O'Reilly Ideas blog: Doing good data science

The first in a new series of posts on data ethics at O'Reilly Ideas is an easy-to-read taste of the issues, including how data science relates to corporate culture:

https://www.oreilly.com/ideas/doing-good-data-science

“Data scientists, data engineers, AI and ML developers, and other data professionals need to live ethical values, not just talk about them.”

"FAIR in practice" - Jisc report now available

To provide a better understanding of current practice with regard to the use of FAIR principles in the UK and the potential these have to enhance research operations and output, Jisc assigned work to Robert Allen and David Hartland of Hapsis Innovation Ltd. They worked in close collaboration with esteemed experts in the UK such as Simon Coles, Susanna Sansone and Melissa Terras and investigated practices and views in various research disciplines.

The results have now been published in the report ‘FAIR in practice - Jisc report on the Findable Accessible Interoperable and Reuseable Data Principles’, that can be found using DOI 10.5281/zenodo.1245567

A blog on the report, titled ‘Open science is all very well but how do you make it FAIR in practice?’ by Rachel Bruce and Bas Cordewener, is available here.

Open Science By Design: Realizing a Vision for 21st Century Research released today by the National (USA) Academies of Sciences, Engineering and Medicine.

http://sites.nationalacademies.org/pga/brdi/open_science_enterprise/index.htm 

A funder-imposed data publication requirement seldom inspired data sharing - Plos article

Available: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0199789

Abstract

Growth of the open science movement has drawn significant attention to data sharing and availability across the scientific community. In this study, we tested the ability to recover data collected under a particular funder-imposed requirement of public availability. We assessed overall data recovery success, tested whether characteristics of the data or data creator were indicators of recovery success, and identified hurdles to data recovery. Overall the majority of data were not recovered (26% recovery of 315 data projects), a similar result to journal-driven efforts to recover data. Field of research was the most important indicator of recovery success, but neither home agency sector nor age of data were determinants of recovery. While we did not find a relationship between recovery of data and age of data, age did predict whether we could find contact information for the grantee. The main hurdles to data recovery included those associated with communication with the researcher; loss of contact with the data creator accounted for half (50%) of unrecoverable datasets, and unavailability of contact information accounted for 35% of unrecoverable datasets. Overall, our results suggest that funding agencies and journals face similar challenges to enforcement of data requirements. We advocate that funding agencies could improve the availability of the data they fund by dedicating more resources to enforcing compliance with data requirements, providing data-sharing tools and technical support to awardees, and administering stricter consequences for those who ignore data sharing preconditions.

Available: http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0199789