Management and mining contract classification method based on semantic understanding

The combination of DeepCNN, BiLSTM, and BERT models in the proposed method effectively captures and fuses semantic features for improved contract classification, overcoming the limitations of single models and enhancing classification accuracy in the government procurement domain.

CN119719365APending Publication Date: 2025-03-28QILU UNIVERSITY OF TECHNOLOGY (SHANDONG ACADEMY OF SCIENCES) +2
0 Cites 0 Cited by

Patent Information

Application Number
CN202411356196.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-09-27
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

Current contract classification technologies, particularly in the government procurement domain, face inefficiencies due to reliance on handcrafted features, slow training speeds, gradient vanishing issues, and the inability of single models to effectively capture both global and local semantic features, leading to low classification accuracy for government contracts with similar text structures and specialized terminology.

Method used

A method combining DeepCNN and BiLSTM networks for feature extraction, integrated with a Trans-Encoder and BERT pre-trained models, and utilizing Softmax classifiers for enhanced semantic understanding and classification, addressing the limitations of single models by fusing global and local semantic features and improving feature generalization.

Benefits of technology

This approach significantly enhances classification accuracy and robustness by effectively extracting and fusing text features, improving the precision of contract classification and addressing the limitations of existing methods in the government procurement domain.

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

Abstract

The invention relates to the technical field of contract classification, and provides a semantic understanding-based government purchasing contract classification method, which comprises the following steps of S1, collecting government purchasing contract data, performing category labeling to form a data set, preprocessing the data set, and dividing the data set into a training set and a test set; s2, establishing a contract classification model which comprises an input layer, a feature extraction layer and an output layer; and S3, inputting the data set into the contract classification model for training and prediction. The method has the beneficial effects that the efficient extraction of the text features is realized, and the probabilities of different categories are predicted through the Softmax classifier in the output enhancement layer, so that the extracted text features are further accurately classified, and meanwhile, the generalization ability of the model to complex features is also enhanced.
Need to check novelty before this filing date? Find Prior Art