A methodology of partner selection for sustainable industry-university cooperation based on LDA topic model

Jiho Kang, Junseok Lee, Dong Sik Jang, Sangsung Park

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

In today's knowledge-based society, industry-university cooperation (IUC) is recognized as an effective tool for technological innovation. Many studies have shown that selecting the right partner is essential to the success of the IUC. Although there have been a lot of studies on the criteria for selecting a suitable partner for IUC or strategic alliances, there has been a problem of making decisions depending on the qualitative judgment of experts or staff. While related works using patent analysis enabled the quantitative analysis and comparison of potential research partners, they overlooked the fact that there are several sub-technologies in one specific technology domain and that the applicant's research concentration and competency are not the same for every sub-technology. This study suggests a systematic methodology that combines the Latent Dirichlet Allocation (LDA) topic model and the clustering algorithm in order to classify the sub-technology categories of a particular technology domain, and identifies the best college partners in each category. In addition, a similar-patent density (SPD) index was proposed and utilized for an objective comparison of potential university partners. In order to investigate the practical applicability of the proposed methodology, we conducted experiments using real patent data on the electric vehicle domain obtained from the Korean Intellectual Property Office. As a result, we identified 10 research and development sectors wherein Hyundai Motor Company (HMC) focuses using LDA and clustering. The universities with the highest values of SPD for each sector were chosen to be the most suitable partners of HMC for collaborative research.

Original languageEnglish
Article number3478
JournalSustainability (Switzerland)
Volume11
Issue number12
DOIs
Publication statusPublished - 2019 Jan 1

Fingerprint

patent
methodology
industry
university
Industry
concentration research
electric vehicle
Intellectual property
intellectual property
technical innovation
Electric vehicles
applicant
Clustering algorithms
research and development
quantitative analysis
innovation
Innovation
Decision making
decision making
allocation

Keywords

  • Collaborative research planning
  • Industry-university cooperation
  • Latent Dirichlet Allocation
  • Patent analysis
  • Topic model

ASJC Scopus subject areas

  • Geography, Planning and Development
  • Renewable Energy, Sustainability and the Environment
  • Management, Monitoring, Policy and Law

Cite this

A methodology of partner selection for sustainable industry-university cooperation based on LDA topic model. / Kang, Jiho; Lee, Junseok; Jang, Dong Sik; Park, Sangsung.

In: Sustainability (Switzerland), Vol. 11, No. 12, 3478, 01.01.2019.

Research output: Contribution to journalArticle

@article{66fa40c458dd40b39a0b09cfe6889078,
title = "A methodology of partner selection for sustainable industry-university cooperation based on LDA topic model",
abstract = "In today's knowledge-based society, industry-university cooperation (IUC) is recognized as an effective tool for technological innovation. Many studies have shown that selecting the right partner is essential to the success of the IUC. Although there have been a lot of studies on the criteria for selecting a suitable partner for IUC or strategic alliances, there has been a problem of making decisions depending on the qualitative judgment of experts or staff. While related works using patent analysis enabled the quantitative analysis and comparison of potential research partners, they overlooked the fact that there are several sub-technologies in one specific technology domain and that the applicant's research concentration and competency are not the same for every sub-technology. This study suggests a systematic methodology that combines the Latent Dirichlet Allocation (LDA) topic model and the clustering algorithm in order to classify the sub-technology categories of a particular technology domain, and identifies the best college partners in each category. In addition, a similar-patent density (SPD) index was proposed and utilized for an objective comparison of potential university partners. In order to investigate the practical applicability of the proposed methodology, we conducted experiments using real patent data on the electric vehicle domain obtained from the Korean Intellectual Property Office. As a result, we identified 10 research and development sectors wherein Hyundai Motor Company (HMC) focuses using LDA and clustering. The universities with the highest values of SPD for each sector were chosen to be the most suitable partners of HMC for collaborative research.",
keywords = "Collaborative research planning, Industry-university cooperation, Latent Dirichlet Allocation, Patent analysis, Topic model",
author = "Jiho Kang and Junseok Lee and Jang, {Dong Sik} and Sangsung Park",
year = "2019",
month = "1",
day = "1",
doi = "10.3390/su10023478",
language = "English",
volume = "11",
journal = "Sustainability",
issn = "2071-1050",
publisher = "MDPI AG",
number = "12",

}

TY - JOUR

T1 - A methodology of partner selection for sustainable industry-university cooperation based on LDA topic model

AU - Kang, Jiho

AU - Lee, Junseok

AU - Jang, Dong Sik

AU - Park, Sangsung

PY - 2019/1/1

Y1 - 2019/1/1

N2 - In today's knowledge-based society, industry-university cooperation (IUC) is recognized as an effective tool for technological innovation. Many studies have shown that selecting the right partner is essential to the success of the IUC. Although there have been a lot of studies on the criteria for selecting a suitable partner for IUC or strategic alliances, there has been a problem of making decisions depending on the qualitative judgment of experts or staff. While related works using patent analysis enabled the quantitative analysis and comparison of potential research partners, they overlooked the fact that there are several sub-technologies in one specific technology domain and that the applicant's research concentration and competency are not the same for every sub-technology. This study suggests a systematic methodology that combines the Latent Dirichlet Allocation (LDA) topic model and the clustering algorithm in order to classify the sub-technology categories of a particular technology domain, and identifies the best college partners in each category. In addition, a similar-patent density (SPD) index was proposed and utilized for an objective comparison of potential university partners. In order to investigate the practical applicability of the proposed methodology, we conducted experiments using real patent data on the electric vehicle domain obtained from the Korean Intellectual Property Office. As a result, we identified 10 research and development sectors wherein Hyundai Motor Company (HMC) focuses using LDA and clustering. The universities with the highest values of SPD for each sector were chosen to be the most suitable partners of HMC for collaborative research.

AB - In today's knowledge-based society, industry-university cooperation (IUC) is recognized as an effective tool for technological innovation. Many studies have shown that selecting the right partner is essential to the success of the IUC. Although there have been a lot of studies on the criteria for selecting a suitable partner for IUC or strategic alliances, there has been a problem of making decisions depending on the qualitative judgment of experts or staff. While related works using patent analysis enabled the quantitative analysis and comparison of potential research partners, they overlooked the fact that there are several sub-technologies in one specific technology domain and that the applicant's research concentration and competency are not the same for every sub-technology. This study suggests a systematic methodology that combines the Latent Dirichlet Allocation (LDA) topic model and the clustering algorithm in order to classify the sub-technology categories of a particular technology domain, and identifies the best college partners in each category. In addition, a similar-patent density (SPD) index was proposed and utilized for an objective comparison of potential university partners. In order to investigate the practical applicability of the proposed methodology, we conducted experiments using real patent data on the electric vehicle domain obtained from the Korean Intellectual Property Office. As a result, we identified 10 research and development sectors wherein Hyundai Motor Company (HMC) focuses using LDA and clustering. The universities with the highest values of SPD for each sector were chosen to be the most suitable partners of HMC for collaborative research.

KW - Collaborative research planning

KW - Industry-university cooperation

KW - Latent Dirichlet Allocation

KW - Patent analysis

KW - Topic model

UR - http://www.scopus.com/inward/record.url?scp=85069770593&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85069770593&partnerID=8YFLogxK

U2 - 10.3390/su10023478

DO - 10.3390/su10023478

M3 - Article

AN - SCOPUS:85069770593

VL - 11

JO - Sustainability

JF - Sustainability

SN - 2071-1050

IS - 12

M1 - 3478

ER -