See this project live at twaiku.net!
Twaiku is a natural language processing web app that identifies accidental haikus from public tweets. It is able to reliably detect if it is possible to make a 5/7/5 haiku from the text in a tweet by checking the syllable counts of the words from the most common pronunciations in the CMU pronunciation dictionary. Haikus are then saved in a database, tweeted by a bot, and can then be served to a web page.
Twaiku was made using Spring Web MVC, Hibernate ORM, Bootstrap, twitter4j/Twitter's free APIs, and a bit of elbow grease. It was initially built as the final project for a Grand Circus bootcamp by Tristan Mortimer, Marcus Perez, and Matt Davis.