日本語WordNetの辞書(SQLite3)から同義語を取り出す。

WN4P.php

<?php
/**
 * WN4P: To access the Japanese WordNet thesaurus dictionary.
 *
 * @package    Sugina
 * @copyright  Copyright 2012, Toshio HIRAI <[email protected]>
 * @license    http://opensource.org/licenses/BSD-3-Clause The BSD License
 *
 */
/*
 * 辞書の再配布に関しては、以下に示すリンク先の記載内容を遵守すること。
 * About the redistribution of the dictionary,
 * Comply with the restrictions that are described in the following links.
 *
 * http://nlpwww.nict.go.jp/wn-ja/
 * http://nlpwww.nict.go.jp/wn-ja/jpn/downloads.html
 * http://nlpwww.nict.go.jp/wn-ja/license.txt
 */

class WN4P {

	/** Connection to SQLite3 databse.*/
	private $db;

	/**
	 * Constructor.
	 *
	 * @param Pass to `wnjpn.db`(A thesaurus dictionary)
	 * @throws Exception: throws when connection failed.
	 */
	public function __construct($dict) {
		try {
			$this->db = new SQLite3($dict, SQLITE3_OPEN_READONLY);
		} catch (Exception $e) {
			throw new Exception("Unable to connect to the SQLite3 database: ".$dict, 0, $e);
		}
	}

	/**
	 * Close Databse.
	 *
	 */
	public function close() {
		$this->db->close();
	}

	/**
	 * Destructor.
	 *
	 */
	public function __destruct() {
		$this->close();
	}

	/**
	 * Returns the list of synonym of the specified word.
	 *
	 * @param string $word The word.
	 * @param string $lang The language of the word to capture. ('jpn'|'eng'|null:default(both))
	 * @return array The array of synonyms.
	 */
	public function getSynonyms($word, $lang=null) {
		$synonyms = array();
		$words = $this->getWord($word);
		foreach($words as $w) {
			$senses = $this->getSynset($w['wordid']);
			foreach($senses as $sense) {
				$wordIds = $this->getWordId($sense['synset'], $lang);
				foreach($wordIds as $wordId) {
					$synonym = $this->getLemma($wordId['wordid']);
					$synonyms[] = $synonym[0]['lemma'];
				}
			}
		}
		return $synonyms;
	}

	/**
	 * Returns the list of WordId.
	 *
	 * @param string $lemma The lemma (an entry word).
	 * @return array The array of word id.
	 */
	private function getWord($lemma) {
		return $this->getRows("select wordid from word where lemma = :lemma", array(':lemma'=>$lemma));
	}

	/**
	 * Returns the lemma.
	 *
	 * @param string $wordid The word id.
	 * @return array The array of lemma.
	 */
	private function getLemma($wordid) {
		return $this->getRows("select lemma from word where wordid = :wordid", array(':wordid'=>$wordid));
	}

	/**
	 * Returns the synset.
	 *
	 * @param string $wordid The word id.
	 * @return array The array of synset (A set of one or more synonyms).
	 */
	private function getSynset($wordid) {
		return $this->getRows("select synset from sense where wordid = :wordid", array(':wordid'=>$wordid));
	}

	/**
	 * Returns the list of wordid.
	 *
	 * @param string $synset The synset.
	 * @param string $lang The Language.
	 * @return array The array of wordid.
	 */
	private function getWordId($synset, $lang) {
		if (!empty($lang)) {
 			return $this->getRows("select wordid from sense where synset = :synset and lang = :lang",
 								array(':synset'=>$synset, ':lang'=>$lang));
		} else {
			return $this->getRows("select wordid from sense where synset = :synset", array(':synset'=>$synset));
		}
	}

	/**
	 * Executes SQL, and returns the result.
	 *
	 * @param string $sql Skeleton of the SQL.
	 * @param string $params The variable of binding params and values.
	 */
	private function getRows($sql, $params) {
		$stmt = $this->db->prepare($sql);
		foreach($params as $param=>$value) {
			$stmt->bindValue($param, $value);
		}
		$result = $stmt->execute();
		$rows = array();
		while($res = $result->fetchArray(SQLITE3_ASSOC)){
			$rows[] = $res;
		}
		return $rows;
	}
}
?>