On 16 September, we shall be announcing the winners of the 2020 ALPSP Awards for Innovation in Publishing. In this series of posts, we meet our finalists and learn more about them.
We asked Emma Warren-Jones, co-founder of Scholarcy to tell us about the organization and their submission for this year's Awards.
Scholarcy was conceived as a result of the real-life experience of its founder, Phil Gooch, while he was researching his PhD. He became overwhelmed with the amount of reading and while there were lots of tools available that, given one paper would recommend several others, there was nothing that could help him distil the research in front of him and make it easier to digest. Eight years later, he built Scholarcy - an online summariser and knowledge extraction tool that aims to make everyday research faster and more productive.
Scholarcy reads documents in any format (PDF, Word, XML, TeX, html) and distils them into key highlights, structured summaries, key concepts, pulling out other important information such as study participants and statistical analysis.
Researchers, publishers and university libraries are all actively using Scholarcy’s knowledge extraction and summarisation technology to help manage the growing volume of published articles, preprints and manuscript submissions.
What is the product that you submitted for the Awards?
Scholarcy Library is a SaaS consumer app that helps academics sift the masses of research in their field faster, without missing any significant findings. It converts research documents into a unified, interactive summary flashcard format - which can be saved, shared, annotated on any device and exported to multiple formats. See examples at https://app.scholarcy.com/flashcard-generator.html.
- Highlights: distils a research paper, books chapter or web article into a tweetable headline, five key highlights, and a structured summary, with links to cited sources.
- Key terms: extracts the most relevant keywords from a document, and links these to their definitions to provide more background to a subject. Scholarcy keywords are also used by discovery services to promote content and aid discovery.
- Mobile friendly format: our unified PDF/Word to HTML conversion technology makes the fine detail of a paper or chapter clearer and easier to digest on phones and tablets.
- Connects to RSS feeds to generate summaries of breaking research and news.
Tell us a little about how it works and the team behind it
Scholarcy uses its own custom-built machine-learning models and proprietary knowledge distillation engine to automatically extract, clean and structure text from any document, while classifying each word according to its structural and semantic function. The output is then fed into our summarisation engine which creates multiple levels of synopses for different types of readers. We’ve fine-tuned the engine on a large corpus of research papers, government reports and book chapters.
Phil built Scholarcy’s technology. Before this he led the NLP team at Babylon Health. He has a PhD in clinical NLP and many years’ experience developing AI solutions for the publishing, EdTech, and healthcare sectors.
Emma has 20 years’ commercial and marketing experience in the EdTech, academic publishing and information industries, launching discovery platforms and analytics tools to the global research community. She has also worked as a freelance writer for academic publishers.
In what ways do you think it demonstrates innovation?
Scholarcy works on a wide range of formats and distils articles, book chapters and preprints into a unified summary flashcard layout that can be read on any device. It also serves a diverse audience, including: students, researchers, libraries, journalists and publishers.
Very few tools out there are tackling this problem specifically in the context of academic research and serving the needs of the academic community. There are a lot of lightweight summarisation tools available but they are unable to effectively work with the length, complexity or range of content (PDF, Word, HTML, XML, ePub, plain text) that Scholarcy does, nor can they extract the range of information we do.
3. Addresses a pressing global challenge
Scholarcy aims to help solve a significant and growing global challenge: the public understanding of science. From the beginning, one of Scholarcy’s core goals was to help make scientific literature and primary research more accessible to the lay community amidst growing levels of misinformation. We began addressing this by building extractive summarisation technology that could break long, complex documents down into easy-to-digest key highlights and sections that capture the essence of a piece of research. We have recently developed this technology further to be able to generate original plain-language summaries from scientific research (see https://summarizer.scholarcy.com/).
What are your plans for the future?
Up to this point, Scholarcy’s technology has focused on parsing and distilling individual documents. We’re now building an engine that will synthesise summaries, key facts and findings from entire collections of articles that could help automatically draft a review paper on a given topic, for example.
We’re also continuing to develop our deep summarisation technology that will create research explainers for non-experts.
Flashcard generator: https://app.scholarcy.com/flashcard-generator.html
Lay summary generator: https://summarizer.scholarcy.com/
You can hear from Scholarcy and the other finalists at the ALPSP Awards for Innovation Lightning Session on Tuesday 8 September. Visit the ALPSP website to register and for full details of the ALPSP Virtual Conference and Awards 2020.
The 2020 ALPSP Awards for Innovation are sponsored by PLS.