1Doctoral Candidate, Department of Intellectual Property Convergence, Chungnam National University, Republic of Korea
2Manager, R&D Planning Team, Korea Institute of Patent Information, Republic of Korea
3Manager, R&D Team, Korea Institute of Patent Information, Republic of Korea
4Graduate Student, Department of Intellectual Property Convergence, Chungnam National University, Republic of Korea
5Korea Invention Promotion Association, Republic of Korea
Correspondence to Dong-Hun Noh, E-mail: laborh@kipi.or.kr
Volume 19, Number 1, Pages 159-177, March 2024.
Journal of Intellectual Property 2024;19(1):159-177. https://doi.org/10.34122/jip.2024.19.1.7
Received on December 21, 2023, Revised on January 30, 2024, Accepted on February 29, 2024, Published on March 30, 2024.
Copyright © 2024 Korea Institute of Intellectual Property.
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives (https://creativecommons.org/licenses/by-nc-nd/4.0/) which permits use, distribution and reproduction in any medium, provided that the article is properly cited, the use is non-commercial and no modifications or adaptations are made.
Intellectual property customer counseling is an important public service that supports the creation of intellectual property rights and protection of the rights and interests of applicants and rights holders. To effectively support customers and secure the use of counseling content as a policy, counseling contents are classified according to certain criteria. Until 2020, it was professional counselors who directly classified these contents, but 2021 saw a shift toward automatic classification based on text analysis (TA) of the consultation texts. However, an investigation as to the distribution of counseling case classification over the past five years showed some differences between the 2018–2020 distribution, classified by professional counselors, and the 2021–2022 distribution, automatically classified by TA. Therefore, this study investigated how to improve the performance of the automatic classification system using BERT, a transformer-based AI model. After fine-tuning the BERT model, which was pre-trained using patent counseling text data and professional counselor classification values data, it was observed that the BERT’s automatic classification distribution was more similar to that of professional counselors than the classification distribution of the existing TA. These results show that the future application of the “Patent Consultation Classification BERT,” a tentative name for the model, to automatic patent consultation classification may yield a better performance than the current TA method. Furthermore, if the automatic classification results become more reliable through the use of this AI model, the purpose behind the policy for the automation of this procedure—namely easing the burden and improving the efficiency of professional counselors—may be achieved with improved continuity and stability. This may then enable a more accurate identification of the current status of patent customer counseling services and customer needs.
Intellectual property consultation, automatic classification, artificial intelligence model, BERT model, model learning
No potential conflict of interest relevant to this article was reported.
The author received manuscript fees for this article from Korea Institute of Intellectual Property.