Each member of the uKit AI ecosystem, if they wish, may contribute to the uData common database by providing datasets that combine the following information:
At the end of each reporting period (1 day or 1 hour), the uData system generates a record in a centralized public transaction log that contains a summary about the data added to the system by each member and the estimated number of tokens that will be transferred to each participant as a reward at the end of the HOLD period.
The estimated reward is calculated based on the percentage of data contributed by a participant, relative to the total amount of data in the system. The actual reward is calculated at the end of the HOLD period.
The record in the public transaction log is signed with a key that is added to the blockchain of the uData smart contract in the Ethereum system, which ensures that there’s no way to falsify the information in the public transaction log, despite the information storage being centralized.
At the same time, the centralized storage of the transaction log makes it possible to provide any level of detail in the report, since there’s limitation for the total amount of data stored in the Ethereum blockchain.
To prevent the misuse of the system, both the community and moderators have the right to block datasets they find suspicious during the moderation period (from 1 week to 1 month).
The list of rules and mechanisms through which certain datasets can be blocked includes but is not limited to the following:
When blocking a dataset which is on HOLD, lists of datasets that are to be blocked, the IDs of those who initiated blocking and description of the reasons are added to the public transaction log.
Blocked datasets, in turn, are excluded from the machine learning process. The goal is to avoid an effect similar to a "51% attack", where a small group of participants negatively influence the behavior of the system, in this case, providing biased datasets and affect the machine learning process.
At the end of the HOLD period, UKT tokens are recalculated and transferred from the central uData wallet to the wallets of all active participants, and a new entry containing the list of the datasets that passed the HOLD period, the actual amount of transferred tokens and the balance of the сentral wallet at the moment of payments is added to the public transaction log.
After passing the HOLD period, excluding datasets from the uData cluster becomes impossible.
The data accumulated in the uData system are used for machine learning of the SMART models that allow community members to benefit from dynamic adaptation of the web pages for visitors. The SMART models will be constantly improving and learning by using new data received from the members of the system.
In order to provide users with the right to use uData in the proportion each user contributed to the development of the system, there's a mechanism of utilising tokens which allows processing a certain volume of traffic by using SMART models, which is proportional to the total amount of convertible traffic for the last completed reporting period prior to the current one.
Each batch of 10 000 tokens entitles you to a one-time conversion in the amount of 0.001% of all traffic that went through the system.
As the capacity of each token directly depends on the volume of data that went through the traffic system, it's necessary to add reporting data on traffic flows and utilized tokens to the public transaction log.
So, at the end of each reporting period, detailed information about all traffic that went through the system (depending on the intended level of privacy, either limited to volumes of data and IDs of the system users, or containing the full information about the websites that the stated percentage of the traffic went through) will be added to the public transaction log.
Also, the log will contain records on the total amount UKT tokens returned to the central uData wallet, this amount can be compared with the total amount of rewards given for contributing new datasets to uData during this period.