This is the R code to the tweetbot that produces 365 daily tweets with fun facts taken from the Cleveland Museum of Art's open data. A project by @HxxxKxxx
Update: I have updated my tweetbot in March 2023! Now, it uses GPT-4 to compose tweets by directly sourcing from the API. Using the elements "title", "description", and "fun_fact", it generates a tweet and sends it to Twitter along with the URL and the first image. This is not represented in the code here.
The open data contain a filed "fun_fact" to some of the objects as well as URLs to images of the objects. 365 have both a fun_fact and an image.
Information about the museum's open access can policy be found here: http://www.clevelandart.org/open-access
Possible errors found in data:
- Some Fun_fact contains “"” , “<em>”, “</em>”, “<U+0085>”
- Some of both Tombstone and Title contain “\n” and “\r”
- Dates BC are a positive number in creation_date_earliest
- Rousseau has a fun_fact on Pollock
With the CSV file imported, this code creates a subset with non-empty fields of "fun_fact", downloads the images, creates flashcards with the image, fun fact, title and date on a background with the images' average color and subsequently a TSV file for upload to autochirp. It makes use of the package Imager.
For scheduled tweets, autochirp from the University of Cologne is used.
It requires a TSV with the fields date, time, tweet content, image attachment, latitude, longitude in UTF8. Line breaks are allowed if the escape sequence is retained.
The resulting twitter account can be found here: https://twitter.com/CArt_fun_facts