Haiku Generation

From dandi08

Haiku (Ameriku) Generator

Probabilistic/ Poetry generation/ Styled after art/

Todd Davies davies.t.o@gmail.com Clay Showalter clshowalt@gmail.com Daniel Kervran dkervran@gmail.com Michael Odland mahodara@mac.com

Conceptual Issue Addressed by the Project: Programing an interactive approach to poetry as transformation of existing text.

Relationship to DandI programs: We will utilize and generate python code to aid in the decomposition and reconstruction of text with artistic aims. We will apply our understanding of phonetics to syllable parsing, utilize the WordNet for semantic natural language processing, expand and apply our knowledge of natural expressions, and combine these in order to analyze and extract from multiple corpora.

The Project Itself: Our project will deliver a means of transforming text into the form of sensical haikus. To achieve this end our activities will include syllable parsing, including phonology; natural expressions for haiku template generation; corpus uploading and analysis; and others as needed.

Our strategy is: First recognize the meter of syllables in order to count them, to develop a haiku (5,7,5) format using syllable count, utilizing NLTK's text generate function to produce text within our parameters, moving on to accessing custom corpora via NLTK for the above process, (Room for expanding the project in accordance with resources.)

Learning goals: This project will concretize lessons in phonology, NLTK and python through real world application. Additional learning is needed to upload custom texts, develop natural expressions and create text of satisfactory meaning.