Patent grant rate prediction method, system, device and storage medium

By using BERT model and convolutional neural network combined with patent literature subject information enhancement method, the problem of degradation in patent authorization prediction accuracy caused by the inability to fully utilize disclosed technologies in the prior art, and more efficient patent authorization prediction is achieved and improved The accuracy and efficiency of patent examinations are provided.

CN117371576BActive Publication Date: 2024-11-01HARBIN HUIWEN JIECHUANG ARTIFICIAL INTELLIGENCE TECHNOLOGY CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311171887.6
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-09-12
Publication Date
2024-11-01
Estimated Expiration
2043-09-12

Smart Images

  • Figure CN117371576B_ABST
    Figure CN117371576B_ABST
Patent Text Reader

Abstract

A patent grant rate prediction method, system, device and storage medium belong to the field of information processing technology, and solve the problem that the existing patent examination process cannot fully retrieve and use the disclosed prior art related to the patent, resulting in a decrease in the accuracy of the prediction. The method includes: obtaining the abstract of the patent document to be tested, vectorizing the abstract text, and obtaining the vector of the abstract; calculating the cosine similarity of the patent document to be tested and the cosine similarity of the public patent data set; selecting N public patents with the highest cosine similarity to the patent document to be tested, and training them on the patent document to be tested to obtain the vector representation of the processed text and the selected public patent; collecting the subject of the public patent document; calculating the distance between the public patent document and its subject vector as a data distribution representation; obtaining the patent grant prediction result through the convolution layer, the pooling layer and the fully connected layer. The present invention is applicable to the prediction scenario of patent grant rate.
Need to check novelty before this filing date? Find Prior Art