With nearly 4,000 biomedical research papers published every day, there is an information overload in the scientific research and knowledge world. Inundated with manuscript submissions, publishers are struggling to determine which articles are the best and in which journals they should be published—amid other challenges that often result in lost revenues and missed papers. Researchers, while trying to keep up with newly published papers and information, are missing those articles that are pertinent to their areas of focus.
And this is where Meta, formerly Sciencescape, comes in. A scientific knowledge network and aggregator rolled into one, Meta was established in 2010 by Toronto-based founders Sam and Amy Molyneux. It currently offers 20
cutting-edge artificial intelligence–enabled data services for researchers, publishers, and life science companies that will make the world’s scientific papers discoverable to the people—and industries—who rely on them.
An AI company for science, Meta partners with 30 major STM book and journal publishers to scan their entire libraries of full-text academic articles. Meta also reads the entirety of PubMed, and continuously crawls the Web to identify all of the people and entities mentioned in the literature. The result is essentially the largest scientific knowledge graph in the world, one so powerful that it knows every relevant paper, person, and entity.
PW spoke with CMO Jeff MacGregor about the data services and the ways publishers and their digital solutions partners can leverage some of these solutions.
Which of your 20 data services are most relevant to the publishing and digital solutions industry?
Jeff MacGregor: Bibliometric Intelligence and Universal Recommendations are two that immediately come to mind.
How does Bibliometric Intelligence work?
It scans a submitted manuscript and identifies 210 unique features, including standard metadata elements like authors, affiliations, and reference lists, as well as deeper concepts such as diagnostic procedures, medical devices, and regulatory activities. It then compares these features to over one million published manuscripts, and, based on this analysis, generates a detailed report that includes projected impact, citation count, and the best journal for its publication.
And what are the benefits of the report to publishers and their partners?
Firstly, it helps to pinpoint high-impact manuscripts the moment they are submitted for publication. Secondly, it allows publishers to prerank manuscripts based on deep predictive profiling and publish more content by intelligently cascading rejected manuscripts to more appropriate sister journals within their portfolio.
On average, what is the accuracy of Bibliometric Intelligence?
Our tests show that Bibliometric Intelligence performs 94% better than thousands of human editors at predicting article-level impact for new manuscripts, prior to publication. It also improved the pre-release identification of “superstar articles”— those that represent the top 1% of high-impact papers — by 118%.
Let’s move on to Universal Recommendations. What prompted this service?
Well, researchers are spending less time on publisher pages. As global research output continues to rise, it is becoming increasingly more difficult for publishers to attract, engage, and retain readers online. And, with new sites being created daily to report the latest scientific discoveries, the competition for researcher engagement is only getting tougher.
So what does it offer in terms of discoverability?
Universal Recommendations provides expert-validated recommendations spanning four core entities of scientific research: articles, topics, researchers, and journals. Researchers can discover more key papers, find more influential authors, and subscribe to more important journals without ever having to leave a publisher’s site. As such, it will increase article readership, usage, and citations as well as increase time on site and page views. It will also increase discoverability of high-value related content.
How would publishers use Meta?
There are two ways: through our full-text publisher partnership program and through our commercial services. The full-text partnership program allows publishers to use Meta’s AI to expand their reach to greater audiences. Meta scans their full-text articles and factors the contents into its Knowledge Graph of science, a continuously evolving scientific intelligence network. Through Meta, researchers are discovering papers that would otherwise remain buried and undiscovered, while publisher manuscripts are reaching even greater audiences.
Who are some of your full-text partners?
The American Medical Association, BioMed Central, Elsevier, Karger, Sage Publishing, Taylor & Francis, and Wolters Kluwer, and most recently the Royal Society, are among our 30 full text–mining partners. Combined, Meta covers 38,000 serial titles that offer around 19 million closed-access full articles—the largest commercial STM text-mining collection on earth.
Your commercial services would be those 20 data solutions?
Yes, and that includes what we just talked about—Bibliometric Intelligence and Universal Recommendations—and others such as Intelligent Manuscript Matching and Horizon Scanning.
With Intelligent Manuscript Matching, for instance, researchers can pre-triage their own manuscripts directly on a publisher’s site and have their manuscript scanned through Bibliometric Intelligence to understand what journal within a publisher’s portfolio is the best match. This helps speed up the path to publication.
Horizon Scanning, on the other hand, exposes emerging technologies and concepts that are still years away from materializing. It allows for reliable early detection of scientific and technical emergences across disciplines and deeply within areas of focus. So publishers can develop new journals and filter for manuscripts that represent the leading edge.
Horizon Scanning would be tremendously helpful to digital solutions providers as they are taking over the peer review, manuscript development, and discoverability parts of the publishing process. Do you agree?
Yes, these data services and the others that we talked about are definitely applicable to publishers’ digital solutions providers, which are functioning as an extension of their editorial and production teams.
Which other Meta services would you recommend for publishers?
There is Article Trajectory, which projects the three-year impact trajectory of a manuscript as well as its postpublication citation performance. This helps to prioritize submissions during triage. Citation Enrichment, on the other hand, identifies key citations that may have been overlooked to enhance the quality of a manuscript or article. Optimal Reviewers is also useful to publishers, assigning ideal combinations of manuscript reviewers based on parameters that include domain, prominence, and collaboration histories.
How does your business model work?
Our primary licensing model is subscription—data as a service, whether that is provided via an API, plug-in, or Analytics Dashboard. Meta Science, our AI-enabled literature-discovery engine, is free for all researchers.
What differentiates Meta Science from engines such as PubMed?
Search engines like PubMed and Google Scholar are great if a researcher already knows what they are looking for. But what about papers that aren’t on their radar? Meta Science is a literature-discovery engine designed for unstructured exploration and real-time streaming. It lets researchers explore science through the entities that make up their research world. This cuts down on the volume of papers within a certain topic and ensures that nothing important is ever missed.
How would digital solutions providers utilize Meta or incorporate it into their services or solutions for the publishers?
One way would be to embed Meta’s data services into their projects and products. For instance, Meta’s Author Profiles can be used to disambiguate authors with similar names and understand the full publishing history, papers, and topics of a given author. This information can be useful for the formation of editorial boards or identifying key opinion leaders, as examples.
We are partnering with publishing-workflow providers to embed bibliometric predictions within the workflow tools for editors, making relevant information available right at the moment of manuscript submission. In another partnership, an editing service provider provides Meta Bibliometric Intelligence reports directly to authors who use their editing services.
So what’s next for Meta?
Biomedicine is just the start for us. We want to expand into physics, chemistry, information science, social science, even patents. Our mission is to organize and deliver the world’s scientific and technical information, and we’ll keep pushing until we realize that goal.