Twitter Bots
When opting for a project idea, I was rather stumped, feeling as if those ideas I wanted to pursue were out of my more general range of knowledge with Cheap Bots Done Quick. I knew I wanted to include images in my bots to add a further layer of complexity, with pulling from databases or random generation being goals.
My initial idea for this project was based around the University of Mary Washington logo and brand change that was relevant at the time, with the bot creating and posting a new logo every day “without the half-a-million dollar investment.” However, after experimentation, I came to the conclusion that such project was perhaps too complex for the time available.
@ArbuckleBot
Introduction
Inspired by @RoofSlappingBot, a Twitter bot based on the popular internet meme, I decided on a more simplified project to begin with, @ArbuckleBot. The meme associated with the bot is based off of a 1986 Garfield comic strip combined with a popular internet copypasta into the following format:
"origin" : ["[noun:#second#] #[combination:#first#-#noun#][temp:#third#]sentence# #image#"]
Screenshot taken from Github page for ArbuckleBot
Given the limited format of the meme, I found adding further complexity rather difficult. My original iteration utilized lists of nouns, adjectives, and verbs from Darius Kazemi’s Corpora database, though I found that without proper curation the results were not to my liking. For that reason, I decided to create a curated list of words, totaling to a possible 1042041867048 total permutation count.
Iterations and Improvements
- Added an expanded word-list for adjectives, nouns, and verbs
- Reduced post frequency to every 6 hours (from 3)
My iterations and improvements for this project were completed by adding more curated words to the assigned lists. This created greater diversity.
To gain more information on the code associated with this project and further reflection, visit the associated Github page for ArbuckleBot.
@mariocasted
Introduction
Building off of my knowledge with @ArbuckleBot and given that my iterations to that bot were rather small, I decided to build an entirely separate bot based off of another meme format. This format was based around the recent news that Chris Pratt would be the voice of both Mario in the 2022 Super Mario Bros. film, a surprising casting.
In its current state, the bot takes images from @mariosmirror (an image of a celebrity) and @mariohitlist (an image of a character). These are then meshed together into the following format:
"origin": ["BREAKING: #person# casted as #character# in \n 'Super Mario Bros.' Movie" ],This results in a Tweet such as the following:
Screenshot taken from Github page for mariocasted
Iterations and Improvements
- Added an expanded list of celebrities and characters
- Reduced post frequency to once a day (from twice a day)
I hope to one day update this bot to take images from databases instead of manually having to add new celebrities and characters, making repeat castings essentially nonexistant.
To gain more information on the code associated with this project and further reflection, visit the associated Github page for mariocasted.