GNN-based COVID-19 scientific literature fine-grained classification method

Abstract

The invention discloses a GNN-based COVID-19 scientific literature fine-grained classification method. The method comprises the following steps: a) constructing a COVID-19 scientific literature knowledge graph; a1) dividing knowledge categories; a2) designing entities in scientific literatures; a3) designing a relationship in scientific literature; a4), constructing a COVID-19 scientific literature knowledge graph; b) constructing a western medicine treatment knowledge graph; c) constructing a traditional Chinese medicine treatment knowledge graph; d) constructing a graph neural network model;and e) carrying out text classification. According to the GNN-based COVID-19 scientific literature fine-grained classification method, an effective screening and classification method is provided formedical workers to quickly find needed knowledge class literatures in massive (usually more than ten thousand) related COVID-19 scientific literatures, the beneficial effects are remarkable, and themethod is suitable for application and popularization.

Description

technical field [0001] The present invention relates to a fine-grained classification method for scientific literature, and more specifically, to a fine-grained classification method for COVID-19 scientific literature based on GNN. Background technique [0002] Text classification is a common task in natural language processing, which automatically classifies and marks text sets according to certain classification systems or standards. Text classification methods fall into three categories: rule-based methods, machine learning-based methods (data-driven methods), and hybrid methods. Rule-based text classification methods (Rule-based methods) require human participation in formulating rules, which are usually more accurate. However, when the rules are changed or updated, the rules need to be manually re-summarized, and the maintenance cost is high. Moreover, when there are many rules, multiple rules may conflict, which is difficult to maintain. In terms of scalability, est...

AI Technical Summary

Problems solved by technology

[0008] (3) Poor interpretability of deep learning

[0014] Specifically, in the face of COVID-19 scientific literature data, there is currently no method for directly fine-grained classification of COVID-19 scientific literature

Images