A graph network patent retrieval method based on meta-learning ipc classification

By combining meta-learning and graph networks, the problem of low patent search efficiency was solved, achieving efficient and accurate patent similarity judgment and narrowing the scope of relevant patent searches.

CN119128142BActive Publication Date: 2026-07-24NANJING UNIV OF POSTS & TELECOMM
3 Cites 0 Cited by

Patent Information

Application Number
CN202410856212.3
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-06-28
Publication Date
2026-07-24
Estimated Expiration
2044-06-28

AI Technical Summary

Technical Problem

Existing patent search methods are inefficient when faced with large amounts of patent data, require extensive professional knowledge, and are difficult to accurately determine the novelty and relevance of patents.

Method used

A meta-learning model is adopted to classify patent texts into IPCs. A graph network is used to construct a patent similarity judgment. The graph network is constructed through noun phrases and combined with BERT word embeddings and GraphSAGE to learn the patent representation, so as to achieve unsupervised patent similarity judgment.

Benefits of technology

It improves the efficiency and accuracy of patent searches, reduces reliance on specialized knowledge, better reflects the correlation between patents, and narrows the scope of relevant patent searches.

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

Abstract

The application discloses a graph network patent retrieval method based on meta learning IPC classification, comprising constructing a binary classification task of each IPC category on a patent dataset, using a meta learning mode to enable a meta model to learn shared prior of the task, and enabling the model to adapt to a specific IPC category determination task with only a few samples; through IPC level-down sequential hierarchical classification, the IPC category to be determined is circled layer by layer; a graph is constructed based on a noun phrase, and structural information is fused on the basis of semantics to meet the requirement of accurately judging patent similarity.
Need to check novelty before this filing date? Find Prior Art