User quality detection method and device

Abstract

The invention discloses a user quality detection method and device, and the method comprises the steps: building a user set which comprises a user with known user quality and a user with unknown user quality; determining the users with the known user quality in the user set as seed users, and respectively assigning quality score initial values for all users; obtaining correlation information of all users in the user set in at least one dimension, and carrying out quantification; employing a preset prediction model for multi-iteration calculation according to the correlation information, the quality scores of all users and the judgment whether all users are the seed users or not, ending the iteration calculation till an algorithm is convergent, and determining the user quality of all users according to the finally obtained quality scores of all users. According to the embodiment of the invention, the method and device can detect the user equality more accurately, improve the reliability of detection results, and reduce the waste of calculation resources.

Description

technical field [0001] This application relates to the technical field of user quality analysis in e-commerce transaction platforms, in particular to a user quality detection method and device. Background technique [0002] On e-commerce trading platforms, there is a need to identify user quality in many scenarios. For example, the activities between buyers and sellers are accompanied by contradictory relationships such as cheating and anti-cheating from the very beginning, such as: fraud and anti-fraud, false transactions and anti-fraud transactions, and so on. In the process of managing the platform, a very important content is to identify and manage bad users. Its premise is to have a good identification of bad users. In addition, there are some scenarios where it is necessary to identify high-quality users. For example, when some users need to be invited to give voting results for a certain voting event, it is generally necessary to select some high-quality users to ens...

AI Technical Summary

Problems solved by technology

Compared with the manual detection method, this automatic detection method improves the processing efficiency of a single problem. However, it is necessary to develop a different business model for each business problem, which also cannot meet the challenges of diversified business problems.

For example, to develop a false transaction user identification and detection model for fraudulent transactions, and to develop a fraudulent user identification and detection model for fraud, the cost of development and maintenance is very high

[0006] In addition, in the prior art, it is also possible to detect the quality of each user by developing a general-purpose detection model for the entire platform. However, it is often necessary to set some parameters in the detection model. For the setting of these parameters, the present In existing technologies, it is often set through experience values, or through some training samples in advance. However, in practical applications, there are often cases where the accuracy of detection results is not high, and it cannot be directly applied to specific scenarios. , causing waste of computing resources, etc.

Images