중소기업 R&D 사업 유형화를 위한 기초연구 : 클러스터링 기법을 중심으로

0kr60k1HmMB2RYhIHSbsmQ==

연구성과

연구보고서

중소기업 R&D 사업 유형화를 위한 기초연구 : 클러스터링 기법을 중심으로 A Foundational Study on the Typology of SME R&D Programs : Focusing on Clustering Methods

중소기업 R&D 사업 유형화를 위한 기초연구 : 클러스터링 기법을 중심으로원문 다운로드 중소기업 R&D 사업 유형화를 위한 기초연구 : 클러스터링 기법을 중심으로원문보기

보고서명
중소기업 R&D 사업 유형화를 위한 기초연구 : 클러스터링 기법을 중심으로
보고서명(영문)
A Foundational Study on the Typology of SME R&D Programs : Focusing on Clustering Methods

책임자 진승화
소속기관과학기술정책연구원
내부연구참여자이현익,신기윤,김선우,조혜진
외부연구참여자
발행기관 과학기술정책연구원
ISBN979-11-24145-10-4
출판년도2025
페이지54
보고서유형 수시연구보고서
연구유형 기초
표준분류 과학기술 > 과학기술일반
자료유형연구보고서
공공누리유형 4유형 (출처표시+상업적이용금지+변경금지)
주제어중소벤처기업,중소기업R&D,중소기업R&D지원사업,중소기업기술혁신지원(KOSBIR),정부R&D지원

조회수

조회수 반영까지 최대 10분 정도 소요될 수 있습니다.

5
다운로드

다운로드 수 반영까지 최대 10분 정도 소요될 수 있습니다.

0

■연구 목적
○ 중소기업 R&D 지원이 양적 확대에서 질적 성장·사업화 중심으로 전환되면서 미국 SBIR의 3단계(PoC-R&D-사업화) 체계 적용 가능성 진단 필요
○ 부처별로 산재된 중소기업 R&D 사업의 텍스트 정보를 DB화하고 텍스트 기반의 클러스터링 분석을 수행하여, 단계별 유형화 및 현행 구조 진단하고자 함

■주요 내용
○ ‘2025년 KOSBIR 통합공고’ 대상 사업의 각 사업별 텍스트 정보를 분석 대상으로 설정하고, 텍스트 전처리–단계별 분류 모델–클러스터링 기반 토픽 모델링의 분석방법론을 활용
- PoC-R&D-사업화 3단계로 분류하되, 키워드 편향 문제를 보정하기 위해 PoC·사업화 우선 분류 및 핵심 키워드 가중치 조정을 적용하는 등의 방식을 적용
○ 2025년 KOSBIR에 해당되는 중소기업 R&D 사업은 R&D단계를 기반하여, 위 논리구조에 따라 PoC 29개(30.5%), R&D 11개(11.6%), 사업화 55개(57.9%)로 분류
- (PoC) 실증, 산업, 에너지, 확보, 반도체, 시스템, 혁신, 환경, 중소기업, 수요 등 상위 10개 키워드하에, 분류의 핵심키워드는 ‘실증’으로 나타남
- (R&D) 안전, 항만, 저장, 핵심, 산업, 관리, 지역, 환경, 공연, 니켈 등이 주요 키워드로 도출되며, 핵심키워드는 ‘안전’, ‘항만 과 같은 기술영역으로 나타남
- (사업화) 글로벌, 시장, 제품, 투자 등 개발된 기술을 시장으로 연결하는 후속 단계의 목적성이 강하게 관찰되며, 디자인, 중견 등의 키워드는 동 유형의 사업이 기술 외적인 상용화, 스케일업 요소를 지원한다는 것을 시사
○ 토픽별 유형화 분석 결과, 중소기업 R&D 사업은 대규모 공공·에너지, 신산업·첨단제조, 바이오·서비스 융합, 수출·전략산업, 자본집약형 제조, 지역·산학연 협력, 그리고 창업·환경·표준 등 독자적 미션의 이상치 유형으로 구성된 다층적 포트폴리오 구조를 형성

■정책 대안
○ (정책) 중소기업 R&D 지원체계를 단계별(PoC-R&D-사업화) 연계형 구조로 전환하여, 부처 간 역할 분담을 전제로 한 성과창출 중심의 정책 포트폴리오로 재설계 필요성을 확인
○ (제도) KOSBIR 통합공고를 중소기업 관점에서 표준화·단순화하여, 실질 지원 규모와 대상이 명확히 드러나는 투명하고 접근 가능한 정보 제공 체계로 개선 필요

A Foundational Study on the Typology of SME R&D Programs: Focusing on Clustering Methods

ㆍProject Leader: Seunghwa Jin
ㆍParticipants: Hyunik Lee · Kiyoon Shin · Sunwoo Kim · Hyejin Jo

The R&D support system for SMEs has long been operated based on quantitative expansion, but in recent years, a qualitative transition from R&D results to practical market performance has been emphasized. The KOSBIR system was introduced in 1998 by benchmarking the US SBIR model, and the Ministry of SMEs and Startups announces the integrated R&D projects for SMEs by 19 ministries and public institutions every year. However, it is necessary to systematically diagnose whether the step-by-step support structure of proof of concept, R&D, and commercialization, like the US SBIR, is actually working 20 years after its operation.
The SME R&D project is a government-led technology innovation support policy to strengthen the technological competitiveness of SMEs with limited technological innovation capabilities, and was introduced as a KOSBIR system by benchmarking the US SBIR in 1998. In accordance with the Framework Act on SMEs, the Ministry of SMEs and Startups announces the integrated R&D projects for SMEs by 19 ministries and public institutions at the beginning of every year, and large-scale R&D operating institutions provide a certain percentage of the budget to SMEs. Based on the announcement of the KOSBIR integration in 2025, a total of 95 projects were set up with a budget of 1.256 trillion won and a budget of 4.5924 trillion won for continuing projects, with the Ministry of Industry accounting for the largest amount with about 583.3 billion won in 46 projects, while the Ministry of SMEs and Startups is pushing for a private investment-linked and customized support strategy for each stage of growth with about 299.5 billion won in nine projects. The average development period of the entire project is 3.6 years, with mid- to long-term projects accounting for more than half of the total, while the Ministry of Environment has the longest development period with an average of 4.1 years, while the Ministry of SMEs and Startups has an average of 2.1 years, reflecting the characteristics of the project that supports small and medium-sized businesses to quickly enter the market.
This study constructed a project DB by extracting the source text from the announcement and project explanation data of the project subject to the KOSBIR integration announcement in 2025, and integrated 9-10 text fields such as project name, project overview, support purpose, and support content and reorganized them into analysis documents. After pre-processing text such as morpheme analysis, a three-step (PoC-R&D-commercialization) classification model was applied to construct a conditional classification structure and perform a BERTopic-based clustering analysis. Considering the technological maturity and innovation cycle, PoC is defined as a stage of verifying the technical feasibility of ideas, such as proof of concept, application of trials, and construction of testbeds, R&D is a full-fledged technology development stage such as core source technology, basic and applied research, and commercialization is defined as a stage of creating market performance after technology development such as commercialization, market entry, marketing, scale-up, and globalization.
When general keyword analysis was applied, 92 out of 95 projects were classified into R&D stages, and to correct this, keywords that cause bias, such as ‘research’, ‘development’, and ‘technology’, were separated and excluded from the list of key keywords in the R&D stage. After calculating the keyword scores of the PoC and commercialization stages first, giving high weights to the key keywords in each stage, those that did not belong to the two were classified as “technology development-oriented projects without the nature of PoC and commercialization” exclusively and specifically redefined as R&D. As a result, the KOSBIR business in 2025 was reorganized into 29 PoCs (30.5%), 11 R&D (11.6%), and 55 commercialization (57.9%), with the commercialization stage accounting for the largest portion, the PoC stage showing a homogeneous structure composed of small business groups, and the R&D stage showing the largest budget deviation by project.
As a result of c-TF-IDF-based text analysis, the policy functions were separated for each step, and the differences reflecting the characteristics of each project were identified. PoC is composed of a project that applies and verifies specific industrial technologies such as energy, semiconductors, and environment to the field with ‘evidence’ as a key keyword, and it is interpreted that the frequency of SME keywords is very low, so it is focused on technology verification within a specific industry. R&D showed social problem-solving R&D characteristics as very specific and specialized technical domain keywords such as safety, port, nickel, and performance ranked high. Commercialization strongly revealed the purpose of a subsequent step of connecting technologies developed through keywords such as global markets, products, investments, design, and mid-sized markets, and suggested that it supports commercialization and scale-up factors outside of technology.
The Ministry of Trade, Industry and Energy and the Ministry of SMEs and Startups analyzed that the form of emphasis on commercialization is similar, but the Ministry of Industry focuses on providing empirical infrastructure centered on national strategic industries and supporting commercialization of mid-sized companies using design and standards, while the Ministry of SMEs and Startups showed a commercialization-oriented structure such as innovation strategies through private cooperation and globalization and export support for startups. This suggests that the two ministries are targeting different stages (PoC vs commercialization), targets (medium vs start-ups), and budget size in the R&D innovation cycle, contrary to concerns about overlapping roles between ministries.
As a result of categorization analysis by topic, it was classified into six key types and one outlier, and it was found that small and medium-sized R&D projects are currently being operated according to specific standards (industry, ministries, etc.). Type 1 is large-scale public and energy R&D, 25 projects are characterized by large-scale national-led projects, and Type 2 is new industry and advanced manufacturing R&D, and 14 projects are focused on national strategic technology fields such as semiconductors, AI, and next-generation. Type 3 is bio-service convergence R&D, which is a mixture of traditional bio and service R&D in 13 projects, and Type 4 is export and strategic industry R&D, which includes 12 projects for commercialization and overseas expansion. Type 5 is capital-intensive manufacturing R&D, which combines traditional manufacturing and manufacturing-based technologies in nine projects, and Type 6 is regional and industry-academic cooperation R&D, which consists of nine projects that support regional-based industry-academic consortiums rather than individual companies.
As a result of the primary categorization analysis, 92 out of 95 projects are classified into R&D stages, indicating that most of the current SMEs R&D projects are concentrated in the R&D stage, and support for each R&D stage may be difficult. The pure R&D stage project showed the smallest and homogeneous structure among the three stages, with an average budget of 4.56 billion and a median of 3 billion almost identical. These projects focused on the development of small-scale niche technologies such as ports, nickel, and groundwater, which contradicted the conventional wisdom that R&D support projects would lead the large budget. This means that when R&D performance in the commercialization stage accounts for a large proportion, government financial investment in terms of pure R&D is not high, and suggests that it is necessary to review the role and scale of pure R&D support projects in the policy portfolio.
In a large framework, it was classified into R&D stages, so the structure of three-stage linkage support such as SBIR could not be confirmed, and it is necessary to enhance R&D performance creation for SMEs by strengthening the linkage of government support. If inter-ministerial linkage is insufficient, there may be a gap in policy linkage from the perspective of SMEs, so inter-ministerial linkage is important in the current structure and it is expected that the transition to an R&D-level linkage support system is essentially appropriate for sustainable growth. In the analysis by type, the Ministry of Trade, Industry and Energy was found to play a dominant role in most commercial and industrial R&D types, including new industries, high-tech manufacturing, regional/industrial cooperation, and capital-intensive manufacturing, and it was confirmed that types that require demonstration and infrastructure construction, such as energy/safety or equipment/robot, have a very high average budget, while regional/cooperation types have the lowest budget, so the scale of support is clearly differentiated and distributed according to policy purposes. The fact that 13 projects are classified as outliers cannot be explained only by a few large types of KOSBIR portfolios, but they have a structure composed of a number of unique projects that perform specific missions such as start-ups, national standards, and environmental diseases, which can be interpreted as indicators of the health and diversity of portfolios.
In terms of information provision, it is necessary for the Ministry of SMEs and Startups to standardize information between ministries and prepare unified guidelines for the preparation of information and to establish a transparent information provision system that clearly separates and announces the actual scale of support from the perspective of SMEs. Currently, the method of providing information is different due to simple information collection in the process of preparing an integrated announcement for 95 projects by government agencies, making it difficult to make consistent comparisons, and there is a limit to active information collection and participation support for SMEs. Although the same item name of support size is guided as new + continuous budget, some ministries mark only the budget for new tasks, others include the budget for continuing projects, and 100% of the contribution is actually paid to non-profit institutions and research institutes, causing confusion due to inaccurate information. Although the Ministry of SMEs and Startups announced more than 1 trillion won in R&D investment, it sometimes differs from the actual scale of SME support, including labor costs, management costs, and budgets for small and medium-sized companies. The technical terms and policy terms of RFP in detailed tasks make it difficult for even SMEs with the technology to judge the suitability of their technology, limiting practical access without the help of external experts.
In conclusion, the KOSBIR-based small and medium-sized enterprise R&D project in 2025 is in a transitional phase of transitioning to market performance as the commercialization stage occupies the majority, but the systematic support path from PoC to R&D to commercialization has not been identified as the step-by-step linkage support structure such as the US SBIR has not been identified. Additionally, the policy accessibility of SMEs is limited due to inconsistencies in the way information is provided by ministries and the opacity of the actual scale of support for SMEs, which urgently requires improvement of the system by standardizing information between ministries and establishing a step-by-step linkage support system. It is expected that the two goals of continuous growth of SMEs and strengthening national innovation capabilities can be achieved at the same time by shifting to a linked support structure that considers the entire innovation life cycle from the perspective of SMEs in the future and establishing a transparent information provision system.

콘텐츠 큐레이션 : 동일주제 가장 많이 이용된 자료 추천

현재 페이지에서 제공되는 연구보고서에 대하여 평가해 주세요

별점5점만점에 5점 별점5점만점에 4점 별점5점만점에 3점 별점5점만점에 2점 별점5점만점에 1점

의견(0)

연구성과에 따른 의견과 무관한 글, 선정적인 글 및 비방글 등의 게시물은 관리자에 의해 언제든 삭제 조치 될 수 있으며, 주민등록번호 형식 및 연속된 숫자 13자리는 입력할 수 없습니다.

입력 가능 Byte : 4000 Byte 현재 입력 Byte : 0 Byte

자동등록방지 숫자를 순서대로 입력하세요.

NKIS 국가정책연구포털

연구성과

국책연구동향

NKIS 통계

게시판

NKIS소개

연구성과

국책연구동향

NKIS 통계

게시판

NKIS소개

연구성과

연구보고서

중소기업 R&D 사업 유형화를 위한 기초연구 : 클러스터링 기법을 중심으로

A Foundational Study on the Typology of SME R&D Programs : Focusing on Clustering Methods

국문초록

영문초록

목차

콘텐츠 큐레이션 : 동일주제 가장 많이 이용된 자료 추천

연구보고서

정책연구자료

세미나

현재 페이지에서 제공되는 연구보고서에 대하여 평가해 주세요

의견(0)