• 한국컴퓨터정보학회논문지
• /
• 제27권2호
• /
• pp.15-23
• /
• 2022

특허 분야에서 자연어처리(Natural Language Processing) 태스크는 특허문헌의 언어적 특이성으로 문제 해결의 난이도가 높은 과제임에 따라 한국 특허문헌에 최적화된 언어모델의 연구가 시급한 실정이다. 최근 자연어처리 분야에서는 특정 도메인에 특화되게 사전 학습(Pre-trained)한 언어모델을 구축하여 관련 분야의 다양한 태스크에서 성능을 향상시키려는 시도가 지속적으로 이루어지고 있다. 그 중, ELECTRA는 Google이 BERT 이후에 RTD(Replaced Token Detection)라는 새로운 방식을 제안하며 학습 효율성을 높인 사전학습 언어모델이다. 본 연구에서는 대량의 한국 특허문헌 데이터를 사전 학습한 KorPatELECTRA를 제안한다. 또한, 특허 문헌의 특성에 맞게 학습 코퍼스를 정제하고 특허 사용자 사전 및 전용 토크나이저를 적용하여 최적화된 사전 학습을 진행하였다. KorPatELECTRA의 성능 확인을 위해 실제 특허데이터를 활용한 NER(Named Entity Recognition), MRC(Machine Reading Comprehension), 특허문서 분류 태스크를 실험하였고 비교 대상인 범용 모델에 비해 3가지 태스크 모두에서 가장 우수한 성능을 확인하였다.

• 한국문헌정보학회지
• /
• 제53권2호
• /
• pp.179-199
• /
• 2019

본 연구의 목적은 특허 문헌 분류에 가장 적합한 방법론을 발견하기 위하여 다양한 자질 추출 방법과 기계학습 및 딥러닝 모델을 살펴보고 실험을 통해 최적의 성능을 제공하는 방법론을 분석하는데 있다. 자질 추출 방법으로는 전통적인 BoW 방법과 분산표현 방식인 워드 임베딩 벡터를 비교 실험하고, 문헌 집합 구축 방식으로는 형태소 분석과 멀티그램을 이용하는 방식을 비교 검토하였다. 또한 전통적인 기계학습 모델과 딥러닝 모델을 이용하여 분류 성능을 검증하였다. 실험 결과, 분산표현 방법과 형태소 분석을 이용한 자질추출 방법을 기반으로 딥러닝 모델을 적용하였을 경우에 분류 성능이 가장 우수한 것으로 판명되었으며 섹션, 클래스, 서브클래스 분류 실험에서 전통적인 기계학습 방법에 비해 각각 5.71%, 18.84%, 21.53% 우수한 분류 성능을 보여주었다.

• 한국의상디자인학회지
• /
• 제17권4호
• /
• pp.67-76
• /
• 2015

The purpose of this study is to set a direction for the development of men's underwear after analyzing current applications for patent regarding men's special-purpose underwear. In terms of a research method, the disclosed patents and utility models were investigated using the patent information database provided by Korea Institute of Patent Information (KIPRIS, http://www.kipris.or.kr). For this, the patents applied from 1990 to October 2015 were targeted. The keywords used for patent search were 'men's underwear' and 'men's special-purpose underwear.' When searched by the keywords above, a total of 1,089 cases were found. Except for expired or cancelled ones, 243 cases were investigated. Then, annual application trends, current registrations on literature records, classification of utility model right holders and contents by topic were analyzed. In terms of data analysis, frequency analysis, crosstabulation analysis and multiple response analysis were conducted, using SPSS 18.0. The results found the followings: In terms of annual application trends, the number of applications for patent started to gradually increase since 2007. Since 2011, it has rapidly increased. In terms of the number of patent registrations, literature registration was far higher than utility model registration. In terms of application rights, 'individually registered (58.8%)' was higher than 'registered by the organization (41.2%).' Among 243 cases, 'underwear (58%)' was the highest, followed by 'men's underwear-related items (29.2%)' and 'thermals (8.2%).' According to analysis on the details of the patent applied for men's underwear, 'penis-scrotum separation' was most focused, followed by 'disposable product' 'airy features,' 'scrotum protection' and 'structure of underwear.'

• 산경연구논집
• /
• 제6권3호
• /
• pp.27-33
• /
• 2015

Purpose - This paper's aim is to analyze the technological information in patent databanks as a strategy in prospecting for new technologies. Research design, data, and methodology - We detail the major free electronic database sources for patent information, the patent documents, the patent document structures, INID codes (Internationally Agreed Numbers for the Identification of Data), indexation, references, and classification notions. Additionally, we review and analyze information on the activities of the Center of Dissemination Documentation and Technological Information (CEDIN) from the National Institute of Intellectual Property (INPI) of Brazil for the period 2000 to 2011. Results - The research shows that the technological information contained in the patents could provide a wide range of functionality within companies and universities. Conclusions - In recent years, (CEDIN), a specialist in intellectual property, has been serving internal and external users by providing guidance on the basis of patents and other literature, but the number of users served is still small. In order to familiarize more potential users of such technological information, task forces should be created among INPI, universities, and companies.

• Korean Journal of Acupuncture
• /
• 제34권4호
• /
• pp.191-208
• /
• 2017

Objectives : The purpose of this study was to analyze the trend of pharmacopuncture in Korean patent in order to establish database for patent technology. Methods : Electronic literature searches for Korean patents related to pharmacopuncture were performed in two electronic databases (Korea Intellectual Property Right Information Service and National Digital Science Library) to June 2017. Patents that were not Korean ones, did not use medicinal herb, only described method of manufacture, or had nothing to do with pharmacopuncture were excluded in this study. The status and application date of patents, Medicinal herb, target diseases, International Patent Classification (IPC), model of experiment and extracting methods were analyzed. Results : A total of 379 patents were retrieved. Based on our inclusion/exclusion criteria, 297 patents were excluded. Of 82 included patents, 27 patents did not include experiments using pharmacopuncture, and 9 patents were invented for treating animals such as pig or calf. In IPC analysis, Bee Venom, Panax (ginseng), Angelica, and Paeoniaceae were used frequently. Musculoskeletal diseases were the most targeted diseases followed by nervous diseases. For extracting, hot water extraction, distillation extraction, and solvent extraction using alcohol, ethanol, or methanol for solvent were commonly used. Conclusions : These data are useful for inventing new patent and extending range of pharmacopuncture in clinical use, however, more systematically analyzed patent studies and pharmacopuncture-related studies for new application on various diseases are needed in further studies.

• Tribology and Lubricants
• /
• 제38권4호
• /
• pp.143-151
• /
• 2022

This study analyzes patents related to oil diagnosis of construction equipment. Oil diagnosis is extremely important for maintaining construction equipment properly. Through the evaluation of existing patents, a patent strategy for the future construction equipment market is presented. The related patents are classified and selected in several steps. Finally, 16 valid patents are selected and analyzed in detail. In the classification process, patents are classified by country, year, and company. A market analysis shows that the top 10 companies have a market share of more than 50. In addition to patents related to the oil analysis of construction equipment, patents related to automobile oil analysis and development of oil sensors are investigated to identify the contents of patents in other fields that can be applied to oil diagnosis technology for construction equipment. Moreover, not only the contents of research articles of two Korean construction companies, but also the research trends in the literature in this field are used in the analysis. The related patents of the two Korean companies are few. Companies with a high market share, including Caterpillar, hold many patents, and patents for diagnosis algorithms using such technologies as artificial intelligence and artificial neural networks, along with oil sensor-based condition monitoring technology, are gradually expanding.