This paper talks about (and reports) experimental data

Automatic discrimination of useless papers via machine learning of abstracts