Towards a More Fine Grained Analysis of Scientific Authorship: Predicting the Number of Authors Using Stylometric Features