Project #16
Title: An improved poetry generator using semantic relations.
Leader's Name: Alexander Tingström
Member2 Name: Daniel Kruczek
Member3 Name: Robert Vegh
Member4 Name: Patrik Wallin
Related paper: Discovering semantic relations from the web and organizing them with PATTY
Presentation Day: May 20
Model: ES
Abstract: In previous work the authors have developed an automatic poetry generator which implements template based poetry generation using a combined approach with Hidden Markov and N-gram models. This implementation generated poetry which passed as human in 30\% of the time. In this paper an implementation of the PATTY system to augment the model with semantic relation patterns will be designed and evaluated in respect to the "humanness" of the poems. This is expected to significantly increase the amount of poems capable of passing as human.