Sample labeling resource allocation method, device and application

Abstract

This application proposes a sample tagging resource allocation method, device and application. The number of taggers is preset. For any transaction sample in the sample set, multiple taggers with the preset number of taggers are selected from the tagger list to generate taggers. Collect and assign to each annotator in the annotator set to annotate to obtain the annotating result; record the annotating completion time of all annotators on the transaction samples, and take the latest annotating completion time as the audit start time when the annotating result enters the audit node; Calculate the time interval between two adjacent audit start times and the historical audit interval to obtain the idleness of audit resources; adjust the number of markers according to the idleness of audit resources, and re-generate markers for the remaining transaction samples based on the adjusted number of markers member assembly. Through the present invention, the allocation of sample labeling resources can be automatically performed based on various risk control rules, which replaces the existing manual screening method, so the accuracy and efficiency are higher.

Description

technical field [0001] The present application relates to the field of data labeling, and in particular, to a method, device and application for allocating sample labeling resources. Background technique [0002] Along with the rapid development of online financial business, while the related transaction volume has grown rapidly, abnormal transactions have also appeared more and more. Abnormal transactions not only cause huge losses to users themselves, but also seriously damage the financial order, damage the reputation of financial institutions, and have a very bad impact on society. [0003] At present, with the rapid development of artificial intelligence, the identification of abnormal transactions has gradually changed from manual identification to automatic model identification, and data labeling in model identification is a basic but highly important task. The labeling accuracy will directly affect the recognition accuracy of the model. The traditional labeling tas...

AI Technical Summary

Problems solved by technology

[0006] The embodiment of the present application provides a sample labeling resource allocation method, device and application, which can solve the problem of how to allocate labeling resources in the prior art. According to this solution, the labeling efficiency is also fast, and the labeling results in the review queue are getting more and more backlogged. Many features, by controlling the idleness of audit resources in the audit node, dynamically adjust the number of markers for each transaction sample, so as to achieve a dynamic balance between improving the accuracy of labeling and improving audit efficiency

Images