1. 云栖社区>
  2. PHP教程>
  3. 正文

权重计算,稍加修改亦可用于分词,词频统计,全文和spam检测等

作者:用户 来源:互联网 时间:2017-12-01 14:24:38

分词统计权重修改计算检测词频用于亦可稍加

权重计算,稍加修改亦可用于分词,词频统计,全文和spam检测等 - 摘要: 本文讲的是权重计算,稍加修改亦可用于分词,词频统计,全文和spam检测等, 效率非常客观,你要是改成其他用处那效率我就不保证了<?php// +------------------------------------------------------------------------// Name :


效率非常客观,你要是改成其他用处那效率我就不保证了

<?php
// +------------------------------------------------------------------------
// Name : 权重计算
// Description: 稍加修改,亦可用于分词,词频统计,全文检索和垃圾检测
// Date : 2013/12/16 08:51
// Authors : latel <[email protected]>
// +------------------------------------------------------------------------
//
class weight {
protected $aDict = array(array());
protected $aItems = array();
protected $sLastRule;
protected $aMatchs = array();
protected $aShow = array();
private function init() {
//清空记录的匹配表和输出结果
unset($this->aShow);
}
public function newItems($mItems) {
//导入新的项目
$this->aItems = (is_array($mItems))? $mItems: array($mItems);
$this->init();
}
public function newTable(array $aTable) {
//导入新的对照表,并生成字典
foreach($aTable as $iTableKey=>$sTableLine) {
$aTableLine = explode(',', str_replace('|', ',', $sTableLine));
$setter = function($v, $k, $paraMeter) {
$k1 = $paraMeter[0]; $oWeight = $paraMeter[1];
$oWeight->genDict($v, $k1);
};
array_walk($aTableLine, $setter, array($iTableKey, $this));
}
$this->init();
}
public function getShow($sRule = 'max') {
//获取最终的显示结果
if(empty($this->aItems) || empty($this->aDict))
return array();
if (empty($this->aShow) || $sRule != $this->sLastRule)
return $this->genShow($sRule);
return $this->aShow;
}
public function genShow($sRule) {
$aShow = array();
$aMatchs = array();
$getter = function($v, $k, $oWeight) use(&$aShow, &$aMatchs, $sRule) {
$t = array_count_values($oWeight->matchWord($v));
$aMatchs[] = $t;
switch ($sRule) {
case 'max':
$aShow[$k] = array_keys($t, max($t));
break;
}
};
array_walk($this->aItems, $getter, $this);
$this->aShow = $aShow;
$this->aMatchs = $aMatchs;
return $aShow;
}
private function genDict($mWord, $iKey = '') {
$iInsertPonit = count($this->aDict);
$iCur = 0; //当前节点号
foreach (str_split($mWord) as $iChar) {
if (isset($this->aDict[$iCur][$iChar])) {
$iCur = $this->aDict[$iCur][$iChar];
continue;
}
$this->aDict[$iInsertPonit] = array();
$this->aDict[$iCur][$iChar] = $iInsertPonit;
$iCur = $iInsertPonit;
$iInsertPonit++;
}
$this->aDict[$iCur]['acc'][] = $iKey;
}
function matchWord($sLine) {
$iCur = $iOffset = $iPosition = 0;
$sLine .="";
$iLen = strlen($sLine);
$aReturn = array();
while($iOffset < $iLen) {
$sChar = $sLine{$iOffset};
if(isset($this->aDict[$iCur][$sChar])) {
$iCur = $this->aDict[$iCur][$sChar];
if(isset($this->aDict[$iCur]['acc'])) {
$aReturn = array_merge($aReturn, $this->aDict[$iCur]['acc']);
$iPosition = $iOffset + 1;
$iCur = 0;
}
} else {
$iCur = 0;
$iOffset = $iPosition;
$iPosition = $iOffset + 1;
}
++$iOffset;
}
return $aReturn;
}
}
?>

以上是云栖社区小编为您精心准备的的内容,在云栖社区的博客、问答、公众号、人物、课程等栏目也有的相关内容,欢迎继续使用右上角搜索按钮进行搜索分词 , 统计 , 权重 , 修改 , 计算 , 检测 , 词频 , 用于 , 亦可 稍加 ,以便于您获取更多的相关知识。