Description
Philosobees was a commissioned art instillation for AUT University.
It is a silly, nonsensical bee talking about philosophy.
This work was a take on the seriousness of academia, the very institution that was paying for it's development. It was designed as an interactive art installation that sits on a 4M screen in the science and engineering building. Passerbyers could tweet @PhilosoBees and there tweet would be turned into a small be asking a question. PhilosoBee would respond and tweet back at them.
Talking to PhilosoBee could trigger them to change appearance, 12 animals where made that philosobee could swap body parts with. Along with an assortment of silly hats. This all added to the obscurity and satire taking place. 
Information
     •     Genre:  Art Instillation, Twitter Bot
     •     Platform:  PC, Mac
     •     Engine:  Unity
     •     Status:  Delivered to Client in 2019
     •     Role:  Game Designer, Developer, & Artist
Highlights
     •     Meaning Making algorithm
     •     Twitter integration
     •     Procedural replies
     •     Absurd character combinations
Feature Breakdown – Meaning Making
This project has several spectacular features I'd love to sit and talk about, but the one feature that was the hardest challenge to overcome was the meaning making algorithm. 
As a tweet would come in, it could say anything. To make some sense of this and provide a response that tries to incite discussion proved very difficult. For it to work several features needed to work together and talk to each other, including the twitter integration that received the tweet. To the procedural reply that was put together with the tracery plugin. 
At the heart of the meaning making algorithm is a very simple understanding of the english language. Each word of the tweet is first run through a custom profanity checker that makes sure we're not going to repeat anything foul in a public space. This used googles list of profanity so was up to date at the time. Each word is than checked against a table of verbs, nouns, adverbs, etc. The meaning maker than decides what to respond to based on what words it understood. 
One annoying hurdle was parsing plurals which was worked around by first converting plurals to singular nouns. 
The biggest problem to overcome was checking each word against 100,000 lists of the english language. This was achieved using a hash table of most words in the english language. 
Back to Top