Data Trading and Management
Last updated
Last updated
The capabilities of AI models are fundamentally reliant on both the quantity and quality of data. However, significant disparities in data rights regulations across countries and regions have led to a fragmented and disordered landscape for data acquisition within the AI industry. This lack of standardization has made it difficult to authenticate and protect the value of datasets, thereby infringing upon the rights of data creators and diminishing their motivation to contribute valuable data. As a result, the circulation of datasets has been severely restricted, leading to widespread data silos and hindering the overall progress of AI development.
The TAGGER Marketplace addresses this challenge by establishing a decentralized data market built upon robust data authorization and rights protocols. It facilitates data trading, authorized data usage, and agent-assisted data management. The platform offers access to professional, industry-specific datasets for model developers in fields such as biomedical research, autonomous driving, finance, industrial inspection, the Internet of Things, and education. By breaking down the barriers of data isolation, the TAGGER Marketplace aims to accelerate the transformation of AI technology productivity across a wide array of industries, fostering greater innovation and collaboration in the global AI ecosystem.
Dataset Trading
The decentralized data rights protocol developed by TAGGER enables the secure authentication and protection of datasets. Within this framework, data and data annotation files are encrypted and stored across decentralized networks, with dataset and annotation indices recorded in the form of NFTs. This ensures that only the NFT owner has full access to the dataset. All transfers and interactions involving on-chain data rights credentials are permanently recorded, guaranteeing the secure implementation of data rights. With this protocol, dataset owners can experience fast, cross-border transactions and rapid value realization, incentivizing global users to participate in this new era of data value creation.
The TAGGER Marketplace redefines the way data is transacted, offering comprehensive functions for dataset display, retrieval, and trading. Accessible to all, the platform enables anyone to realize the value of their datasets through seamless data rights delivery, thereby monetizing their data assets. Our goal is to accelerate the global development of the AI industry by overcoming the challenges posed by data silos.
Dataset Authorization for Use
TAGGER has developed a groundbreaking technology for dataset authorization, understanding the importance of preserving data privacy for industry-specific or personal datasets in AI applications. Direct data transactions often pose privacy risks to data owners, but through the data rights protocol built into the TAGGER Marketplace, we have implemented a federated learning architecture using deep learning encryption protocols and Trusted Execution Environment (TEE) hardware encryption. This enables a data authorization usage mode where data can be utilized while remaining invisible.
In this mode, data owners can authorize third-party users to train AI models with their datasets while maintaining full ownership. For third-party users, the data remains encrypted and invisible, safeguarded by our encryption protocols and hardware encryption technologies. Users will not have access to specific information about the dataset; they will only use it for model training and updating model weights. This ensures that data privacy and security risks are mitigated, while still enabling AI development.
By leveraging our platform, data owners can confidently authorize the use of their data without compromising privacy, while the TAGGER Marketplace works to break down data barriers and facilitate the transformation of AI productivity across a wide range of verticals.
Agent-Assisted Data Management System
As vast amounts of data continue to circulate globally, efficient data retrieval and management are crucial to the success of data trading marketplaces. The TAGGER Marketplace incorporates a lightweight Agent-Assisted Data Management engine based on RAG+LLM (Retrieval-Augmented Generation + Language Model) technology. This system provides real-time assistance in data processing, large-scale analysis, and event-based computing. The agents offer robust computing and analytical capabilities, from handling large datasets to responding instantly to user queries.
Parallel Processing: The agent engine takes full advantage of parallel computing by breaking down complex query tasks into sub-tasks, which are executed concurrently within a cluster. This dramatically enhances query processing efficiency, reducing response times and increasing throughput for multiple simultaneous queries.
Data Indexing and Optimization: The agent’s query engine utilizes advanced indexing structures and optimization algorithms to quickly locate required data, minimize the query scan range, and improve query efficiency. By combining precise indexing design with optimization, the platform accelerates complex queries while maintaining high accuracy.
Caching Mechanism: An intelligent caching mechanism stores inference results from frequently asked queries. When the system receives repeat queries, it retrieves the answers directly from the cache, bypassing the need for recomputation. This reduces redundant processing and speeds up response times, delivering fast and efficient query results.
