Computational Media – Final Project

Meme Collider

In collaboration with Johnny Lu, we set out with a mutual goal: to create a basic interactive platform for creating new Han-based graphemes or ideographs. —this notion grew from a simple question: how are new words made in Kanji / Hanzi? The answer is generally that there are no actual “new” graphemes being made in some foundry of logic, but rather the inefficient recombination of old ones.

UI sketch for our Kanji / Hanzi builder

We understood it was a daunting task, but not nearly how daunting it would eventually become. Depending on who you ask the complete database comprises anywhere from 40 to 80,000 characters (downloadable here 28MB). With subsets of strokes ranging from 2 to 30+ the exponentiation begins to present itself. Even with much of the dirty work having been done by others like the Wen Lin Institute with CDL, it was clear that this was not a project to be completed in weeks, but more likely years.

Realizing this we decided to revisit a previous project; The Meme Collider. This was conceived of as being an interactive space for textual collisions and clustering. The underlying logic and reference engine would be derived simply from Google search results numbers. While true, these numbers alone may appear overly simplified signifiers for what might benefit more from carefully scraped data, our belief was that when thinking about interaction: Speed wins, and nothing is faster than realtime.

In its present form Meme Collider is essentially a slightly more sophisticated and animated version of Google Fight, however it will soon incorporate clustering functionality that will allow different search terms to cluster together, and de-cluster based on their collective search values, —in effect creating a simple but progressive text-based model for meme interaction.

Applet and Source here

We used a PHP proxy to handle the HTTP requests to Google:

<?php
 
function getResultsForSearch($str) {
    if (!empty($str)) {
        $code = file_get_contents('http://www.google.com/search?q=' . urlencode($str));
        preg_match('/About\s([0-9,]+)\sresults/', $code, $match);
        return str_replace(',', '', $match[1]);
    }
    return false;
}
 
if (!empty($_GET['q'])) {
    print getResultsForSearch($_GET['q']);
}

This PHP script should probably be handled by a separate thread, perhaps with an array of different URL’s to decrease the existing latency when entering search terms rapidly.

We hope to develop Meme Collider further with greater functionality over the the Winter break.