A selective updating tool for researchers

A few months ago, I had the pleasure of interviewing Thomas Krichel, about his role in setting up RePEc, the working paper repository for economics. During the interview, I learned about a publication alert system he created at the same time, and which, I subsequently discovered, is today in use both for economics (where it is entitled NEP, short for New Economics Papers), based on the content in RePEc, and for life sciences (where it is called Biomed News, or BIMs), using content from PubMed. I’ve since spoken to users of the tool, and heard very positive comments. I found it a fascinating, but little-known, pair of services.

For both domains, the way it works is very similar. So for this description, I will just use BIMs, which can be found here. The way it operates is simple, a combination of machine- and human input:

1. A human (described as an “editor”) suggests a new topic to the system, by identifying an article or number of articles they think are relevant to that topic.
2. The system, using machine learning most likely via concept extraction, identifies some similar recent content from PubMed.
3. The human then corrects the selection for relevant/not relevant papers. The system subsequently uses these choices to refine the selection.
4. Those papers that the editor believes are interesting and relevant are then sent to subscribers in an email newsletter.

The entire system runs free of charge and uses only volunteer editors, of whom there are around 95 for economics, and 116 for biomedical topics. For each editor, there is a number of “subscribers”, people who sign up to receive the alerts. For economics, the number of subscribers for each newsletter is in the hundreds; while it is considerably fewer in life science, this may be because the service in this domain is more recent.

What I liked about it:

• It’s free to use, and non-intrusive (no promotion, no cookies)
• It involves the community of researchers
• It is clearly a success.

What I’m less happy about:

• It’s a very blunt tool. All the human editor does is select an article – not state why that article is relevant.
• The value of the system is not very apparent. Although there are many thousands of economics researchers benefitting from the service, there is no public metric that demonstrates its value.
• Since there is no advertising, and no promotion, the service spreads by word of mouth – and that is a very slow and imperfect way to disseminate an idea. Many less valuable tools, as measured by their take up by the researchers, are heavily funded and hence promoted, which may produce short-term impact but be less certain of long-term viability.
• Although I stated that it involves the community, and so can be described as part of Open Science, it is nonetheless entirely one-way traffic. The editor provides indications to subscribers; the subscribers have no way of knowing who the other subscribers are; there is no option to share their identity. The decision to keep all subscribers anonymous is I think very limiting. If you are a researcher in a field of a few hundred people, at most, you would want, I would guess, to know about what others in that area are doing – and you probably know a fair number of them. Any discussion about the papers tends to take place on social media (Mastodon or Bluesky), but this is outside the system and so is not tracked.
• The corpus used is limited to PubMed, which does not contain full text, and AI tools work better with full text than on abstracts and title. It could also include grey literature, grant applications, preprints, and many other types of content.
• It should not be difficult to find a bigger corpus to enable the tool to provide better results.  

I recommend that you look at either or both of the tools, and perhaps contact an editor or two of a subject area to get their views. The interface is underwhelming, but the whole thing is run with no ongoing funding. The service delivers and has earned the admiration of many researchers. I suspect most of the editors will agree with  Gavin McStay, one of the administrators, who stated: “I was looking for a  way to keep up with the literature in my subject, and BIMs solves this problem”.