out of context quote from Hansard forming a haiku
Hansardku is a public database of quotes from the Hansard which form haiku. Assuming everything has gone to plan, the exact quote making up the haiku will be in the linked source Hansard document.
I wrote this to try and open up the Hansard to a broader audience. Our politicians often say interesting things, talk about things we wouldn't expect, sometimes they even tell jokes. All of this is faithfully copied into the record, but perhaps rarely read. With any luck this site will change that.
This is a personal project of myself, Grahame Bowland. It is intended to be non-partisan, and the software which generates the haiku makes no special allowance for particular Members of Parliament, or for particular parliamentary parties.
Hopefully this project will open up the Hansard and spread awareness that it exists. Of course, the quotes making up the haiku are computer generated and possibly wildly out of context. To understand what a politician was really saying, check the linked source document.
Great! If you happen to be looking for a programmer, I'm looking for work. I'm an experienced software engineer based in Perth with a focus on data visualisation. If you're interested in employing me, contact email@example.com, or prod me on twitter, or over at linkedin.
If you've got some suggestions or feedback, get in touch :-)
The Hansard is made available by the Parliament as XML documents. I have written a computer program which parses these documents, finds the paragraphs spoken by members, and scans for haiku in them.
A haiku is a poem of form 5-7-5; five syllables, seven syllables, five syllables. To find haiku the program must determine the syllable counts of words. This is achieved by checking if the word (or a similar word with a common suffix) is in the GNU Collaborative International Dictionary of English. If not, the program attempts to estimate the number of syllables.
Estimating syllable counts is hit and miss, as the pronounciation rules for English and other latin-script languages are complex. The estimation works by breaking a word up by its constituent sequences of vowels, with some allowance for silent trailing "e"'s and the like.You may find occasional haiku that don't quite add up - apologies!
Some additional tricks are going on. Often politicians quote what other people have said; the program attempts to filter long quotes out, to avoid a haiku attributed to a politician being comprised entirely of a quote of someone else. Sometimes numbers are entered as a string of digits, or as roman numerals. In such a case, the program converts them to words (115 becomes one-hundred-and-fifteen, iv becomes four) for the purpose of calculating the syllable count.