中文分词处理方法

发布时间：2015-05-27 19:27:24作者：知识屋

[php]
<?php

/**
* 中文分词处理方法
*+---------------------------------
* @param stirng $string 要处理的字符串
* @param boolers $sort=false 根据value进行倒序
* @param Numbers $top=0 返回指定数量，默认返回全部
*+---------------------------------
* @return void
*/
function scws($text, $top = 5, $return_array = false, $sep = ',') {
    include('./pscws4/pscws4.php');//去下面给的网址把pscws4下载下来
    $cws = new pscws4('utf-8');
    $cws -> set_charset('utf-8');
    $cws -> set_dict('./pscws4/etc/dict.utf8.xdb');
    $cws -> set_rule('./pscws4/etc/rules.utf8.ini');
    //$cws->set_multi(3);
    $cws -> set_ignore(true);
    //$cws->set_debug(true);
    //$cws->set_duality(true);
    $cws -> send_text($text);
    $ret = $cws -> get_tops($top, 'r,v,p');
    $result = null;
    foreach ($ret as $value) {
        if (false === $return_array) {
            $result .= $sep . $value['word'];
        } else {
            $result[] = $value['word'];
        }
    }
    return false === $return_array ? substr($result, 1) : $result;
}
print_r(scws('青花夔龙纹香炉'));

<?php

/**
* 中文分词处理方法
*+---------------------------------
* @param stirng $string 要处理的字符串
* @param boolers $sort=false 根据value进行倒序
* @param Numbers $top=0 返回指定数量，默认返回全部
*+---------------------------------
* @return void
*/
function scws($text, $top = 5, $return_array = false, $sep = ',') {
    include('./pscws4/pscws4.php');//去下面给的网址把pscws4下载下来
    $cws = new pscws4('utf-8');
    $cws -> set_charset('utf-8');
    $cws -> set_dict('./pscws4/etc/dict.utf8.xdb');
    $cws -> set_rule('./pscws4/etc/rules.utf8.ini');
    //$cws->set_multi(3);
    $cws -> set_ignore(true);
    //$cws->set_debug(true);
    //$cws->set_duality(true);
    $cws -> send_text($text);
    $ret = $cws -> get_tops($top, 'r,v,p');
    $result = null;
    foreach ($ret as $value) {
        if (false === $return_array) {
            $result .= $sep . $value['word'];
        } else {
            $result[] = $value['word'];
        }
    }
    return false === $return_array ? substr($result, 1) : $result;
}
print_r(scws('青花夔龙纹香炉'));