Categories
Japan IP News

July 2, 2022. JPO 2022-2026: Action Plan for Utilization of Artificial Intelligence (AI) Technology.

The JPO’s “Action Plan for Utilization of Artificial Intelligence (AI) Technology” to improve the efficiency and quality of the JPO administrative work. This is an update from the 2017 six-year plan: in fact, the JPO has promoted the project while revising the Action Plan annually.

Features an automated patent classification tool for foreign patent documents. The USPTO also announces updates on its AI/ET Partnership Program. July 2 2022 JPO and USPTO announce AI Plans.

The rapid advancement of AI technology has caused the JPO to examine how AI should be utilized over the next five years or so.  The USPTO has its own AI/ET Partnership Program.

The JPO’s “Action Plan for Utilization of Artificial Intelligence (AI) Technology” to improve the efficiency and quality of the JPO administrative work. This is an update from the 2017 six-year plan: in fact, the JPO has promoted the project while revising the Action Plan annually.

Under the AI Action Plan, the JPO has developed tools such as an automated patent classification assigning tool for foreign patent documents, a re-ranking tool for patent documents, an and image search tool for trademarks.  These tools were provided to examiners on a trial basis to improve the efficiency and quality of the JPO administrative work.

Kimura Commentary

Advances in AI are rapidly being adopted in every aspect of business and personal life. Globally of course, as well as in Japan.

On June 7, the USPTO provided an update on its AI/ET (Artificial Intelligence / Emerging technologies) Partnership. Some additional details are available here. In its own words, the USPTO states”:

“The USPTO seeks to engage the AI/ET community on ongoing and future USPTO AI/ET efforts, such as using AI and ET within the agency to enhance the quality and efficiency of patent and trademark examination. Additionally, the USPTO seeks the public’s views on various intellectual property (IP) policy issues that uniquely affect the AI/ET community.”

While we need not, perhaps, be so interested in the specific details of these AI Action Plans (although readers may well be), this is important news for Japanese patent seekers from all countries, including Japan. As we can see, the JPO and USPTO are leading the way. This is good news for patent creators – whose very livelihoods are also connected to the creation of innovative technologies.

As Patent Attorneys, we welcome such advances.  They ensure that the JPO remains at the forefront of technology to serve its clients: domestic and foreign entities seeking to protect their businesses through IP registration.

JPO AI Action Plan Background & Details per July 2 2022 JPO and USPTO announce AI Plans 

All supporting materials can be found here on the JPO website.

Major AI Achievements To Date

The JPO isolates these specific AI achievements under the Action Plan

Patent Classification

  • Predict JPO patent classifications for foreign patent documents using machine-translated Japanese text as input. By assigning the classifications, it has become possible to search foreign patent documents using JPO classifications and Japanese text queries. 

Prior Art Search (Patent Image Search)

  • Based on the “Drawings” and the brief description of drawings in the “Description” of the application being examined, patent documents with highly relevant drawings are displayed preferentially. The efficiency and quality of the prior art searches by examiners have been improved.

Prior Art Search (Re-Ranking Patent Documents)

  • Calculate scores representing the similarity based on text data and metadata (claims, description, bibliographic information, patent classifications, number of citations, etc.) contained in the application being examined and the target prior patent documents and re-rank the documents in order of the scores. The efficiency and quality of the prior art searches by examiners have been improved.

Search for Prior Graphic Trademarks (Trademark Image Search)

  • Calculate scores representing the similarity based on the image feature of the applied trademark and that of the prior trademarks and display them in order of the scores. Machine learning competition was held in FY2021 to improve search accuracy. The predictive models of the top winners will be integrated into the image search system of the JPO.

By Takaaki Kimura

Managing Partner and Patent Attorney with over thirty-five years of IP law experience.