PDF files are not suitable for communicating computational (geo)scientific results

Basic data for this talk

Information about the event

Abstract

Many geoscientific articles are based on computational methods, e.g. a spatiotemporal analysis. It is common practice to describe the analysis as part of the methodology section in a paper. Nevertheless, it is less common to make these materials publicly accessible. Besides issues such as the time needed to prepare code and data, a key issue is that the “classic” way of publishing a paper (i.e. as a PDF file) is not suitable for describing and publishing computational research. It is challenging (if it is even possible) to describe the computational steps in such a detail that others can understand and reuse the analysis and the data. However, even if the materials, i.e. the source code and data, are attached to the paper, they are largely disconnected from the actual text. Besides the lack of connections between the paper, the source code, and the data, a further issue is that PDFs are static. To solve this issue, interactive geoscientific papers could help to better understand how the results were achieved. For example, an interactive paper could show which source code lines and which data subsets were used to produce a specific result. Besides the possibility to inspect the materials underlying the results in the paper, a further benefit is the creation of interactive results. We found a number of papers in which the authors suggested alternative values for parameters that were used in the analysis but they were not able to show the effects on the results in a static PDF file. This talk describes the concept of a binding which assists authors in creating interactive geoscientific papers by creating links between the paper, the analysis code and the data. Based on these links, the authors can specify controls (e.g. a slider) allowing readers to manipulate analysis parameters within a certain range. Giving readers the possibility to inspect and explore the results might facilitate a reader’s, reviewer’s, or journalist’s understanding of the reported results - a key goal of researchers who publish papers. An essential requirement for this alternative way of publishing research is open reproducible research, i.e. code and data underlying the results in the paper are publicly available and executable. Hence, interactive geoscientific papers not only help authors to communicate their research but also to follow good scientific practice.

Speakers from the University of Münster

Konkol, Markus

Professur für Geoinformatik (Prof. Kray)