My projects are almost always focused on an interesting hypothesis about the scientific enterprise that can only be answered either by analyzing very large datasets (“big data”) or a unique dataset given to me. So my work requires a great deal of big data analysis using state-of-the-art machine learning and artificial intelligence as well as multidisciplinary collaborators.
I always try to investigate questions whose answers I intuitively feel go against the current beliefs held by the scientific community. I try to provide knowledge rather than science policies.
If you want to collaborate on a project, feel free to send me an email to firstname.lastname@example.org or fill out the form bellow at the end of this page.
This projects aims at understanding the predictors of success or failure of a paper during peer review process. Specifically, I’m studying systematic biases introduced by reviewers and authors, including their collaboration network, their geographic location, seniority, and demographics (e.g., gender, race, and age). I’m currently working with a dataset of reviews from PLoS ONE journal, its neuroscience subset, and cross-linking authors and reviews to Neurotree and publication and citation networks.
Comparison of scientific “success” indexes
There is a large number of numerical indexes that attempt to the “success” of a scientists, the most famous of course being the h-index. Almost all indices make a lot of sense but how “good” are they for real decisions performed by committees, funding agencies, and scientific societies? There have been some attempts at quantifying how well these indices are predictive of “real” indicators of success, such promotion or grants. In this project, we will attempt to combine many different indexes by the success they have at predicting such “real” indicators. Our result should be an “uber” index tradeoffs “simplicity” and accuracy.
Predicting scientific success
See my Konrad Kording lab webpage about this project and also the only h-index predictor. You can read the actual article as well.
Other projects unrelated to science of science
Over the years, I’ve worked on many projects that are unrelated to science of science. Here is a list of some projects worth mentioning
Analysis of motor chunking
Chunking is an phenomenon described during the early work in memory studies. Essentially, when we repeatedly perform a task composed of many subtask, memorize a telephone number or a story, the brain implicitly “chunks” those memories into subsets.
One of most interesting aspects of this projects is that we proposed a new algorithm for inferring chunking structure solely based on response times during the task and we made this algorithm openly available online.