Contrasting Publications in Design and Scientific Research

This paper explores the differences between 'design' and 'science' papers published at eCAADe conferences through use of automatic classification. The latter is conducted using a set of differentiating criteria (e.g. number of figures determines a paper to be either 'design' or 'science') which are calibrated with the help of a manual selection of papers from eCAADe 2015 as ground truth. Results show that we predict 83% of the papers correctly; experiments using data from eCAADe 2014 until eCAADe 2016 furthermore show the stability of our results. However, we are not so much after the development this automatic classification but rather want to characterize the two research cultures of design and science. This is achieved by taking a close look at the differentiating criteria, which can inform tools such as ProceeDings over possible future directions and adaptation needs.