
Abstract:The Japan Patent Office (JPO) provides IT services to users for accessing patent information, and through the J-PlatPat platform, English machine translation is offered, which is publicly accessible to both domestic and international users. Recently, J-PlatPat has been updated to improve functionality for users, enhancing various features.
Additionally, the JPO is creating bilingual dictionaries and parallel corpora for higher-priority languages to expand machine translation capabilities. They are also conducting research on the effectiveness of improving translation quality through training the machine translation engine, and the research reports are publicly available.
In this presentation, he will discuss the updates to J-PlatPat, report on the results of the research project related to machine translation, and outline the initiatives regarding machine translation of patent information at the JPO.
Abstract:In this presentation, Mr. Bruno Pouliquen from the World Intellectual
Property Organization (WIPO) will discuss the advancements in patent
and scientific literature translation, with a focus on WIPO's in-house
tool: WIPO Translate. This is specifically trained for patents and is
fully integrated in WIPO search engine Patentscope, it currently
covers 17 languages and is also used for non-patent literature (NPL)
translation. The presentation will delve into the development and
application of WIPO Translate, highlighting its role in improving
accessibility and understanding of intellectual property across
linguistic barriers. In addition, Mr. Pouliquen will present automatic
classification using the International Patent Classification (IPC)
system, also used to classify NPL documents. He will also share
insights on WIPO's experiments with image similarity,
speech-processing and Large Language Models (LLMs) in the intellectual
property field.