Source information classification method, device, electronic equipment, storage medium

Abstract

The present invention provides a source information classification method, device, electronic device and storage medium. The method includes: judging whether the source information to be classified is quasi-valid information based on rule matching; if so, extracting features of the source information to be classified to obtain a feature vector ; Input the feature vector into at least two basic learners respectively; Input the output of the at least two basic learners to a primary learner; And according to the output of the secondary learner, determine whether the source information to be classified is not for valid information. The invention combines rule matching and classification model algorithm to realize the classification of the source of goods information released by the consignor, so as to improve the generalization ability, fault tolerance ability, self-learning and self-adaptive ability of the classification algorithm, so as to realize the identification of non-source information, so as to avoid Advertising information, sensitive information, complaint information and other information unrelated to the source of goods are published on the platform.

Description

technical field [0001] The present invention relates to the field of computer technology, and in particular, to a method, device, electronic device and storage medium for classifying source of goods information. Background technique [0002] With the development of the Internet and information technology, online freight platforms have become more and more popular among cargo owners and drivers. The cargo owner releases the source information through the cargo platform, and the driver browses and accepts orders to realize the docking process before the cargo. [0003] In the freight platform, the source of goods information is input by the owner and released. There are often some advertisements or abusive complaints, or illegal information containing prohibited keywords. The platform needs to manage and control the released supply information, so as to prevent advertising information, sensitive information, complaint information and other information unrelated to the supply ...

AI Technical Summary

Problems solved by technology

However, manual recognition is time-consuming and labor-intensive, so algorithms are usually used to solve this problem

[0004] However, simply using the addition of special rules for identification will lead to lengthy rules, and in order to achieve the matching of each rule, there may also be cases where the matching time is too long

At the same time, there are too many situations that need to be identified, and it is difficult to summarize comprehensively. Therefore, the generalization ability and fault tolerance ability of rule matching are not strong enough, and there is no self-learning and self-adaptive ability

In addition, if only a single type of neural network is used to classify, because the approximation and generalization capabilities of the network model are closely related to the typicality of our training samples, it is currently difficult to ensure that the collected training data is typical, and some The computational cost of the neural network model is relatively high, and it is difficult to achieve the same excellent effect in the industry in the industry

Images