Preserving information on mathematical software via web archives

Abstract

Software is an essential ingredient in mathematical research especially in numerical analysis, mathematical modelling, and statistics. However, the traditional publication process has not come up with a satisfyingly neat solution to reference software yet, taking into account its dynamics in order to allow for reproducibility. While software on GitHub might be ”cited” by linking to the repository including the precise commit given by its SHA signature, commercial software can only be referenced by its web presence most of the time. In 2016, we evaluated swMath, a database for information on mathematical software, on how a software’s archived web site reflects its development. We found, although some web sites are already archived and indeed provide a temporal overview of their corresponding software’s development including documentation, there is need for improvement. In 2017, we presented a demo framework to archive and cite software homepages adding a time stamp. Our vision is a service to archive semantically linked web sites of mathematical software on the basis of swMATH. The resulting archive contains all information on the software at a specific time and can be cited in a traditional publication via DOI.