Semester recently ended and I thought it would be nice to try out a little visualisation of article citation metrics that I’ve been thinking about for a little while. This is a little approximate: I manually extracted papers which cite mine (using Google Scholar) and counted up how many of those citations come from each year. Then I time-locked this to years following publication, rather than absolute year.
The first plot (left) shows the number of citations received in the years following publication. Each line corresponds to a different paper. The plot is a little messy, but if you squint you could maybe make an argument that a paper will receive most citations a few years after it’s publication date and then start to trail off into insignificance.
However, it is a frustrating mess, so I then plotted the same data but in the form of cumulative citations in the years following citation (right). I’ve not seen a plot of this form before, but I doubt I am the first person to think of doing it. This cumulative citation plot more clearly shows the different citation performance of each paper, with the confound of publication date removed. Longer lines are from papers published a long time ago, and vice versa. The variability in the slope clearly demonstrates the different impact of the papers – as measured by citations.
One of the more interesting observations is that there might be a correlation between the citation performance and the topic of the paper. I constructed a quick and dirty classification of my papers into 3 categories:
- DPhil related papers on metabolic energy efficiency in the visual system.
- Eye movements. Some purely empirical, some combined empirical + modelling.
- Bayesian, optimal observer modelling. Mostly covert attention, but one eye movement study.
It would be great there was some code to automatically extract the relevant data from Google Scholar, or Web of Science.
Procrastination rating: This all took about 1.5 hours.