IN53B-1846
Software Attribution for Geoscience Applications in the Computational Infrastructure for Geodynamics

Friday, 18 December 2015
Poster Hall (Moscone South)
Lorraine Hwang1, Joseph Dumit1, Alison Fish2, Laura Soito3, Louise H Kellogg4 and MacKenzie Smith4, (1)University of California Davis, Davis, CA, United States, (2)Indiana University, Bloomington, IN, United States, (3)University of New Mexico, Albuquerque, NM, United States, (4)University of California - Davis, Davis, CA, United States
Abstract:
Scientific software is largely developed by individual scientists and represents a significant intellectual contribution to the field. As the scientific culture and funding agencies move towards an expectation that software be open-source, there is a corresponding need for mechanisms to cite software, both to provide credit and recognition to developers, and to aid in discoverability of software and scientific reproducibility.

We assess the geodynamic modeling community’s current citation practices by examining more than 300 predominantly self-reported publications utilizing scientific software in the past 5 years that is available through the Computational Infrastructure for Geodynamics (CIG). Preliminary results indicate that authors cite and attribute software either through citing (in rank order) peer-reviewed scientific publications, a user’s manual, and/or a paper describing the software code. Attributions maybe found directly in the text, in acknowledgements, in figure captions, or in footnotes. What is considered citable varies widely. Citations predominantly lack software version numbers or persistent identifiers to find the software package. Versioning may be implied through reference to a versioned user manual. Authors sometimes report code features used and whether they have modified the code. As an open-source community, CIG requests that researchers contribute their modifications to the repository. However, such modifications may not be contributed back to a repository code branch, decreasing the chances of discoverability and reproducibility. Survey results through CIG’s Software Attribution for Geoscience Applications (SAGA) project suggest that lack of knowledge, tools, and workflows to cite codes are barriers to effectively implement the emerging citation norms. Generated on-demand attributions on software landing pages and a prototype extensible plug-in to automatically generate attributions in codes are the first steps towards reproducibility.

<< Previous Abstract | Next Abstract >>