Haiku news is an application that generates haikus by referencing its own vocabulary to the wordage of incoming environmental or news-based RSS feeds. The project emerged as an attempt to counter or re-pose the onslaught of news and data we are subject to each day in hopes and offer a pithy and poetic way to take in the happenings of the world.

We created our own very large database of words related to the environment or current news and stored them by syllable count and part of speech. When initialized, the application compares the word content of an incoming RSS feed with words in the database. If there is a match, the program then uses the word in generating a haiku. Pre-formatted templates ensure that the results are generated true to traditional Haiku form. The haikus are then superimposed on top of images found within the feed or from linked websites.

The resulting haikus are random juxtapositions playing with and re-positioning textual meaning. Since they are based out of the vocabulary of the feed, the haikus lend themselves to analyzing the rhetoric and discursive nature of the feed itself.

The haiku’s are frequently ironic, often provocative, sometimes bafflingly true, and sometimes utterly meaningless. When things do fall into place, though, these random poems serve as a chance commentary on the natural world and its relationship to man.

We live in a day and age when the news is updated by the minute. While this is a great resource to us, it can cloud the mind with an overabundance of information. If news were made into poetry, one might have any easier time sifting through the data, generating an emotional response, and see what's presented to us from a new angle.

Generative poetry is a difficult but interesting concept for us as well, so this project allowed us to experiment in generative poetry and discover all the trappings of the practice.

While attempting generating accurate haikus, writing a code that could discover a word's syllable count gave us lots of trouble. So we created our own very large database of pertinent words to which the program will compare the word content of incoming RSS feeds. If there is a match, then the program uses the word in the haiku.

This example shows a small portion of the source text:

“…green low carbon or renewable energy sources conservation and
sustainable use of scarce resources such as water land and air
protection of representative or unique or pristine ecosystems
preservation and expansion of threatened or endangered species or
ecosystems from extinction the establishment of nature and biosphere
reserves under various types of…” — Wikipedia

We used a java concordance program to break this content into individual words along with their frequency count. We then went through the words and systematically broke them down into categories (adjective, noun, etc.) and subcategories of syllable count for each word type. As well, we obtained five sample haikus for building syntactical templates.

We hope that readers will be able to gather meaning from the text, be
it in the form of summary, satire, or other form of word play that
results.

readers, environmentalists, politicians

A user observes a screen that displays a new Haiku every minute or so. A more interactively inclined user might choose to run the process on a different RSS feed via mouse-controlled web interface. Haikus and accompanying images can be appreciated for what they are or--out of one's own curiosity--be traced to the source content on the internet.

It's both a program and a website for people to interact with as described in the user scenario. Other than the computer, we'd like to hang prints of selected haikus and images in and around the floor.

Generating formal poetry, or any human language for that matter, is complicated to coherently generate using computer coded due to things like grammar, syntax, context and relevance to current issues--it's possible, though, but only through very specific rules and templates.