November 8, 2017
You might remember that a few months back, we had the pleasure introducing our first AI for Everyone Challenge winners. They were each image projects, with common good focuses like creating more equitable facial recognition algorithms and advancing disease research with molecular imagery. Both are up and running, of course, but that’s not the point of our post today. No, today, we’re announcing our second pair of AI for Everyone awardees. So first: a big congratulations to Sarah Sternman and Ani Nenkova! We’re thrilled to have them on board and are really excited to see where their projects end up a year from now. In fact, let’s dig into what they’re doing. We’ll start with Sarah.
Literary style–or, more plainly, writing style–is a complex and often subjective. We know some of the ingredients–word choice, rhythm, tone–and that authors have antecedents and similarities, but research in this field is fraught with challenges. Style is currently studied in many disciplines, through techniques like close reading and computational approaches that look at features like parts of speech or grammar.
Sarah Sternman, a Ph.D. student from UC Berkley, is looking at just this field. She and her team at LiterAIry are looking to use AI to identify the distinct features that influence our perception of style so we can effectively analyze texts based on the characteristics they’ll be discovering in their project.
An AI that can identify and map an author’s style has a host of fascinating uses. It could allow for better, subtler search throughout massive corpuses of text. It could aid predictive text assistants to understand personal style and thus massively improve their suggestions and intuitions. Really, it could facilitate smarter communication via all manner of textual interface.
Our second awardee is Ani Nenkova, an associate professor at the University of Pennsylvania. Her project, like Sarah’s, also centers around text analysis but in a much different field. That field? Medical literature.
To level set here: there are millions of pages of medical research published every year. Staying on top of even a single field or speciality can be difficult, if not impossible. A big part of the reason is that there’s really no effective way to search through vast libraries medical literature. And that’s where Ani’s project comes in.
Ani will be using CrowdFlower to help identify key phrases in clinical trial literature to help alleviate this problem. She hopes to create an AI that can output abstracts, summaries, and reviews of any particular paper based on the identification of those exact key phrases and concepts. An AI that can actually understand this information can help organize and query it, allowing doctors and researchers to locate and source the specific advances they’re most concerned with, effectively making large libraries of medical text more searchable and usable for the medical, as well significantly accelerating medical research in a variety of fields.
We’ll be bringing you updates on both these ventures in the future and we’re thrilled to have both Ani and Sarah aboard. These are truly ambitious projects, ones that require vast amounts of annotated data to run effectively. As such, they really embody what we’re trying to with AI for Everyone: give innovative researchers the tools and support they need to follow their passion and further their research.
If you’re working a project that needs some data and AI help, we’re be accepting new submissions for our next round until early December. Head here to apply. Our finalists are always selected by a group of distinguished judges including members of CrowdFlower’s Scientific Advisory Board: Barney Pell, founder at Moon Express; Pete Warden, Staff Research Engineer at Google; Monica Rogati, independent data science advisor; Adrian Weller, Senior Research Fellow at the University of Cambridge; Jack Clark, Director of Strategy and Communications at OpenAI; Adrien Treuille, VP of Simulation at Zoox Inc.; and Lukas Biewald, founder at CrowdFlower. Selection is based on the innovation of the project, its importance to the advancement of AI and the overall potential impact of the proposed initiative.