jump to navigation

“Miraculous crispy pig ear”: an analysis of the Charcutepalooza tweets January 3, 2012

Posted by inspiredbywolfe in Charcutepalooza.

This isn’t a food post – apologies – but it is sort of food related, as it contains an analysis of tweets using the #charcutepalooza hashtag. Never fear! Food posts will resume shortly.

When I’m not making brain-shaped headcheese or silly detective cookies, I work as an archivist. Some of the stereotypes are true – I do work in a basement, and sometimes I wear cardigans. However, archivists and records managers aren’t just interested in old stuff, and serious time and research has gone into looking at methods of preserving electronic information – which increasingly includes information contained on social media sites. The capturing and use of social media records is something I’m interested in, and the charcutepalooza twitter archive provided me with the opportunity to conduct some analysis.

I originally set up the charcutepalooza twapperkeeper archive in February so I could look back on the scheduled twitter chats as they were always held at a time when I couldn’t attend. In December, I was reminded that I’d set this thing up, and sure enough, it had continued to merrily tick along, capturing all tweets using the #charcutepalooza hashtag. Since I had this great data source to use, I thought I’d see if I could do some basic analysis and see if any trends emerged. I closely followed some of the analysis methods used by Dr Axel Bruns as part of the Mapping Online Publics project, after I saw him speak earlier this year.

Capturing the tweets

I originally created the Twapperkeeper archive on 14 February 2011. At that time, Twapperkeeper was one of the main ways I’d heard about to capture tweets on a particular keyword or hashtag. However, in March of 2011, the ability to export Twapperkeeper archives was removed, as this was deemed a violation of Twitter’s terms of service. This removed one of the most useful features of Twapperkeeper, but at least the tweets were still being captured. Then in early December 2011, Twapperkeeper announced it had been purchased by HootSuite, and that existing twapperkeeper archives would not be available after January 6, 2012, unless you moved to the (paid) HootSuite model.

Luckily (and thanks to twitter!), I was directed to the blog of Martin Hawksey, who had written a script to extract twapperkeeper archives into a google spreadsheet. I lost no time in immediately exporting the existing charcutepalooza archive to a google spreadsheet, and then set up a new google spreadsheet to continue capturing tweets using the charcutepalooza hashtag (thanks again to Martin Hawksey). This will continue to capture any tweets using the #charcutepalooza hashtag into the future.

Basic information

  • The following analysis was done on tweets captured from 14 February – 28 December 2011 (so does not include the announcement of the finalists on 29 December and subsequent tweets)
  • The archive contains all tweets tagged with the hashtag #charcutepalooza but not tweets which do not have this hashtag. This means if you have replied to a #charcutepalooza-tagged tweet but have not used the hashtag, it won’t be in this collection
  • It contains manual retweets (ie if you have used “RT”), but not retweets done using the automatic twitter retweet button

Some statistics

  • Total number of tweets: 15,244 – which means an average of 1270 tweets per month and 48 tweets per day
  • Number of words contained in tweets: over 235,000
  • Number of retweets (using “RT”): 5042, meaning about 33% of the tweets were retweets
  • Total number of tweeters: 1524 (including people who only used the hashtag once)
  • Number who used the hashtag more than 10 times: 149
  • Number who used the hashtag more than 50 times: 50 (while I haven’t cross-checked, I would suggest this would correspond with the final 33 participants, those who completed the majority of the challenges, plus leaders/advisors such as Mrs Wheelbarrow, Kim Foster, Michael Ruhlman and Bob Delgrosso.

Trends over time

Not surprisingly, there were many more tweets towards the start of charcutepalooza than in November-December. I suggest this is because many people signed up for charcutepalooza, and were very enthusiastic towards the start, but as people dropped out, and there was less publicity surrounding charcutepalooza month by month, the overall number of tweets reduced.

Tweets over time - click through for a bigger image

While the overall quantity of tweets reduced as the months progressed, there was a definite spike on the 15th of every month – corresponding, of course, with the time that everyone’s posts for that month had to be in. There was a larger dip in posts in July (while it still spiked on the 15th it was less dramatic than other months): it seems that the blending challenge did a few people in.

Top topics

When I first took a grab of the Twapperkeeper archive on December 11, I created this chart of the top words in the charcutepalooza tweets. This was done using wordle then some simple manipulation to get it into an appropriately-themed image. The size of the words relates to how often they were used.

Charcutepalooza keywords - click through for a larger version

I removed several symbols and words, such as “RT”, “@” and URLs. As you can see above, the top words also included people’s twitter names. I used a program called wordstat (you need to pay for it – I used the demo version) to analyse the text in the tweets. Here’s the top words in the tweets, once the word “charcutepalooza” and the names of participants were removed.

Top keywords - click through for a larger image

Not surprisingly, everyone announcing their new posts each month meant that “post” was the top word (I assume this was pushed higher by other people commenting saying “great post” and similar). Both Ruhlman and BobdelGrosso were also mentioned quite a lot; I suggest this is because of the twitter chats conducted each month, and people generally asking them questions. Most of the other words related to charcuterie ingredients or techniques (Wordstat also let me look at the top phrases – things like “corn beef” is probably no surprise, but my overall favourite was “miraculous crispy pig ear”!).

Here’s another look at the top words – the size of the bubbles indicates how frequently the words were used, and the different colours indicate the words were used together, or grouped frequently.

Click through for a larger version

Top tweeters

Finally, I wanted to look at who was tweeting, and the frequency. This is a graph with all tweeters who used the charcutepalooza hashtag more than 50 times.

Top tweeters - click through for a larger version

Not surprisingly, Mrs Wheelbarrow led the tweeting charge, by a fair amount. Some tweeters changed user names during the year; I grouped them together as the twapperkeeeper harvest had no way of knowing these were the same people (apologies if I missed anyone changing their name).

Another way of looking at the network between tweeters is to use the tags explorer, again developed by Martin Hawksey. This is set up to run off the google spreadsheet, so is only from December 16 onward, and does not contain the full charcutepalooza tweet set.

Network of charcutepalooza tweeters - click through to the live version

This is a live network – it will continue to expand with charcutepalooza tweets, and allows you to replay any users’ tweets and connections from December 16 onward (click on the username to be taken to the replay screen).

This map also tracks top contributors, which we already know, and something we don’t already know: top conversationalists. This looks at how many connections participants have with other participants (ie, how often they are replying, retweeting and generally interacting with other charcutepalooza participants), rather than total number of tweets.

Click through for a larger version

Here we have the interesting case of Janis Tester, who, while not being the person with the top number of tweets, has the most number of connections within the charcuteplaooza twitter network. This means she is probably directly replying to charcutepalooza tweets (and replying using the #charcutepalooza hashtag), and generally actively engaging with other charcutepalooza participants by commenting and retweeting their tweets. On a purely anecdotal level, I would say that Janet being the top conversationalist will not come as a surprise to any charcutepalooza participant 🙂

There’s definitely more analysis to be done on the charcutepalooza twitter collection but I’ll leave it there for now. Thanks for putting up with a non-food post!


1. mrswheelbarrow - January 3, 2012

I love this analysis! Fascinating!

2. Rachel Tayse - January 3, 2012

Wow! Thanks for your work putting this together!

3. Mardi@eatlivetravelwrite - January 3, 2012

Oh the stats geek in me is loving this!!! Thanks for putting it together!

4. Nic - January 3, 2012

Fascinating! Thanks for doing this. I feel like in need a badge of honour for being a good Charcutepalooza Tweeter!

5. inspiredbywolfe - January 3, 2012

Thanks everyone! I’m glad other people think this is interesting too – I was a bit concerned I’d be the only one who would find this interesting!

6. mosaica - January 4, 2012

Ah, happy geek bits to keep me going as I pack and pack and pack boxes and boxes and boxes. I did a graduate degree in information science, and this sort of thing just makes me beam. Nice work, Captain K 🙂

7. John O'Brien - January 5, 2012

Wow… really enjoyed this analysis. Thanks for sharing. Always love(d) to see TwapperKeeper in use… 🙂

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: