While reproducibility has been a requirement in natural sciences for centuries, computational experiments have not followed the same standard. Often, there is insufficient information to reproduce computational results described in publications, and in the recent past, this has led to many retractions. Although scientists are aware of the numerous benefits of reproducibility, the perceived amount of work to make results reproducible is a significant disincentive. Fortunately, much of the information needed to reproduce an experiment can be obtained by systematically capturing its provenance. In this paper, we give an overview of different types of provenance and how they can be used to support reproducibility. We also describe a representative set of provenance tools and approaches that make it easy to create reproducible experiments.
Science is conducted collaboratively, often requiring the sharing of knowledge about computational experiments. When experiments include only datasets, they can be shared using Uniform Resource Identifiers (URIs) or Digital Object Identifiers (DOIs). An experiment, however, seldom includes only datasets, but more often includes software, its past execution, provenance, and associated documentation. The Research Object has recently emerged as a comprehensive and systematic method for aggregation and identification of diverse elements of computational experiments. While a necessary method, mere aggregation is not sufficient for the sharing of computational experiments. Other users must be able to easily recompute on these shared research objects. Computational provenance is often the key to enable such reuse. In this paper, we show how reusable research objects can utilize provenance to correctly repeat a previous reference execution, to construct a subset of a research object for partial reuse, and to reuse existing contents of a research object for modified reuse. We describe two methods to summarize provenance that aid in understanding the contents and past executions of a research object. The first method obtains a process-view by collapsing low-level system information, and the second method obtains a summary graph by grouping related nodes and edges with the goal to obtain a graph view similar to application workflow. Through detailed experiments, we show the efficacy and efficiency of our algorithms.
A heated debate surrounds the significance of reproducibility as an indicator for research quality and reliability, with many commentators linking a "crisis of reproducibility" to the rise of fraudulent, careless and unreliable practices of knowledge production. Through the analysis of discourse and practices across research fields, I point out that reproducibility is not only interpreted in different ways, but also serves a variety of epistemic functions depending on the research at hand. Given such variation, I argue that the uncritical pursuit of reproducibility as an overarching epistemic value is misleading and potentially damaging to scientific advancement. Requirements for reproducibility, however they are interpreted, are one of many available means to securere liable research outcomes. Furthermore, there are cases wherethe focus on enhancing reproducibility turns out not to foster high-quality research. Scientific communities and Open Science advocates should learn from inferential reasoning from irreproducible data, and promoteincentives for all researchers to explicitly and publicly discuss (1) their methodological commitments, (2) the ways in which they learn from mistakes and problems in everyday practice, and (3) the strategies they use to choose which research component of any project needs to be preserved in the long term, and how.
Reproducibility and repeatability dramatically increase the value of scientific experiments, but remain two challenging goals for the experimenters. Similar to the LAMP stack that considerably eased the web developers life, in this paper, we advocate the need of an analogous software stack to help the experimenters making reproducible research. We propose the EnosStack, an open source software stack especially designed for reproducible scientific experiments. EnosStack enables to easily describe experimental workflows meant to be re-used, while abstracting the underlying infrastructure running them. Being able to switch experiments from a local to a real testbed deployment greatly lower code development and validation time. We describe the abstractions that have driven its design, before presenting a real experiment we deployed on Grid'5000 to illustrate its usefulness. We also provide all the experiment code, data and results to the community.
In this case study, the authors present one library’s work to help increase awareness of reproducibility and to build capacity for our institution to improve reproducibility of ongoing and future research.
In recent years, evidence has emerged from disciplines ranging from biology to economics that many scientific studies are not reproducible. This evidence has led to declarations in both the scientific and lay press that science is experiencing a “reproducibility crisis” and that this crisis has significant impacts on both science and society, including misdirected effort, funding, and policy implemented on the basis of irreproducible research. In many cases, academic libraries are the natural organizations to lead efforts to implement recommendations from journals, funders, and societies to improve research reproducibility. In this editorial, we introduce the reproducibility crisis, define reproducibility and replicability, and then discusses how academic libraries can lead institutional support for reproducible research.