Bibliometric Analysis and the Use of Machine Learning for Identifying Latent Topics in ICT Research from Finland
Acharya, Ashim (2025)
2025
All rights reserved. This publication is copyrighted. You may download, display and print it for Your own personal use. Commercial use is prohibited.
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:amk-2025061122324
https://urn.fi/URN:NBN:fi:amk-2025061122324
Tiivistelmä
This research was conducted regarding the complexity of navigating a massive volume of research output from Finland in the ICT sector. It aimed to identify and conduct a bibliometric analysis and uncover latent topics within Finnish ICT research so stakeholders can navigate its landscape meaningfully. Finnish universities are popular and well-ranked internationally in their ICT research output. However, there is little in-depth mapping of Finnish ICT research.
Thus, with the primary objective of navigating the complex Finnish ICT research landscape, this research conducted a bibliometric analysis and an unsupervised machine learning with the latest BERTopic modeling technique. This study collected data from the Scopus database using a keyword search, which had 19500 records from 2005 to 2025 from Finnish ICT research. It carried out a bibliometric analysis at different levels with different units of analysis using VOSviewer software, followed by BERTopic modelling using Llama 2 language model.
Findings showed that in Finland, ICT researchers are collaborating frequently, while machine learning dominates, deep learning is an emerging topic, and sustainability is a key focus area. Regarding highly prominent authors, Mehdi Bennis tops the list, the United States has the most connections with Finland, and IEEE Access is a highly cited journal. Findings also showed the top 10 latent topics, which, despite being scattered, share some connection at a point. Interestingly, these topics are receiving fewer publications in 2025, signaling a shift in the focus of researchers. This study showed the value of combining bibliometric analysis with machine learning-based BERTopic modelling.
Thus, with the primary objective of navigating the complex Finnish ICT research landscape, this research conducted a bibliometric analysis and an unsupervised machine learning with the latest BERTopic modeling technique. This study collected data from the Scopus database using a keyword search, which had 19500 records from 2005 to 2025 from Finnish ICT research. It carried out a bibliometric analysis at different levels with different units of analysis using VOSviewer software, followed by BERTopic modelling using Llama 2 language model.
Findings showed that in Finland, ICT researchers are collaborating frequently, while machine learning dominates, deep learning is an emerging topic, and sustainability is a key focus area. Regarding highly prominent authors, Mehdi Bennis tops the list, the United States has the most connections with Finland, and IEEE Access is a highly cited journal. Findings also showed the top 10 latent topics, which, despite being scattered, share some connection at a point. Interestingly, these topics are receiving fewer publications in 2025, signaling a shift in the focus of researchers. This study showed the value of combining bibliometric analysis with machine learning-based BERTopic modelling.