/**
* Calculate the PageRank for all nodes in $graph.
* @param Graph $graph
* @param bool $keepAllRoundData [optional]. Default: false. If true, the values for every round of calculation is kept. Great for understanding what's going on during the calculation.
* @return PageRankResult
*/
public function calculatePagerank(Graph $graph, $keepAllRoundData = false)
{
$round = 0;
$rounds = [];
$ns = $graph->getNodes();
$edges = $graph->getEdges();
/** @var PageRankNode[] $nodes */
$nodes = [];
//transform to PageRankNodes
foreach ($edges as $edge) {
$prFromNode = $this->getPageRankNode($edge->getFrom(), $nodes);
$prNodeTo = $this->getPageRankNode($edge->getTo(), $nodes);
$prFromNode->addLinkTo($prNodeTo);
$prNodeTo->addLinkFrom($prFromNode);
}
//transform (remaining) disconnected Nodes (that are not part of an Edge) to PageRankNodes
foreach ($ns as $node) {
$this->getPageRankNode($node, $nodes);
}
$this->getLogger()->debug("Got " . count($nodes) . " nodes in total");
do {
$distance = 0;
$currentRound = new PageRankNodeHistory($round);
$newPrCache = [];
foreach ($nodes as $key => $node) {
$node->setOldPr($node->getPr());
$newPr = $node->calculatePageRank($this->dampingFactor, count($nodes));
$curDistance = abs($node->getOldPr() - $newPr);
if ($curDistance > $distance) {
$distance = $curDistance;
}
$entry = new PageRankNodeHistoryEntry($node, $node->getOldPr(), $newPr);
$newPrCache[$key] = $newPr;
$currentRound->addEntry($entry);
}
foreach ($nodes as $key => $node) {
$newPr = $newPrCache[$key];
$node->setPr($newPr);
}
if (!$keepAllRoundData) {
$rounds = [];
}
$rounds[$round] = $currentRound;
$round++;
$this->getLogger()->debug("Calculating round {$round}. Max rounds: {$this->maxRounds}. Last max distance: {$distance}. Required max distance: {$this->maxDistance}");
} while ($round < $this->maxRounds && ($distance == 0 || $distance > $this->maxDistance));
$result = new PageRankResult($graph, $rounds);
return $result;
}
/**
* Imports the CSV file specified by $pathToFile
* @param string $pathToFile
* @return Graph
*/
public function import($pathToFile)
{
$lines = IOUtil::readCsvFile($pathToFile, $this->hasHeader, $this->encoding, $this->delimiter, $this->enclosure, null, null, $this->offset);
$urls = [];
$graph = new Graph();
foreach ($lines as $nr => $line) {
if (!array_key_exists($this->sourceColumn, $line) || !array_key_exists($this->destinationColumn, $line)) {
continue;
}
if (!parse_url($line[$this->sourceColumn]) || !parse_url($line[$this->destinationColumn])) {
continue;
}
$urlFrom = trim($line[$this->sourceColumn]);
if (!array_key_exists($urlFrom, $urls)) {
$urls[$urlFrom] = new Node($urlFrom);
}
$urlTo = trim($line[$this->destinationColumn]);
if (!array_key_exists($urlTo, $urls)) {
$urls[$urlTo] = new Node($urlTo);
}
$edge = new Edge($urls[$urlFrom], $urls[$urlTo]);
$graph->addEdge($edge);
}
return $graph;
}
/**
* See example image at http://de.wikipedia.org/wiki/PageRank
* >> http://de.wikipedia.org/wiki/PageRank#/media/File:PageRank-Beispiel.png
* Difference to the englisch example: No normalization is used - that means PageRankNode "a" has no outlinks at all instead if links to any other node including itself.
*/
public function test_ShouldYieldSameResultsAsTheDeWikiExample()
{
$a = new Node("a");
$b = new Node("b");
$c = new Node("c");
$d = new Node("d");
$e = new Node("e");
$f = new Node("f");
$x1 = new Node("x1");
$x2 = new Node("x2");
$x3 = new Node("x3");
$x4 = new Node("x4");
$x5 = new Node("x5");
$graph = new Graph();
$graph->addEdge(new Edge($b, $c));
$graph->addEdge(new Edge($c, $b));
$graph->addEdge(new Edge($d, $a));
$graph->addEdge(new Edge($d, $b));
$graph->addEdge(new Edge($e, $b));
$graph->addEdge(new Edge($e, $d));
$graph->addEdge(new Edge($e, $f));
$graph->addEdge(new Edge($f, $b));
$graph->addEdge(new Edge($f, $e));
$graph->addEdge(new Edge($x1, $b));
$graph->addEdge(new Edge($x1, $e));
$graph->addEdge(new Edge($x2, $b));
$graph->addEdge(new Edge($x2, $e));
$graph->addEdge(new Edge($x3, $b));
$graph->addEdge(new Edge($x3, $e));
$graph->addEdge(new Edge($x4, $e));
$graph->addEdge(new Edge($x5, $e));
$damping = 0.85;
$maxRounds = 1000;
$maxDistance = 1.0E-6;
$pageRank = new PageRank($damping, $maxRounds, $maxDistance);
$result = $pageRank->calculatePagerank($graph);
$history = $result->getHistory();
$actual = [];
foreach ($history as $key => $historyElement) {
$actual[$key] = $historyElement->toArray();
$round = $actual[$key];
foreach ($round as $node => $nodeValue) {
$actual[$key][$node]["oldPr"] = round($nodeValue["oldPr"], 4);
$actual[$key][$node]["newPr"] = round($nodeValue["newPr"], 4);
}
}
$expected = [92 => ["b" => ["node" => "b", "oldPr" => 0.3242, "newPr" => 0.3242], "c" => ["node" => "c", "oldPr" => 0.2892, "newPr" => 0.2892], "d" => ["node" => "d", "oldPr" => 0.033, "newPr" => 0.033], "a" => ["node" => "a", "oldPr" => 0.0276, "newPr" => 0.0276], "e" => ["node" => "e", "oldPr" => 0.0682, "newPr" => 0.0682], "f" => ["node" => "f", "oldPr" => 0.033, "newPr" => 0.033], "x1" => ["node" => "x1", "oldPr" => 0.0136, "newPr" => 0.0136], "x2" => ["node" => "x2", "oldPr" => 0.0136, "newPr" => 0.0136], "x3" => ["node" => "x3", "oldPr" => 0.0136, "newPr" => 0.0136], "x4" => ["node" => "x4", "oldPr" => 0.0136, "newPr" => 0.0136], "x5" => ["node" => "x5", "oldPr" => 0.0136, "newPr" => 0.0136]]];
$this->assertEquals($actual, $expected);
}
use paslandau\PageRank\Graph;
use paslandau\PageRank\Node;
require_once __DIR__ . "/bootstrap.php";
// define the nodes
$a = new Node("a");
$b = new Node("b");
$c = new Node("c");
$d = new Node("d");
$e = new Node("e");
$f = new Node("f");
$x1 = new Node("x1");
$x2 = new Node("x2");
$x3 = new Node("x3");
$x4 = new Node("x4");
$x5 = new Node("x5");
$graph = new Graph();
// uncomment to get the results of the german example http://de.wikipedia.org/wiki/PageRank
// leave commented for the results of the english one http://en.wikipedia.org/wiki/PageRank
//$graph->addEdge(new Edge($a,$a));
//$graph->addEdge(new Edge($a,$b));
//$graph->addEdge(new Edge($a,$c));
//$graph->addEdge(new Edge($a,$d));
//$graph->addEdge(new Edge($a,$e));
//$graph->addEdge(new Edge($a,$f));
//$graph->addEdge(new Edge($a,$x1));
//$graph->addEdge(new Edge($a,$x2));
//$graph->addEdge(new Edge($a,$x3));
//$graph->addEdge(new Edge($a,$x4));
//$graph->addEdge(new Edge($a,$x5));
// define the links between the nodes
$graph->addEdge(new Edge($b, $c));
