Patent Office Opens 450,000 Data Sets for Pharmaceutical Experiments and AI Training Free of Charge

[Asia Economy (Daejeon) Reporter Jeong Il-woong] The Korean Intellectual Property Office announced on the 12th that it will publicly release 450,000 pieces of information free of charge, including pharmaceutical experimental data contained in patent gazettes and artificial intelligence training data for extracting experimental data.

The related data will be available through KIPRIS Plus starting from the 13th.

The pharmaceutical experimental data to be opened is information constructed as a database by processing and analyzing tables in image (figure) form included in patent gazettes in the pharmaceutical field, such as pharmaceutical ingredient names, experimental methods, and experimental values.

The artificial intelligence training data includes image classification information to extract only table forms from various image types of experimental data such as tables, graphs, and chemical formulas; table structure information (rows × columns) to accurately understand table data; and experimental data attribute classification information to automatically classify ingredient names and experimental values used in experiments.

By utilizing pharmaceutical experimental data and artificial intelligence training data, ‘table experimental data’ in image form included in patents and papers can be converted and extracted into text format data that can be analyzed.

Based on this, intellectual property service providers can develop services for extracting and utilizing experimental data from patent gazettes, and related companies and research institutions are expected to freely analyze experimental examples and comparisons included in patent gazettes to utilize them in research and development of vaccines, new drugs, and more, according to the Korean Intellectual Property Office.

Meanwhile, KIPRIS Plus is a data open platform operated by the Korean Intellectual Property Office that provides data products such as major industrial property rights (patents, trademarks, designs) gazettes and patent administrative information from 13 countries domestically and internationally.

Recently, it has also opened artificial intelligence training data for multilingual translation, image search, and more, offering a total of 115 data products in file or OpenAPI (Open Application Programming Interface) form.

OpenAPI is a public interface that allows anyone to call and utilize standardized data in real time.

This content was produced with the assistance of AI translation services.