FROM CLAIRE COPE IN THE SCIENCE POLICY CENTRE
Scientists collect data; analyse it; try to make sense of it all; and publish their findings in a scientific journal. Right? Well yes, but it’s not always as simple as that.
The way in which science is conducted has changed dramatically over the past couple of decades. Many scientists now collect terabytes of data, which would be impossible to include in a printed scientific article. And so today, the printed article acts more as a summary of the underlying data.
As a hypothesis-driven process, scientific research does not always deliver the anticipated results. Hundreds of experiments, some that are more informative than others, may be necessary to uncover a new phenomenon in science. Some experiments are conducted to optimise experimental conditions – to uncover the appropriate conditions in which to carry out further experiments. Others show no effect, producing so-called ‘negative data’. Such experiments often fail to make the final cut in the published manuscript and the data remain undisclosed. These issues are part of the Society’s current study on Science as a Public Enterprise.
New mechanisms for sharing data allow experiments to be replicated to verify or refine scientific findings. Data sharing prevents duplication of experimental findings and accelerates the scientific process. Making datasets reusable allows data to be ‘repurposed’: imagine temperature measurements made by the Met Office reused and recycled to study the effect of temperature on wildlife; or linking drugs that affect the activity of genes to chemical databases to accelerate drug discovery. Greater openness with scientific data may also lead to interdisciplinary collaborations and attract investment from a wider range of organisations.
But how can data be more openly accessible? There is a range of software developers who are working on projects designed to help scientists make the most of new technologies – bringing more data to more people. A few will be covered below but it may well be that this becomes a fertile field and one which should be closely watched.
Software developers at Digital Science are part of a growing community developing research tools to facilitate data sharing and day to day laboratory management.
One of their products, FigShare, was created by Mark Hahnel during his PhD at Imperial College. He became aware of the duplication that ensues from a lack of data sharing. He created FigShare out of frustration, allowing scientists to publish all of their data – even those data sets that would not make it into a publication – in an “easily searchable, shareable and citable manner”.
Figures, datasets, images, videos are uploaded onto a researcher’s profile on the FigShare website. The data are permanently stored online under a Creative Commons license which allows others to copy and distribute work provided credit is given to the original data creator. Persistent identifiers assigned to each piece of uploaded data enable the data to be cited even if it hasn’t been published in a scientific journal.
Data can be shared on social media websites such as Facebook and Twitter and other users can leave comments or send a message to the data creator. These features stimulate scientific exchange and can help with data interpretation and even foster scientific collaborations. Data can also be tagged so that other researchers can easily search for data of interest. FigShare also recommends links to similar sets of data much like recommendations on Facebook or iTunes.
The application is very much aimed as a tool for the scientific community. The metadata (‘data about the data’) provided will help an audience who already have a working knowledge of the subject but a non-specialist audience may struggle with its interpretation. There is no peer-review of the uploaded data and therefore no control over their quality. The system works on the basis that a researcher would not want their name associated with questionable data that could be easily verified by other scientists using FigShare. But these concerns may be outweighed by the prospects of more openness in science.
Digital Science is developing other tools aimed at assisting the researcher in their daily activities, with a particular focus on text mining, research tools and science metrics. In particular, Altmetric.com, a new piece of software due to be launched in 2012, directly tackles the need for an alternative measure of scientific impact in the digital age. It aims to capture the attention or real life immediate impact of research (as opposed to its citation in scientific journals) by monitoring social media, reference managers and mainstream media including international newspapers and science magazines. Such alternative measures of scientific impact may help provide the incentives and rewards needed to stimulate greater openness in science.
The importance of open science and data sharing is not a new issue. The 17th century saw a major revolution in the way science was communicated with the invention of the scientific journal. The resulting increase in sharing of knowledge and expertise played a major role in accelerating scientific discoveries. This method of communicating scientific results was well suited to communicating scientific research in the 17th century but may be less so in today’s digital world with multiple media outputs.
Platforms such as those developed by Digital Science, Mendeley or Mekentosj, provide solutions that bring laboratories into the digital age. They provide user-friendly, intuitive applications to keep on top of laboratory management and publication organisation, and invaluable tools to increase scientific exchange, simplify scientists’ daily work and promote greater openness in science. Just like the invention of the scientific journal revolutionised science in the 17th century, these new platforms for communication and data management may place us on the verge of another open science revolution.
The Royal Society would be interested in hearing from individuals and software developers who know of or have developed similar packages aimed at promoting greater openness with these kinds of tools. Please contact us at firstname.lastname@example.org.