AskNature.org is making it easier for people to share and organize biological literature so that it can be used to inspire innovators around the world as they search for resilient and regenerative solutions this century's greatest challenges. You can help by training a new learning algorithm to identify organism names and causal relationships from the abstracts of peer reviewed literature that describe how organisms and other living systems survive and thrive on Earth.
For the past several years our team has been investigating ways to improve the quality and breadth of the biology content available on asknature.org. One of the challenges we currently face is that the process of manually assigning functions and living systems to biological strategy records is time consuming and requires specialized knowledge. So earlier this year we started working with a small group of data scientists and programmers to automate these components of AskNature’s content generation process via several machine learning algorithms. If we’re successful, future content contributors will be able to quickly select from ranked lists of suggestions rather than having to manually cull through options from long lists.
The suggestions this tool makes will only be as good as the associations we manually teach it, and with your help, we hope to generate enough associated data (upwards of 10,000 submissions) to get started.
Participating is easy; our colleagues have created a short online survey in which you’ll read the title and abstract of a biology article and then identify key concepts described therein. Each time you take this survey you’ll be training this algorithm to make associations accurately and consistently.
Each time you take the survey you’ll be shown the title and abstract from a peer reviewed article that may be appropriate to include on AskNature. If it is relevant, you’ll be asked to identify relevant functions and living systems, and copy/paste the phrases that led you to make those associations. If it’s not relevant you should still submit, since we’re also teaching the algorithm to distinguish between text that may be useful vs text that isn’t.
