Top 10 Australian research frontiers discovered with the help of AI
AI gave us a new way to look at research. We used machine learning to classify Australia and the world’s research and then looked for the “hot spot” frontier zones.
Let your imagination roam for a minute. Imagine that the attributes of a research paper – its topics and its findings – can be summarised and represented by a point in a vast space. But we don’t stop there. We summarise more research papers and populate this space with more points, countless numbers of them. Papers that are on similar topics are represented by points that are near each other while papers on dissimilar topics are further apart. The more different a pair of papers are, the greater the distance between them.
We end up with a huge array of points resembling the stars in the universe, and we find that many of them cluster into “galaxies” of papers that are on similar topics.
An example of what we end up with is the picture at the top of this page, a universe of “stars” that actually represents more than 60,000 research papers published in the world in the past year.
Clearly though, this picture was not assembled by hand. It was done with machine learning, which analysed the papers and classified them, using more than 100 different variables. (Obviously we can’t show you on a page the full complexity of what this produced. What you are seeing is a two dimensional slice through a space of more than 100 dimensions, one for each variable.)
The interesting thing is that the machine learning process, without human intervention, naturally grouped the papers into the nine “galaxies’’ in the top picture. (Note that the labels were created by humans, not the AI.) Within those nine groups, but not visible at this scale, are 150 smaller clusters, each representing a “hot spot” – a research frontier where frenetic academic activity is taking place and many papers are being published.
Research analytics firm League of Scholars, which created this pictures, is trying out this AI-driven approach, looking for new insights into the research landscape that are not evident when research papers are “pigeonholed” by traditional methods.
“It’s a way of picking up subtle emerging trends,” says League of Scholars CEO Paul McCarthy. He believes it could be particularly useful in identifying cross-disciplinary research hot spots that fall between the cracks in old-school classification structures.
The approach is in its early days. But, as an experiment, McCarthy and his colleagues focused on Australian research papers to discover which of the 150 global research frontiers Australia is doing best in.
They decided to select the 10 best Australian research frontiers based on the number of citations that Australian papers (in each of the global research frontier clusters) had received from other researchers. Below, we name the ten areas where Australia researchers perform best and we highlight one Australian research paper from each of the ten areas, based the paper’s citation performance.
The ten top Australian research frontiers can also be pictured on the map. We have recoloured the global map above to highlight Australian research on maroon, and located and numbered the top ten frontiers with the numbers corresponding to the list below.
Remember this is an early-stage analysis with limits. Because it spans only a single year, citation counts remain small.
Even so, McCarthy believes it demonstrates a new way to read the literature and identify new frontiers, using machine learning to map papers at scale from the texts themselves. Analysing research with machine learning is still in its early days but we will keep refining the method and tracking where Australia leads, he says.
Tim Dodd is the editor of The Australian’s 2026 Research magazine.
1) Large language models
Example paper:Unifying large language models and knowledge graphs: A roadmap, IEEE Trans. Knowl. Data Eng.First author: Shirui Pan, Griffith University
2) Deep learning and time series
Example paper:Autoencoders and their applications in machine learning: A survey, Artificial Intelligence Review. First author: Kamal Berahmand, Queensland University of Technology
3) Climate change and ecosystems
Example paper:300 years of sclerosponge thermometry shows global warming has exceeded 1.5 °C, Nature Climate Change. First author: Malcolm McCulloch, University of Western Australia
4) Bibliometrics and publishing
Example paper:Open access outputs receive more diverse citations, Scientometrics. First author: Chun-kai Huang, Curtin University
5) Student well-being and resilience
Example paper:Need support and need thwarting: A meta-analysis of autonomy, competence, and relatedness supportive and thwarting behaviors in student populations, Personality and Social Psychology Bulletin. First author: Joshua Howard, Monash University
6) Higher education and skills
Example paper:Work-integrated learning: opportunities and challenges in Australia, High. Educ. Res. Dev. First author: Denise Jackson, Edith Cowan University
7) Global health and social change
Example paper:Unfair knowledge practices in global health: a realist synthesis, Health Policy and Planning. First author: Seye Abimbola, University of Sydney
8) Education systems and pedagogy
Example paper:Where has the joy gone? A qualitative exploration of academic university work during crisis and change, Higher Education Research & Development. First author: Craig Whitsed, Curtin University
9) Religion, culture and identity
Example paper:Religion and growth, J. Econ. Lit. First author: Sascha Becker, Monash University
10) Digital Platforms & Society
Example paper:Risky business: How food-delivery platform riders understand and manage safety at work, Journal of Sociology. First author: Qingyu Wang, University of Melbourne
