An algorithmic approach to identifying influential research

Originally posted on The Horizons Tracker.

It’s estimated that several million scientific papers are published every year, with the volume increasing year on year. As a result, anyone minded to try and keep on top of things, whether to understand the latest thinking or indeed to guide their own research efforts, has a considerable job on their hands.

The difficulties inherent in keeping up to date with the latest thinking prompted research¹ from Northeastern, which aims to explore how researchers can better prioritize without succumbing to biases or cutting corners.

“There really is a problem about how we develop scholarship,” the researchers explain. “Right now, scientists will often use a search tool like Google Scholar on a topic and start from there, or, if you’re lucky, you’ll get a wonderful instructor and have a great syllabus. But that’s going to be basically the field through that person’s eyes.”

A problematic approach

Often, researchers will rely on Google Scholar, but while the tool is undoubtedly valuable, it tends to focus only on the most popular papers in a given field, with this popularity determined by the citations that paper received. As a result, if there are crucial subsets of that field that are relevant but not as popular, they can be excluded from the search.

The researchers cite the example of aggression, which can often focus on areas such as video games but exclude equally important research into the role of testosterone or social aggression. They used the topic of aggression as the focus of their research and grouped papers on the topic into distinct communities.

They then used citation network analysis to identify 15 research communities on the topic. Their approach bypassed simply looking at raw citations and instead looked at the community of papers that often cite each other. This revealed that the largest communities did include examinations of the role of the media and video games, but also research on the role of testosterone, stress, traits and aggression.

“If you use community detection, then you get this really rich, granular look at the aggression field,” the researchers explain. “You have sort of a bird’s-eye-view of the entire field rather than [it appearing that] the field of aggression is basically media, video games, and violence.”

Improving diversity

The researchers found that as well as improving the diversity of topics uncovered, their community-based approach also increased the percentage of articles with women first authors that were identified as influential.

They have released the code behind their work in the hope that others can both use it themselves and potentially even improve upon it in their own research.

Article source: An Algorithmic Approach To Identifying Influential Research.

Header image source: Iancarelli et al. 2022, CC BY 4.0.

Reference: