Data auditing method, device and apparatus

Abstract

The embodiment of the invention provides a data auditing method, device and apparatus, and the method comprises the steps: obtaining sample data, enabling the contribution degree of the sample data toa preset auditing model to be greater than or equal to a preset contribution degree threshold value, wherein the contribution degree is a contribution value of the sample data to the optimization training of the auditing model; wherein a labeling result corresponding to the sample data meets a preset standard; performing optimization training on the auditing model according to the sample data; and auditing the data by using the auditing model after optimization training. By obtaining sample data, optimization training is performed on an auditing model by utilizing the sample data. Among numerous sample data, sample data with relatively high contribution degree to the auditing model can be selected. Compared with the prior art, the number of the sample data is reduced, the quality of optimal training of the audit model by the sample data is ensured, the quality and efficiency of auditing the data by the audit model are effectively ensured, and the cost of manual annotation is effectively reduced on the premise of ensuring that the accuracy of the audit model is not reduced.

Description

technical field [0001] The present invention relates to the technical field of communications, in particular to a data review method, device and equipment. Background technique [0002] With the continuous development of multimedia information, e-commerce platforms are developing more and more rapidly, and e-commerce platforms can realize commodity transactions. In order to increase the transaction rate, sellers on the e-commerce platform will edit and publish pictures and / or text information about the products. According to the relevant regulations of the platform, the e-commerce platform will manually review the title and content of the product published by the seller to determine whether it violates the regulations. [0003] However, with the diversified development of commodities, the contents of commodities are becoming more and more abundant. When sellers publish commodities on the e-commerce platform, they can not only include commodity titles, commodity main picture...

AI Technical Summary

Problems solved by technology

[0003] However, with the diversified development of commodities, the contents of commodities are becoming more and more abundant. When sellers publish commodities on the e-commerce platform, they can not only include commodity titles, commodity main pictures, drawings and description pictures, but also Short videos, dynamic pictures and other content can be added, which greatly increases the labor cost of commodity review, and also reduces the quality and efficiency of review

Images