Altmetric Blog

Is a 3rd party trying to game the Journal of Medical Internet Research?

Euan Adie, 9th October 2012

… almost certainly not, is the short answer. Makes for a fun conspiracy theory though.

Maybe once a month we see an article temporarily jump to the top of the attention rankings because it has been randomly selected as filler content by a Twitter spambot network. The intention, I guess, is that if you go to a Twitter user’s page and see only a big list of tweets about cheap car insurance you dismiss them as a spammer immediately but if you see a bunch of interesting, legit links mixed in you may start to wonder if there’s a real person behind the account after all.

It’s easy to spot this kind of event algorithmically (all of the accounts are relatively new, follow each other, have never mentioned a scientific paper before) and even easier with the human eye: astrophysicists don’t tend to use photos of themselves duckfacing in bikinis or list ‘cheap payday loans’ in their interests. Spambots do, big time.

What’s interesting about the article that this has happened to most recently is that it was published in the Journal of Medical Internet Research. Within a couple of hours of publication it had been tweeted about extensively, for the most part by fake users.

The JMIR is edited by Gunther Eysenbach – who has been involved in a lot of cool science on the web type stuff over the years, not least WebCite – and the journal has been assiduously tracking tweets about their papers for years. Eysenbach published a really interesting if contentious paper earlier this year called Can Tweets Predict Citations (you can find some article level metrics for it here).

Given that there are 30k or so scholarly articles mentioned each month on Twitter the odds of spammers selecting one of the 60 or so from the JMIR is pretty low: at this point it’s well known in altmetrics circles though. On top of that this NY Times article about link spamming came out last week and was discussed in the context of altmetrics soon after:

… so a part of me wondered: is this actually somebody doing some research into how easy it could be to game alt-metrics systems by buying $25 worth of cheap linkage? It’d make for an interesting report or blog post. I’d read it.

(I think the Twitter sponsored ad underworld is actually really interesting in and of itself. $25 will get you an army of easily detectable bikini’ed spambots but if you’re happy to invest a bit more you can also get Paris Hilton for $4,000 or Marlon Wayans for $1,300.)

To be absolutely clear I’m not suggesting that the authors or anybody at the JMIR are trying to game anything. They aren’t. I’m wondering if a third party with an interest in article level metrics is experimenting with paid links.

Boringly this almost certainly boils down to a spammy coincidence – the source tweet just happened to be in the public timeline at the wrong time. If anybody is experimenting, though, please ‘fess up and write about the experience somewhere.

The truth is out there… etc.

3 Responses to “Is a 3rd party trying to game the Journal of Medical Internet Research?”

Gunther Eysenbach
October 9, 2012 at 12:00 am

Thanks for this post. We continue mine the tweets related to JMIR articles to learn about spam patterns and to refine our algorithms to detect and filter out spammers. We have noticed similar patterns in a few previous articles before. In addition, I noted that there are 57 facebook likes for this article - which is also an unusual pattern. We have observed similar patterns for articles with "weight loss" in the title, but I am unsure what keywords or coincidence triggered the spam attack for this particular paper.

G. Eysenbach
Editor & Publisher
JMIR Publications

[...] Is a 3rd party trying to game the Journal of Medical Internet Research’s #altmetrics? (cc @eysenbach) 2012-10-10RT @Scopus: Our customers who use #EBSCO, #Primo or #Summon will soon [...]

September 25, 2014 at 12:00 am

@anirvan @lphilpots Odd spamming of an article happened in 2012

Leave a Reply

Your email address will not be published. Required fields are marked *