|
bims: Biomed News
The Open Library Society, Inc. a United States 501(3)(c) charity based in Jackson Heights, New York.
2017‒02‒05
http://biomed.news
bims: Biomed News is an expertise sharing system for new papers in PubMed. PubMed indexes about 30000 news papers each week. Bims has recruited subject expert selectors. They maintain reports on new papers that fall within their area of expertise. The report issues are open data. They are distributed to subscribers every week. Thus we have expertise sharing. Experts stay current with the papers and they share their expertise.
Bims is a finished product that makes a radical innovation in an ancient problem domain.
The ancient problem domain is how new academic papers meet their readers. On the one hand, academic papers have few readers because they are highly specialised. On the other hand, these few readers really need to be aware of the new papers in order to stay abreast in their domains. This ancient problem is most acute in the biomedical domain. It has the fastest turnover. But even in the biomedical domain, there is no standard way to deal with this current awareness problem. Every researcher seems to have her own way to stay current. And every researcher swears that that her method is the best. It’s a lovely mess. And we do not pretend we have the solution. We just innovate in this problem space.
The radical innovation is so radical because it has two interrelated aspects to it. One is technical. The other one is economic.
The technical innovation is the use of machine teaching rather than searching. Bims’ experts do not ever enter any search queries. Instead, they teach a machine learning system. They feed it positive and negative example records. Bims is only the second system in the world to use this approach. While there are other system, generally more recent, systems that use artificial intelligence, they all start with search queries.
The economic innovation is the public dissemination of the papers lists selected using the technical innovation. This is completely at odds with usual information retrieval tools. There, users just look up something privately. They could tell the world about what they found. But this is not part and parcel of what the system is all about. Bims disseminates the list papers curated by the experts immediately and permanently. It sends them via email to mailing lists it maintains. And it also creates a web page for each report issue. Thus, our expert users can demonstrate their expertise. The dissemination creates name recognition. Name recognition is an incentive driver for experts’ participation.
In 1998, Thomas Krichel created “NEP: New Economics Papers”. This service uses a small group of experts to disseminate new economics working papers by subject. Working papers are informally published accounts of recent research. With formal publishing delays counting in years rather than months, reading working papers is important for research-active economists. NEP uses the working paper data from the RePEc digital library that Thomas started in the early 1990s. The first issue of NEP had only 24 papers. But by the early 2000s, Thomas realised that the growth of RePEc would mean that eventually, manual sifting through all new papers would become a masochist's pleasure. In 2003, he used left-over funds from the JISC-funded WoPEc project to have Roman D. Shapiro build “ernad”. This is a specialised report composition tool. It is the first-ever tool for information retrieval that uses no search queries. Instead it uses example records. In the 2010s Thomas rewrote ernad completely to make it a tool that can be adapted to any collection of XML records. The obvious candidate to build a similar system on was PubMed. Thomas worked on indexing software for PubMed. But he knows as much about biomedical research as a cow knows about flying. He looked for a project director for years. After meeting Gavin P. McStay he was able to quickly open Bims for Gavin. Gavin was impressed with what he experienced.
The critical resource is the genius of its creator Thomas Krichel. But the most substantive resource is Thomas’ dedicated work to maintain (1) the inputs from PubMed (2) the machine learning system (3) the report issue composition system, (4) the public-facing website and emailing systems. Since there are no subsidies all this has been done with Thomas’ free labour. Since there are no hopes for any substantive revenue, his efforts concentrate on keeping costs down. The result is a no-frills user interface that belies the robust and sophisticated technology behind it.
There are really three aspects to the significance of Bims. One is its pioneering character. We have covered this aspect already.
The second aspect of significance is that exists at all. Just think of the obstacles we face.
On the last item ALPSP now has the chance to demonstrate how perspicacious it really is!
The final aspect is longevity. We know that start-up funding driven initiatives can provide a more fanciful user experience. But their funding dries up as there just is not enough revenue to be made from researchers. Even a general purposed news service like Twitter struggles. We were told a few years ago that PubMed is old-fashioned, that scholarly communication would go to Twitter, with people sharing papers news about papers there. Now it looks like we may outlive Twitter.
The value that that the system offer to its users is immeasurable. We have testimonials.
We hope to help publishers bring papers to readers. And we hope to improve the process of generating new research through more efficient spread of existing research. In this way, we contribute to equity and inclusion in research dissemination. As long as the paper is in PubMed, large and small publishers are treated the same way.
Our usage adds up over time. Data gathered on 2023‒04‒24 shows that our users announced 155809 papers 206722 times. Obviously, there is some duplication between reports.
Recently Thomas Krichel started to produce a web access stats page. These data are derobotified. Otherwise they would be much more homogeneous. The statistics show quite a bit of attention for report issues that are referenced, say on Twitter. This page does not measure usage from the email subscription system. To measure that, we would need a general redirection mechanism. And even if we had that, we can not guarantee that the data is not used in ways we do not know about. It would monitor access to individual papers. We could then assess the popularity of certain papers in certain reports.
We are actively trying all free means we can think of to get more selectors. But we are also trying to gain a new type of selectors. We want to move beyond serving pompous academics who use the system for their intellectual gratification. In 2021, we get the first loved-one of a patient. In 2022 we recruited the first support organisation for a rare disease to act as a selector. This is an important development for us.
We think that bims shines for watching research on rare diseases. There are many diseases that are so rare that authors writing relevant papers may not mention the disease or may not even know it exists. To find a paper that is relevant to a rare disease that does name the disease is a tough challenge. To repeat that challenge every so often to stay current is a challenge beyond the resources that rare disease support organisations have. Thus we hope that we can make a supportive contribution to advancing research and relieve suffering.
Gavin McStay, PhD
School of Pharmacy and Biomolecular Sciences
Liverpool John Moores University
Liverpool, L3 3AF
United Kingdom
g.p.mcstay@ljmu.ac.uk, +44(0)151 9046849