Prototyping an article recommendation “bot” powered by Altmetric data

Prototyping an article recommendation “bot” powered by Altmetric data

cute robot
Hello, world! I’m @SociologyBot!
Avatar CC-BY-SA clipartkid / Wikimedia

I’ve recently launched a fun new project, @SociologyBot. It’s a Twitter account that recommends recently discussed research in the field of (you guessed it) sociology.

I’ve been wanting to explore the “altmetrics as a filter” idea for a long time. Being able to find not only disciplinary research but also the conversations surrounding research appeals to me, and I bet that other researchers would like access to that kind of information, too.

So, now I’m experimenting with a prototype “bot”, @SociologyBot. What sets @SociologyBot apart from other research recommendation bots on Twitter are a few things:

  • It’s a social sciences bot (which are surprisingly rare!)
  • It tweets out new and old research alike (not just the “recently published” stuff)
  • It surfaces both research and the conversations surrounding research
  • It’s not actually a bot (yet)!

I’m prototyping @SociologyBot right now, meaning it’s powered using a mix of manual and automated means. (Hence the scare quotes I keep putting around “bot”.) That’s because I want to understand if people actually care about this kind of a bot before I put a lot of time and energy into coding it! I guess you could call @SociologyBot a “minimum viable product”.

Here’s how @SociologyBot currently runs:

  1. I set up a search in Altmetric Explorer to find articles from the top ten sociology journals (as identified by SCImago Journal Rank) that have been mentioned in the last day. I use the journal shortlist as a basis this not because I particularly care for finding only research published in the “top” journals, but because it makes the list of articles much more manageable.
  2. Explorer sends me a daily email summary of said articles.
  3. Based on the shortlist provided in the summary email from Explorer, I schedule new daily tweets using TweetDeck that include both the article with the highest Altmetric Attention Score (AAS) and a link to the Altmetric details page, where discussions of the articles can be found.
  4. Using TweetDeck as automation, @SociologyBot then tweets out one scheduled article daily, at 8 am Mountain time.

Here’s how I plan to build @SociologyBot so that it’s fully automated:

  1. I write a script to query the Altmetric API every 24 hours to find sociology articles that have been mentioned online in the past day.
  2. The script takes the article with the most mentions and checks whether it’s already been tweeted about in the past month, as a safeguard against the same popular articles being constantly recommended.
  3. If it hasn’t, the script then composes a tweet that links to the article and its Altmetric detail page. If it has, the script will then check for the article with the next highest AAS that has not been recently tweeted, and will compose a tweet for that one instead.
  4. The script then posts the article and its Altmetric details page immediately to the @SociologyBot Twitter account.

Whether or not @SociologyBot gets a lot of followers, and whether or not those followers actually click on the Altmetric Details Page links, will determine whether @SociologyBot is a success (and thus whether I should bother coding it to be a proper bot!)

So: if you’re interested in sociology research and want to see this little guy come to life, please give @SociologyBot a follow!

Leave a Reply

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