Automatic label system and system

Abstract

The invention belongs to the technical field of label setting, and particularly relates to an automatic label system and system. The automatic label system comprises a data set generation module, a training module, a prediction module, a data uploading module and a label generation module. The data uploading module is used for uploading original data information and generating an original database; and the data set generation module is used for generating a training set and a test set from an original database, and defining a label large class based on an MECE principle. According to the invention, through an automatic label generation mechanism and an automatic label setting mechanism, label creation and setting are automatic and a closed loop is formed; therefore, the problems that in a current service re-operation mode, it is difficult to quantify indexes, conduct activity and operation effect analysis, cannot truly reflect user experience and is low in data analysis efficiency are solved; through the automatic label system, a large amount of manpower work is optimized into a mode with higher precision and efficiency, and the effects of refined operation and refined risk control of offline services are achieved.

Description

technical field [0001] The invention belongs to the technical field of label setting, specifically relates to a set of automatic label system and system, in particular to a complete set of automatic label setting and adjustment system for offline stores, shop assistants and commodities. Background technique [0002] In the label setting scenario, it is often necessary to add human judgment, resulting in information distortion, which affects the product and operation personnel to make correct decisions. At present, there are two common ways of labeling in enterprises: [0003] The first method: manually conduct framework analysis, label group construction, and specific label design, and then employ a large number of content reviewers / operation personnel to analyze the data of business objects and manually perform manual processing on these objects. Label settings. However, this method will cause the labeling results to contain a large number of personal emotions of the labe...

AI Technical Summary

Problems solved by technology

In this way, it is more convenient for enterprises, but generally the third-party organizations output solutions are generally poor in versatility, and the cost of the solutions is generally high, and because it is developed by an external team, the overall communication efficiency is low and the information fidelity is low. lower

[0005] It can be seen that currently, there are defects such as poor versatility, inability to truly reflect user experience, low efficiency, and excessive manpower consumption in terms of business.

Images