• 한국산업융합학회 논문집
• /
• 제25권5호
• /
• pp.817-826
• /
• 2022

Patent information describes the history of technological progress in the relevant field, so it can be usefully used to identify trends in technological development and change and to establish R&D development strategies. This study proposes a method to identify the needs and problems of technology development at the planning stage of the R&D process and to analyze core technologies through patent analysis using Natural Language Processing(NLP) technology. As a big data source, collected patent documents registered in Google Patents for foldable technology, the latest technology in the display industry, and then extracted keywords using NLP analyzer. By classifying the extracted keywords into needs and problems for technology development, developed technology and materials, identified the needs of the market and customers and analyzed the technologies being researched and developed. Unlike previous studies that performed patent analysis, this methodology is different in that it can quickly and conveniently analyze the latest technology trends from big data called patents even if you do not have specialized knowledge and skills in the text mining. This study contributes to the digitalization of the R&D process based on data analysis.

• Genomics & Informatics
• /
• 제17권2호
• /
• pp.16.1-16.4
• /
• 2019

Medical Subject Headings (MeSH), a medical thesaurus created by the National Library of Medicine (NLM), is a useful resource for natural language processing (NLP). In this article, the current status of the Japanese version of Medical Subject Headings (MeSH) is reviewed. Online investigation found that Japanese-English dictionaries, which assign MeSH information to applicable terms, but use them for NLP, were found to be difficult to access, due to license restrictions. Here, we investigate an open-source Japanese-English glossary as an alternative method for assigning MeSH IDs to Japanese terms, to obtain preliminary data for NLP proof-of-concept.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2017년도 제29회 한글 및 한국어 정보처리 학술대회
• /
• pp.56-59
• /
• 2017

CNN(Convolutional Neural Network)을 이용하여 발화 주제 다중 분류 task를 multi-labeling 방법과, cluster 방법을 이용하여 수행하고, 각 방법론에 MSE(Mean Square Error), softmax cross-entropy, sigmoid cross-entropy를 적용하여 성능을 평가하였다. Network는 음절 단위로 tokenize하고, 품사정보를 각 token의 추가한 sequence와, Naver DB를 통하여 얻은 named entity 정보를 입력으로 사용한다. 실험결과 cluster 방법으로 문제를 변형하고, sigmoid를 output layer의 activation function으로 사용하고 cross entropy cost function을 이용하여 network를 학습시켰을 때 F1 0.9873으로 가장 좋은 성능을 보였다.

• 전기학회논문지
• /
• 제67권7호
• /
• pp.954-961
• /
• 2018

In this paper, a chatbot for emotional stress recognition and management using rule-based method and NLP is designed and developed to tackle various emotional stresses of people through questionnaire. For this, Dialogflow as open chatbot development platform and Facebook messenger as chatting platform are used. We can build natural and resourceful conversational experiences through predefined questions by using powerful tools of Dialogflow, and can use developed chatbot on the Facebook page messenger. Developed chatbot perceives emotional stresses of user by user-input which is either text or choice of predefined answer. It also gives user questions according to the user's feeling, and assess the strength of the emotional stresses, and provide a solution to the user. Further research can improve the developed chatbot by using open Korean NLP library and database of emotions and stresses.

• 한국정보과학회 언어공학연구회:학술대회논문집(한글 및 한국어 정보처리)
• /
• 한국정보과학회언어공학연구회 2018년도 제30회 한글 및 한국어 정보처리 학술대회
• /
• pp.199-203
• /
• 2018

음식과 맛집에 대한 사용자의 정보검색 니즈가 나날이 증가하면서 서비스 제공자가 정보 제공의 대상이 되는 맛집 상호명을 파악하는 것은 중요한 이슈다. 그러나 업종의 특성상 점포가 새로 생겨나는 주기는 매우 짧은 반면, 신규 점포의 서비스 등록 시점에는 시간적 차이가 존재하는 문제가 있다. 본 논문에서는 신규 상호명을 능동적으로 파악하기 위해 위치기반 서비스 로그에서 맛집 상호명을 추출하는 문자 기반의 딥러닝 모델 및 방법론을 제시한다. 자체 구축한 학습 데이터셋으로 실험한 결과, 제안하는 모델이 기존 기계학습 모델보다 높은 정확도로 상호명을 분류할 수 있음을 확인하였다. 또한, 사전 학습된 모델을 검색로그에 적용하여 신규 상호명 후보를 추출함으로써 향후 상호명 DB를 능동적으로 업데이트 할 수 있는 가능성을 타진하였다.

• Genomics & Informatics
• /
• 제17권2호
• /
• pp.18.1-18.10
• /
• 2019

Prediction of the relations among drug and other molecular or social entities is the main knowledge discovery pattern for the purpose of drug-related knowledge discovery. Computational approaches have combined the information from different sources and levels for drug-related knowledge discovery, which provides a sophisticated comprehension of the relationship among drugs, targets, diseases, and targeted genes, at the molecular level, or relationships among drugs, usage, side effect, safety, and user preference, at a social level. In this research, previous work from the BioNLP community and matrix or matrix decomposition was reviewed, compared, and concluded, and eventually, the BioNLP open-shared task was introduced as a promising case study representing this area.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2023년도 추계학술발표대회
• /
• pp.1045-1046
• /
• 2023

본 논문은 디지털 소외계층과 사회적 약자를 고려한 보이스피싱 예방 솔루션을 제안한다. 통화 내용을 AWS Transcribe를 활용한 STT와 NLP 알고리즘을 사용해 실시간으로 보이스피싱 위험도를 파악하고 결과를 사용자에게 전달하도록 한다. NLP 알고리즘은 KoBIGBIRD와 DeBERTa 모델 각각을 커스터마이즈하여 보이스피싱 탐지에 적절하게 파인튜닝 했다. 이후, 성능과 인퍼런스를 비교하여 더 좋은 성능을 보인 KoBIGBIRD 모델로 보이스피싱 탐지를 수행한다.

• 산경연구논집
• /
• 제11권2호
• /
• pp.33-39
• /
• 2020

Purpose: The purpose of this study is that victims of technology leaks and people concerned about leaks complain of stress over security concerns. However, there are no psychological treatments among the government's comprehensive plans to prevent technology leaks. Therefore, the government intends to present education methods using the NLP (Neuro Linguistic Program), a collective counseling technique, to heal the psychological injury of the victims. Psychological counseling methods include cognitive behavioral therapy, psychoanalytic behavioral therapy, humanism therapy, art therapy, and other psychological therapies. Among them, NLP (Neuro Linguistic Programming) method was used. NLP has three concepts: neuron, language, and programming, and is used as a general method for group counseling. Research design, data and methodology: In relation to composition, Chapter 1 explained the purpose and necessity of the study, Chapter 2 explained the types of psychological counseling and NLPs to help understand the study, introduced the prior study related to the development of collective counseling programs through NLP, and Chapter 3 developed a security psychological counseling education program. In addition, FGI(Focus Group Interview) was conducted for professionals. Results: Corporate counseling considered most in this study should satisfy client, counselor and manager differently from individual counseling. For this purpose, the result was composed of 11 times. In order to derive personal problems for clients, they consisted of finding, loving, expressing, and emancipating self. And, It solved the leakage anxiety to suggest a professional solution for the counselor. In addition, this course helps them become familiar with counseling techniques for becoming a good security administrator. Lastly, it was configured to leave the result for the manager to suggest the organizational development method through this training. The implication of this study is to derive psychological counseling methods for security officers. Most companies in the field of security counseling complain about technology leakage stress. There is currently no psychotherapy support project under the policy. And It was developed because it can expect sales improvement from security consultation. Conclusions: In conclusion, the results were organized to be left to the manager so that he could suggest how to develop the organization through this time.

• 한국정보과학회:학술대회논문집
• /
• 한국정보과학회 2004년도 가을 학술발표논문집 Vol.31 No.2 (1)
• /
• pp.778-780
• /
• 2004

문법 구축은 NLP 작업에서 중요한 역할을 한다. 이 논문에서는 트리뱅크 코퍼스에서 자동으로 어휘화 문법을 추출하는 시스템을 소개한다 문법 자동 추출 시스템에서 자동으로 추출한 어휘화 TAG 문법, CFG 문법, 의존관계 등 여러 정보는 이후 한국어 파서 구현 및 다양한 NLP 연구에 사용된다.

• Journal of Korean Neurosurgical Society
• /
• 제45권4호
• /
• pp.213-218
• /
• 2009

Objective : This study is a retrospective clinical study over more than 4 years of follow up to understand the mechanism of load sharing across the graft-bone interface in the static locking plate (SLP) fixation compared with non-locking plate (NLP). Methods : Orion locking plates and Top non-locking plates were used for SLP fixation in 29 patients and NLP fixation in 24 patients, respectively. Successful interbody fusion was estimated by dynamic X-ray films. The checking parameters were as follows : screw angle (SA) between upper and lower screw, anterior and posterior height of fusion segment between upper and lower endplate (AH & PH), and upper and lower distance from vertebral endplate to the end of plate (UD & LD). Each follow-up value of AH and PH were compared to initial values. Contributions of upper and lower collapse to whole segment collapse were estimated. Results : Successful intervertebral bone fusion rate was 100% in the SLP group and 92% in the NLP group. The follow-up mean value of SA in SLP group was not significantly changed compared with initial value, but follow-up mean value of SA in NLP group decreased more than those in SLP group (p=0.0067). Statistical analysis did not show a significant difference in the change in AH and PH between SLP and NLP groups (p>0.05). Follow-up AH of NLP group showed more collapse than PH of same group (p=0.04). The upper portion of the vertebral body collapsed more than the lower portion in the SLP fixation (p=0.00058). Conclusion : The fused segments with SLP had successful bone fusion without change in initial screw angle, which was not observed in NLP fixation. It suggests that there was enough load sharing across bone-graft interface in SLP fixation.