Data Sharing
Data Utilization Flow
Requesting Data
Objective: Access validated data tokens for training or inference in AI models (or for 3rd party systems.)
Process:
AI developers or users identify the relevant tokens representing the data they wish to use.
They initiate a request to access the data associated with these tokens.
Verifying Token Ownership
Service Used: Hedera Token Service
Inputs Required:
Token ID(s) of the data assets
User's Hedera account ID
On-chain Recording:
Check that the requesting account owns or has rights to the token associated with the data.
If ownership is verified, proceed to grant access.
Data Retrieval
Objective: Access the actual data files linked to the tokens.
Process:
Once access is granted, the AI model retrieves the data files from the Hedera File Service using the stored file hashes.
Access permissions and licensing terms are checked through smart contracts to ensure compliance.
Utilizing Data in AI Models
Objective: Use the retrieved data for training or inference in AI models.
Process:
Data is processed and fed into the AI model.
The model may produce outputs used for decision-making or further analysis.
Logging Data Usage
Service Used: Hedera Consensus Service
Inputs Required:
Token ID(s) used
Model parameters
Timestamp of usage
On-chain Recording:
Each instance of data usage is logged on the Hedera ledger, capturing which tokens were utilized, model parameters, and time of operation.
This log serves as an audit trail for accountability and transparency.
Performance Monitoring and Feedback
Objective: Gather insights on the effectiveness of the data used.
Process:
Collect performance metrics from the AI model (e.g., accuracy, efficiency).
Provide feedback regarding data relevance and quality.
On-chain Recording:
Record performance metrics and feedback on the Hedera ledger for continuous improvement and data source evaluation.
Last updated