Two-stage academic data web page classification method and system based on multi-dimensional feature fusion

By adopting a two-stage classification method of multi-dimensional feature fusion in web page classification, combining short text logistic regression and deep measurement learning algorithm, the problem of low web page classification accuracy in the existing technology is solved, and higher classification accuracy and generalization performance are achieved.

CN115130601BActive Publication Date: 2025-03-28SHANGHAI JIAOTONG UNIV
2 Cites 0 Cited by

Patent Information

Application Number
CN202210795308.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-07-07
Publication Date
2025-03-28
Estimated Expiration
2042-07-07

AI Technical Summary

Technical Problem

The prior art is difficult to effectively utilize a variety of features in web page classification, resulting in low classification accuracy, especially when the web page structure is complex.

Method used

A two-stage academic data web page classification method based on multi-dimensional feature fusion is adopted. First, a short text logistic regression model is used to perform preliminary classification, and then a text convolutional neural network is used to combine the depth measurement learning algorithm of triple loss to perform the second stage classification, integrating the long text and URL information of the web page.

Benefits of technology

Through multi-dimensional feature fusion, the accuracy and generalization performance of the web page classification model are improved, and academic data web pages can be screened more accurately.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The present invention provides a two-stage academic data web page classification method and system based on multi-dimensional feature fusion, which relates to the technical field of web page classification, including: step S1: based on academic keywords, input search engines for retrieval, and obtain the retrieval page content; step S2: carry out the first stage classification based on the short text logistic regression model; step S3: obtain the web page HTML information labeled as data web page after the first stage classification is completed; step S4: based on the web page long text and website information, carry out the second stage classification, and adopt the text convolution neural network combined with the deep metric learning algorithm of triple loss; step S5: store and organize the final classification results, analyze the necessary information in the web page, and display it on the data portal website. The present invention can quickly and accurately filter out data web pages from the Internet.
Need to check novelty before this filing date? Find Prior Art