misc/cookbook/groupkeypages.php

<?php

# return the "key pages" of a group
# todo
# 	http://phpir.com/pagerank-in-php
# 	http://faculty.ucr.edu/~hanneman/nettext/C10_Centrality.html#Betweenness
#	check if isset() really is necessary, seems it was avoided in Coeditions

function GroupKeyPages($groupname,$selectionmechanism=""){
	$pages = ListPages("/$groupname\./e");

	//note that this could be replaced by proper ListPages call
	$pages = CleanPageList($pages,$groupname);

	$b = GenerateNetworkFromList($pages);
	//var_dump($b);

	switch ($selectionmechanism) {
		case "NetworkLeverageCentrality":
			$keypages = $selectionmechanism($b,0.05); # value threshold
			break;
		case "NetworkOutgoingIncoming":
			$keypages = $selectionmechanism($b,4,5); # outgoing and incoming degree threshold
			break;
		default:
			$keypages = NetworkDegree($b,8); # degree threshold
	}
	return array_unique($keypages);
}


function GenerateNetworkFromList($pagelist){
	foreach ($pagelist as $page) {
		$content = ReadPage($page,$since=0);
		$links = $content["targets"];
		$links_array = explode(",",$links);
		foreach ($links_array as $link) {
			if (($linkindex = array_search($link,$pagelist)) !== false ){
					if (isset($b["$page"]["outgoing"]))
						$b["$page"]["outgoing"]++;
					else
						{ $b["$page"]["outgoing"] = 1; $b["$page"]["name"]=$page; }
					if (isset( $b["$link"]["incoming"]))
						$b["$link"]["incoming"]++;
					else
						{ $b["$link"]["incoming"] = 1; $b["$link"]["name"]=$link; }
					if (isset($b["$page"]["degree"]))
						$b["$page"]["degree"]++;
					else
						{ $b["$page"]["degree"] = 1; $b["$page"]["name"]=$page; }
					if (isset($b["$link"]["degree"]))
						$b["$link"]["degree"]++;
					else
						{ $b["$link"]["degree"] = 1; $b["$link"]["name"]=$link; }
					$b["$page"]["neighboors"][] = $link;
					$b["$link"]["neighboors"][] = $page;
				}
		}
	}
	return $b;
}

function CleanPageList($pagelist,$groupname){
	// rather arbitrary, not necessarily a good choice
	unset($pagelist[array_search("$groupname.$groupname",$pagelist)]);
	unset($pagelist[array_search("$groupname.Template",$pagelist)]);
	unset($pagelist[array_search("$groupname.GroupHeader",$pagelist)]);
	unset($pagelist[array_search("$groupname.GroupFooter",$pagelist)]);
	return $pagelist;
}

function NetworkOutgoingIncoming($b,$IncomingThreshold=0,$OutgoingThreshold=0){
	$keypages = array(); //done to avoid returning notthing if the theshold is too high
	foreach ($b as $bs){
		if (isset($bs["incoming"]))
			if ($bs["incoming"] > $IncomingThreshold )
				$keypages[] = $bs["name"]; 
		if (isset($bs["outgoing"])) 
			if ($bs["outgoing"] > $OutgoingThreshold ) 
				 $keypages[] = $bs["name"]; 
	}
	return $keypages;
}
	
function NetworkDegree($b, $DegreeThreshold=0){
	$keypages = array(); //done to avoid returning notthing if the theshold is too high
	foreach ($b as $bs){
		if (isset($bs["degree"]))
			if ($bs["degree"] > $DegreeThreshold )
				$keypages[] = $bs["name"]; 
	}
	return $keypages;
}

function NetworkLeverageCentrality($b,$Threshold=0){
	$keypages = array(); //done to avoid returning notthing if the theshold is too high
	foreach ($b as $bs){
		if (isset($bs["neighboors"])){
			$bs["leveragecentrality"] = 0;
			$sum = 0; $avg=0;
			foreach($bs["neighboors"] as $neighboor){
				$sum += ($bs["degree"] - $b["$neighboor"]["degree"]) / ($bs["degree"] + $b["$neighboor"]["degree"]);
				$avg += $b["$neighboor"]["degree"];
			}
			if ($bs["degree"] > 0){
				$avg = $avg / $bs["degree"];
				if ($avg > 0){
					$bs["leveragecentrality"] = ($sum/$avg) / $bs["degree"];
				}
			}
			if ($bs["leveragecentrality"] > $Threshold )
				$keypages[] = $bs["name"]; 
		}
	}
	return $keypages;
}

function GroupDegreeDistribution($groupname){
	$pages = ListPages("/$groupname\./e");

	//note that this could be replaced by proper ListPages call
	$pages = CleanPageList($pages,$groupname);

	$b = GenerateNetworkFromList($pages);

	$distribution = array_fill(1,10,0);
	// has no effect

	foreach ($b as $node){
		$degree = $node["degree"];
		if (isset($distribution[$degree]))
			$distribution[$degree]++;
		else
			$distribution[$degree]=1;
		$sumdeg += $node["degree"];
		$sumin += $node["incoming"];
		$sumout += $node["outgoing"];
	}
	$avgdeg = $sumdeg / count($b);
	$avgin = $sumin / count($b);
	$avgout = $sumout / count($b);

	/*
	print "avgdeg = $avgdeg; ";
	print "avgin = $avgin; ";
	print "avgout = $avgout; ";
	strangely equal, probably a mistake there
	*/ 
	ksort($distribution);
	
	/* fails to get the proper last key
	end($distribution);
	$last_key = key($distribution);
	print $last_key;
	for ($i=1;i<$last_key;$i++)
		if (!(isset($distribution[$i])))
			$distribution[$i]=0;
	*/
	return array_unique($distribution);
}

?>