Data labeling method and device and storage medium

Abstract

The invention provides a data labeling method and device and a storage medium, and the method comprises the steps: receiving and executing a data labeling request transaction, recording Hash values and download addresses of a plurality of pieces of to-be-labeled data on a block chain, and freezing to-be-paid loyalty for a requester account in a contract; receiving and executing a labeling task receiving transaction, randomly distributing a plurality of first to-be-labeled data for a labeling party, and recording corresponding Hash values and downloading addresses on a block chain, so that a second terminal accesses the downloading addresses to download the first to-be-annotated data after obtaining the Hash values and the downloading addresses, generates a first labeling result and packages the first annotation result to generate an annotation result to submit the transaction; receiving and executing an annotation result submitting transaction, and judging whether the first to-be-annotated data is successfully annotated or not: if yes, recording an annotation completion state and an authentication annotation result on the block chain so as to be acquired by the first terminal; andaccording to the frozen to-be-paid revenue, issuing the revenue. According to the invention, high-efficiency data labeling is realized, and the risk of leakage of an original data set is avoided.

Description

technical field [0001] The present application relates to the technical field of the Internet, and in particular to a data labeling method, device and storage medium. Background technique [0002] In the era of artificial intelligence, there are more and more application systems based on big data, such as personalized recommendation systems, and so on. The core of this type of system is based on the data model trained by historical big data, and the accuracy of the original data itself is the most critical factor affecting the quality of the model. With the exponential increase in the amount of data, it is particularly important to accurately and effectively label the original data. [0003] In the existing data labeling schemes, the following two types of schemes are usually adopted: [0004] One is to label through the employees or user groups of the enterprise. Due to the limited number of employees or users, it is difficult to quickly and efficiently label the increasi...

AI Technical Summary

Problems solved by technology

[0004] One is to label through the employees or user groups of the enterprise. Due to the limited number of employees or users, it is difficult to quickly and efficiently label the increasingly large data sets;

[0005] The other is to outsource the original data to the service provider for labeling. This type of solution has the risk of the service provider leaking the original data set

Images