// using Erik Borra's very useful Gexf class
require_once "Gexf.class.php";
// search query and api key are in:
include "conf.php";
// prepare term
$term = urlencode($term);
$term = preg_replace("/[^a-zA-Z.]/", "", $term);
$folder = getcwd() . "/json_" . $term;
$list = scandir($folder);
// shave off . and ..
array_shift($list);
array_shift($list);
// initialize gexf object
$gexf = new Gexf();
$gexf->setTitle("NY Times Categories");
$gexf->setEdgeType(GEXF_EDGE_UNDIRECTED);
$gexf->setMode(GEXF_MODE_STATIC);
$gexf->setTimeFormat(GEXF_TIMEFORMAT_DATE);
$gexf->setCreator("polsys.net");
// iterate over all JSON files
foreach ($list as $fn) {
$json = file_get_contents($folder . "/" . $fn);
$json = json_decode($json);
// iterate over keyword matrix (half)
for ($i = 0; $i < count($json->keywords); $i++) {
$node1 = new GexfNode(strtolower($json->keywords[$i]->value));
$node1->addNodeAttribute("type", $json->keywords[$i]->name, $type = "string");
$gexf->addNode($node1);
for ($j = $i; $j < count($json->keywords); $j++) {
$node2 = new GexfNode(strtolower($json->keywords[$j]->value));
$node2->addNodeAttribute("type", $json->keywords[$j]->name, $type = "string");
function getGEXFtimeseries($filename, $series)
{
include_once 'common/Gexf.class.php';
$gexf = new Gexf();
$gexf->setTitle("Co-word " . $filename);
$gexf->setEdgeType(GEXF_EDGE_UNDIRECTED);
$gexf->setMode(GEXF_MODE_DYNAMIC);
$gexf->setTimeFormat(GEXF_TIMEFORMAT_DATE);
$gexf->setCreator("tools.digitalmethods.net");
foreach ($series as $time => $cw) {
$w = $cw->getWords();
$cw = $cw->getCowords();
foreach ($cw as $word => $cowords) {
foreach ($cowords as $coword => $coword_frequency) {
$node1 = new GexfNode($word);
if (isset($w[$word])) {
$node1->addNodeAttribute("word_frequency", $w[$word], $type = "int");
}
$gexf->addNode($node1);
//if ($documentsPerWords[$word] > $threshold)
// $node1->setNodeColor(0, 255, 0, 0.75);
$gexf->nodeObjects[$node1->id]->addNodeSpell($time, $time);
$node2 = new GexfNode($coword);
if (isset($w[$coword])) {
$node2->addNodeAttribute("word_frequency", $w[$word], $type = "int");
}
$gexf->addNode($node2);
//if ($documentsPerWords[$coword] > $threshold)
// $node2->setNodeColor(0, 255, 0, 0.75);
$gexf->nodeObjects[$node2->id]->addNodeSpell($time, $time);
$edge_id = $gexf->addEdge($node1, $node2, $coword_frequency);
$gexf->edgeObjects[$edge_id]->addEdgeSpell($time, $time);
}
}
}
$gexf->render();
file_put_contents($filename, $gexf->gexfFile);
echo '<fieldset class="if_parameters">';
echo '<legend>Your co-hashtag time-series File</legend>';
echo '<p><a href="' . filename_to_url($filename) . '">' . $filename . '</a></p>';
echo '</fieldset>';
}
function getCowordsAsGexf($title = "")
{
include_once 'Gexf.class.php';
$gexf = new Gexf();
$gexf->setTitle("Co-word " . $title);
$gexf->setEdgeType(GEXF_EDGE_UNDIRECTED);
$gexf->setMode(GEXF_MODE_DYNAMIC);
$gexf->setTimeFormat(GEXF_TIMEFORMAT_DATE);
$gexf->setCreator("tools.digitalmethods.net");
foreach ($this->cowords as $word => $cowords) {
if (empty($cowords)) {
$node1 = new GexfNode($word);
if (isset($this->words[$word])) {
$node1->addNodeAttribute("word_frequency", $this->words[$word], $type = "int");
}
$this->addNodeExtraNodeAttributes($node1, $word);
$gexf->addNode($node1);
} else {
foreach ($cowords as $coword => $coword_frequency) {
$node1 = new GexfNode($word);
if (isset($this->words[$word])) {
$node1->addNodeAttribute("word_frequency", $this->words[$word], $type = "int");
}
$this->addNodeExtraNodeAttributes($node1, $word);
$gexf->addNode($node1);
$node2 = new GexfNode($coword);
if (isset($this->words[$coword])) {
$node2->addNodeAttribute("word_frequency", $this->words[$coword], $type = "int");
}
$this->addNodeExtraNodeAttributes($node2, $coword);
$gexf->addNode($node2);
$edge_id = $gexf->addEdge($node1, $node2, $coword_frequency);
}
}
}
$gexf->render();
return $gexf->gexfFile;
}
