Data Authentication Protocol
Last updated
Last updated
The data authentication and data rights confirmation system within TAGGER is a decentralized framework built on cryptography and smart contract technology, leveraging the immutability of digital asset credentials. This system is accessible to anyone, anywhere in the world. By utilizing advanced deep learning encryption algorithms, we ensure that data ownership and rights are securely protected.
All data processed through our network undergoes encryption using a dynamic mixed chaotic system. Once data collection or dataset upload is complete, the platform generates an index file for the dataset. The dataset owner can then mint an NFT to authenticate ownership. The individual holding the NFT corresponding to the dataset has exclusive rights to manage it, including viewing, publishing annotation tasks, trading, authorizing, or even destroying the dataset.
The dataset owner retains the right to publish data annotation tasks. Upon task completion by all participants using the platform’s annotation tools, the labeled data is encrypted using a hybrid encryption algorithm based on a hyper-chaos system and time-series data prediction algorithms. This prevents any tampering with the annotated content. The network generates an index for the labeled data, and the task publisher can mint an NFT to claim ownership of the newly labeled dataset.
As depicted in Figure 2, our encryption and decryption processes rely on dynamic mixed chaotic encryption methods. This method uses duo-layer chaos mapping to generate pseudo-random sequences for data diffusion and permutation through two creation processes. During encryption and decryption, the key used in this dual-chaos digital image encryption scheme consists of the calculation parameters of the first-level chaos mapping and the initial values of the second-level chaos mapping. The encryption process follows a diffusion-then-permutation sequence, while the decryption process reverses this order to ensure the data's integrity and security.
Figure 3 illustrates the platform's encryption system for user-generated labeled datasets using deep learning algorithms. Initially, the system applies a hyper-chaotic system to generate chaotic sequences. A time-series prediction algorithm simulates chaotic characteristics and constructs new chaotic signals, which are then analyzed for their dynamic properties using a multi-parameter approach. This newly generated chaotic signal is applied to the encryption process, securing the labeled dataset and ensuring its protection against unauthorized access or modification.
