Mapping Mr Correcter tweets
As an update to a previous post about Mr Correcter, the Twitter spelling bot here is a post about how I managed to find and geolocate so many offenders. Quick update; Mr Correcter was a script that searched Twitter for 2x common spelling mistakes (beleive and acommodate) and replied back to the tweeter reminding them of the spelling mistake.
- Mr Correcter ran for a total of 9 days
- Mr Correcter tweeted 246 times (236 of these were replies to spelling mistakes)
- I was able to geolocate 105 of those Tweets programmatically
I wanted to geolocate the offenders to generalise which part of the world was the worst at spelling, at least on social media anyway. I previously guessed America due to the number of replies back to Mr Correcter (see previous post for some amusing ones) and as you can see from the map, this turned out to be true.
How I gathered this information
I used three methods in this algorithm to identify where (lat, lng) these Tweets (or accounts) orginated
The obstacles were the Twitter API, getting some of this data is restricted to hourly limits so I had to use a cronjob to schedule the script and aggregate the data after completion.
The script I used is openly available.