model Data provider integrations
MetaVisa Protocol V1 testnet
Q1 2022
MetaVisa Protocol V1 mainnet
Ethereum Layer 2 blockchain integrations Data provider integrations
DeFi integrations
NTFs and delivery channels integrations
Q2 2022
Multi-chain Support: Binance Smart Chain, Huobi Eco Chain, Solana, Polkadot Smart Templates: Pre-built Smart Triggers for specific use cases
MetaVisa Protocol V2 testnet
Q3 2022
MetaVisa Protocol V2 mainnet Identity & Credit Oracle Engine