University of Illinois Urbana-Champaign

Patent citation recommendation based on multimodal content and deep learning

Zhang, Jinzhu; Zhang, Yi; Zhang, Ruoyu

Permalink

https://hdl.handle.net/2142/126216

Description

Title
Patent citation recommendation based on multimodal content and deep learning
Author(s)
  • Zhang, Jinzhu
  • Zhang, Yi
  • Zhang, Ruoyu
Issue Date
2025-03-11
Keyword(s)
  • Patent citation
  • Citation recommendation
  • Deep learning
  • Patent image
  • Patent multimodal content
Abstract
The increase in the number of patents has led to higher costs for patent applicants and examiners to search and cite prior patents. Existing patent citation recommendation methods focus more on textual content such as patent titles and abstracts, while neglecting patent images. Patent images contain detailed technical information and express the core technological content of the patent, making them highly valuable as a source of technical intelligence. Therefore, this paper adopts the multimodal patent data perspective, combines the full-text content with patent images to jointly represent the patent information, and then constructs a deep learning model to accomplish more effective patent citation recommendation. Empirical studies in the field of drones show that combining patent images to represent patent features can significantly improve the effectiveness of patent citation recommendations, thus helping patent examiners or applicants to find relevant prior patents to cite.
Publisher
iSchools
Series/Report Name or Number
iConference 2025 Proceedings
Type of Resource
Other
Genre of Resource
Conference Poster
Language
eng
Handle URL
https://hdl.handle.net/2142/126216
Copyright and License Information
Copyright 2025 is held by Jinzhu Zhang, Yi Zhang, and Ruoyu Zhang. Copyright permissions, when appropriate, must be obtained directly from the authors.

Owning Collections

iConference 2025 Posters PRIMARY
Posters presented at the 2025 iConference https://www.ischools.org/iconference