download0 view78
twitter facebook

공공누리This item is licensed Korea Open Government License

빅데이터 분석을 활용한 연구자 맞춤형 R&D 진단 및 처방 기술 연구
Alternative Title
R&D diagnosis and prescription technology study based on big data analysis
Korea Institute of Science and Technology Information
Publication Year
funder : 과학기술정보통신부
funder : Ministry of Science and ICT
○ 산업현장에서 활용할 수 있는 지식서비스 모델이 부재하고, 해당 산업의 자생적 발전을 위한 체계적 수요조사 및 이에 대한 대응이 부족하였음
○ R&D 진단 및 처방을 위해 중소기업의 문제점 및 정보니즈를 파악하여 구체적인 requirement를 서비스로 매칭시키기 위한 3대 지식서비스 인프라에 대한 로드맵 및 각 인프라간의 서비스 시너지 향상을 위한 로드맵 구축을 완료함
○ 중소기업 지원사업에 유관한 변수를 찾기 위해 프로파일링 분석 등 다변량 분석 방법을 활용하였으며, CAID, CRT, C5.0, Quest 등 4가지 의사결정나무와 몇가지 모델을 결합한 앙상블 모델을 활용하여, 예측력이 가장 높은 모델을 확보하였음
- R&D 기획 수요기업 예측모델은 최근 진행된 설문조사 결과를 바탕으로 진행하였으며, Fisher의 판별분석 통해 8개 핵심변수를 도출하였음
○ R&D진단 및 처방기술연구를 통해 확립된 모델과 판별함수는 기계학습 기반의 지능형 비서시스템인 R&D IA(Intelligence Assistant)에 탑재하여 시범서비스 함
○ 본 연구를 통해 Biz Advisor를 지향하는 R&D진단 및 처방기술의 구현이 가능함.

(출처 : 보고서 초록 5p)

Ⅲ. Content and Result of the study

□ Information supporting environment analysis for SMEs
○ The US, EU and UK aim data competitiveness.
- It can be seen that the government will concentrate its national strategy to create a new society based on data. They are concentrating on building a system that can efficiently manage data distribution.
- It is necessary to prepare for effective data communication in Korea, especially to establish public infrastructure for data production, analysis and management and we are necessary to start social discussion for this purpose.

□ Technical environment analysis for R&D diagnosis and prescription technology platform implementation
○ In the future, everyone becomes a technician, everything becomes intelligent and instant, and all businesses are digitized. The three elements to support this future are Precision, Disposability, and Autonomy.
○ Future architecture is going to be the time of no more button (machine, chatbot) instead of the human intervention (command) process, and the power of super vendor is expected to be higher than ever.

□ Demand analysis and roadmap construction
○ Conduct to analyze the gap between the target platform and reserve resources for ML(Machine Learning) based R&D diagnosis / prescription using big data

○ A road map for three knowledge service infrastructures was established to solve problems and to identify information needs of SMEs for R&D diagnosis and prescription
- Establishing a function-oriented linkage scheme to enable the three major knowledge service infrastructures to serve as individual solutions for the integrated platform.

□ Research results of R&D diagnosis and prescription technology
○ We conducted a profiling analysis to find variables related to SMEs support projects (three major knowledge services). Multivariate analysis methods were used to express the variables that were statistically more robust and easy to use.

○ The modeling of machine learning was conducted using survey data that the government should support R&D planning stage among planning, progress, commercialization.
- We construct the 4 decision tree(CHAID, CRT, C5.0, Quest) model and ensemble model and found some models that can predict highly

○ The R&D planning demanding forecasting model was based on the recent survey results and was derived from the balanced survey results. The discriminant analysis was conducted by Fisher 's linear discriminant function, and step selection was based on Wilks' Lamda Analysis
- R&D planning needs were highly predicted when the proportion of the main product, the duration of the development, the period required for the development of the market, the number of technology development attempts is shorter or smaller

(출처 : SUMMARY 12p)
R&D진단; R&D처방; 연구자 맞춤형; 빅데이터 분석; 지식서비스; R&D diagnosis; R&D prescription; research customized; big data analysis; knowledge service
Files in This Item:
There are no files associated with this item.
Appears in Collections:
7. KISTI 연구성과 > 연구보고서 > 2017
RIS (EndNote)
XLS (Excel)