Today we face an exponential growth of academic literature. The number of journal articles doubles every 9 years; 3.5 million papers are published in 2017 alone. Publishers make every effort to deliver them to appropriate readers. However, 50% of papers are never read by anyone Much of published research has limited readership (read this excellent commentary to understand why we are rephrasing). Simply, the time researchers can spend on their reading is too limited - an average US faculty reads only 20.66 papers per month, spending 32 minutes each. Young researchers and those whose primary language is not English would take a lot longer time. A solution to understand more contents in less time is very much needed.
Can we make an article summary automatically? The answer is “Yes!” The latest development of AI and GPU brings the accuracy of auto-summarization and the calculation speed up to our expectations. We are bringing capable members to take on this challenge. Paper Digest brings benefits to readers, both field experts and non-experts. At the very least, the quick extraction of key sentences will save the time for even expert readers. With further sophistication, Paper Digest can guide the understanding of non-specialist readers, and even produce outreach materials for authors to promote their recent publications, or publishers and institutions for highlighting their latest contents.
Paper Digest uses an AI to generate an automatic summary of a given research paper. You can simply provide a DOI (digital object identifier), or the url to a PDF file, then Paper Digest will return a bulleted summary of the paper. This works only for open access full-text articles that allow derivative generation (i.e. CC-BY equivalent). In case you receive an error message and no summary is generated, it is most likely either the full text is not available to use or the license does not allow derivative generation. Example of journals that work for Paper Digest include but not limited to: PLOS One, Wiley, and others.
For more details we invite you to read the frequent asked questions page
Paper Digest is still in its beta stage and we welcome your contributions to improve our algorithm. We are currently accepting a limited number of pilot users who can participate in our user test, in which you are expected to test Paper Digest for your own published papers and spend up to 10 minutes for each to give us feedback. Please drop an email if you are interested.
Paper Digest is a brainchild of two researchers, Yasutomo and Cristian, both specialized in bibliometrics at the Tokyo Institute of Technology. As we co-authored some papers, our common challenge was to read many papers especially when expanding into new research areas. Leveraging our specialties, we conducted citation network analysis to effectively reduce our reading pile, but the biggest pain to go through the full-text persisted, since English is not our mother tongue. We often had to turn to textbooks to gain basic domain knowledge before spending more hours on a full-text, only to find it irrelevant to our project. Eventually, we asked ourselves: is there a way for a novice to understand the gist of the paper quickly? As we started working on this project - Yasutomo developing deep learning service and Cristian working on web application - Nobuko joined our project to advise product positioning and business strategy backed up by her 15+ years industry experiences. Yasutomo and Cristian are also leading Jiyu Laboratories a data analytics company.
In October 2018, Paper Digest received the Catalyst Grant by Digital Science. We are very honored to receive this global recognition. The generous support provided by Digital Science will help accelerate our development and expand collaboration opportunities.
See the grant announcement.