In a dataset of over 700,000 scientific abstracts from 1881 to 2015, we have found that the readability of scientific texts is steadily declining. Our article was published in the open access journal eLife and was covered in international media (see e.g., Nature News and The Wire ) as well as various blogs (e.g., Neuroskeptic ). All the source code to reproduce the analyses in the published article is available in this Github repository .
I've built a movie and user recommender system for users of the social network movie site Letterboxd . On this site, users who have rated a sufficient amount of movies (currently about 100+) could search for their Letterboxd username and get recommendations for movies and users based on their own movie ratings. Recommendations for a random user can also be given. Semantic similarity of scientific journals Based on the abstracts we have obtained from 123 scientific journals I have used the cortical.io API to
In my PhD I have been using computational models such as Reinforcement Learning and Drift Diffusion models in combination with neuroimaging analyses to make inferences about the decision processes underlying adaptation to errors. Results from these studies are described for example in these two articles with an emphasis on Drift Diffusion modelling and Reinforcement Learning modelling