Leveraging large language models and knowledge graphs to map AI innovations in Finland
Rytky, Mari (2025)
2025
Johtamisen ja tietotekniikan DI-ohjelma - Master's Programme in Management and Information Technology
Johtamisen ja talouden tiedekunta - Faculty of Management and Business
Hyväksymispäivämäärä
2025-02-06
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202502062033
https://urn.fi/URN:NBN:fi:tuni-202502062033
Tiivistelmä
This thesis explores how AI innovations can be mapped from website content by leveraging large language models and knowledge graphs. The study analyzes the adoption of artificial intelligence, or AI, in Finland from an innovation perspective. Traditional methods, such as surveys and patent analyses, often fail to capture innovation processes' rapidly evolving and multidimensional nature. To address this gap, this study introduces a novel methodology that utilizes large language models and knowledge graphs to analyze publicly disclosed information gathered from Finnish companies' websites through web scraping.
The content gathered from Finnish companies' websites between 2020 and 2024 was modeled into a knowledge graph using large language models. This method enabled the identification of organizations, innovations, and their relationships, with a particular focus on AI-related developments across various sectors in Finland. The results show that AI adoption and innovations are concentrated in technology-driven and service-oriented sectors, while in traditional sectors, such as agriculture and food manufacturing, adoption remains limited. Knowledge graphs reveal that organizations utilizing AI engage more actively in collaboration and significantly contribute to innovation development, establishing themselves as pioneers. Digitalization emerges as a fundamental enabler of AI adoption, underscoring its growing importance within the innovation ecosystem.
This thesis provides new insights into the role of AI within Finland's innovation environment, offering practical implications for companies, policymakers, and researchers. Furthermore, it demonstrates the utility of large language models in analyzing extensive textual datasets to identify innovations, presenting a novel methodological contribution to innovation studies. Despite these contributions, the study acknowledges certain limitations, such as the potential bias in web-scraped data, and suggests avenues for future research to strengthen the validity of the findings.
The content gathered from Finnish companies' websites between 2020 and 2024 was modeled into a knowledge graph using large language models. This method enabled the identification of organizations, innovations, and their relationships, with a particular focus on AI-related developments across various sectors in Finland. The results show that AI adoption and innovations are concentrated in technology-driven and service-oriented sectors, while in traditional sectors, such as agriculture and food manufacturing, adoption remains limited. Knowledge graphs reveal that organizations utilizing AI engage more actively in collaboration and significantly contribute to innovation development, establishing themselves as pioneers. Digitalization emerges as a fundamental enabler of AI adoption, underscoring its growing importance within the innovation ecosystem.
This thesis provides new insights into the role of AI within Finland's innovation environment, offering practical implications for companies, policymakers, and researchers. Furthermore, it demonstrates the utility of large language models in analyzing extensive textual datasets to identify innovations, presenting a novel methodological contribution to innovation studies. Despite these contributions, the study acknowledges certain limitations, such as the potential bias in web-scraped data, and suggests avenues for future research to strengthen the validity of the findings.