AI학습용 데이터구조 및 데이터셋의 물건특허 인정방안 = Recognition Measures of Product Patent in Data Structures and Datasets for AI Learning

다국어 초록 (Multilingual Abstract)

To create an AI model, a data structure or data set is essential for deep learning training. Furthermore the data structure is used as comparison data with the data input by the user into the AI model and plays an important role in learning and reasoning. A data structure refers to the logical structure of data that represents the interrelationships between data elements, and the data structure itself cannot be patented because it can be processed by a computer or viewed as a collection that can access or search to/for materials (data).
However, the technological idea as an algorithm in which the data structure is stored in deep learning programs for AI learning and used as an information processing means to solve specific task, can be considered as an invention utilizing the laws of nature. Although the current AI Examination Practice Guide recognizes data structure storage medium claims as product inventions, claims for data structures are being rejected under Article 42, Paragraph 4, Subparagraph 2 of the Patent Act on the grounds that it is not clear whether the category of invention is an object or a method, and there is a problem that effective data protection cannot be achieved.
To address these issues, it is necessary to recognize data structures as product inventions, on the premise that the claim specifies the means of processing information processed by a specific structure in hardware such as computer etc. In cases where claims end with phrases such as “... data having a structure” or “... data structure (set),” an active consideration should be given to revising the patent law or examination standards to recognize them as product inventions.