Technology Trend Forecasting and Idea Support Incorporating Generative AI Using Patent Information

View at publisher Summarise with VIRAT
Authors: Yurie Okuhara, António Oliveira Nzinga René
Cited by: 0
References: 9

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The rapid accumulation of patent documents presents challenges for traditional analysis methods due to the high dimensionality and volume of data. This study proposes an integrated framework that combines Transformer-based Sentence-BERT embeddings, UMAP dimensionality reduction, and k-medoids clustering to extract and visualize meaningful patent clusters. Selected clusters are analyzed through 2D and 3D visualizations to reveal core and peripheral technological concepts, semantic densities, and inter-cluster relationships. Furthermore, generative AI is employed to provide descriptive summaries and predictive insights for each cluster, enabling the identification of emerging technological trends and potential applications. The proposed approach enhances interpretability, supports strategic decision-making in research and development, and demonstrates a scalable method for AI-assisted patent landscape analysis.

Fields of Study & Subjects

Intellectual Property and Patents Topic Modeling Law AI and Intellectual Property

Publication details

Journal / source
International Conference on Artificial Intelligence in Information and Communication (ICAIIC)
Pages
431–436
Publication date
24 February 2026
Language
EN