/src/php/associated-words/get-associated-words.php

<?php
/* Copyright 2012 Aditi Muralidharan. See the file "LICENSE" for the full license governing this code. */

/*****************************************************************************
get-associated-words.php

Return adjectives, nouns, and verbs with high TF-IDF scores that 
tend to occur within 10 sentences of this word
*****************************************************************************/

include_once '../util.php';
$wordseer_instance = getGetParam('instance');
$path = '../../../instances/'.$wordseer_instance.'/config.php';
include_once $path;

$N = 0;

//Query parameters
$wordseer_instance = getGetParam('instance');
$gov = '';
$dep = '';
$relation = '';
$govtype = 'word';
$deptype = 'word';
$searches = decodeGetJson('search');
if (count($searches) <= 1) {
    $gov = getGetParam('gov');
    $govtype = getGetParam('govtype');
    $dep = getGetParam('dep');
    $deptype = getGetParam('deptype');
    $relation = getGetParam('relation');
}
$collection = getGetParam('collection');
$statistics = getGetParam('statistics');
$phrasess = decodeGetJson('phrases');
$metadata = decodeGetJson('metadata');
$timing = getGetParam('timing');

dispatch();

function dispatch(){
	if (array_key_exists('id', $_GET)) {
		$id = getGetParam('id');
		$ids = array();
		array_push($ids, $id);
	} else if (array_key_exists('word', $_GET)) {
		$cls = getGetParam('class');
		if ($cls == 'word-set') {
			$ids = explode(", ", getWordIDsFromWordSet(getGetParam('word')));
		} else if ($cls == 'word') {
			$ids = getWordIDs(getGetParam('word'));
		} else if ($cls == "phrase") {
			$ids = explode(" ", getGetParam('word'));
		}
	}
	$id_string = join(",", $ids); 
	$words = getAssociatedWords($id_string, $ids, getGetParam('word'));
	echo json_encode($words);
}

function getAssociatedWords($idString, $ids, $word){
	global $STOPWORDS; // util.php
	global $query_id;
	global $timing;
	$context_conditions = "";
	$table_identifier = "";
	$query_id_where = "";
	if ($query_id) {
		$table = 'filtered_sent_ids';
		$query_id_where = '';
		global $cache_results;
		global $query_id;
		global $dont_cache_search_results;
		if (($cache_results || $query_id) && !$dont_cache_search_results) {
		    $table = 'cached_filtered_sent_ids';
		    $query_id_where = " AND query_id = $query_id ";
		}
		$table_identifier =  ", $table ";
		$context_conditions = "AND $table.id = sentence_id
			$query_id_where";
	}
	$sql = "SELECT sentence_id, sentence_number, document_id 
		FROM sentence_xref_word $table_identifier 
		WHERE word_id IN (".$idString.")
		$context_conditions;";

	if ($timing) {
		echo $sql."<br>
		";
	}
	$result = mysql_query($sql) or die("get-associated-words.php L39 <b>A fatal MySQL error occured</b>.
				<br/> Query: " . $sql . "
				<br/> Error: (" . mysql_errno() . ") " . mysql_error());
	$sentence_ids = array();
	$sentence_numbers = array();
	$document_ids = array();
	while($row = mysql_fetch_array($result)){
		array_push($sentence_ids, $row['sentence_id']);
		array_push($sentence_numbers, $row['sentence_number']);
		array_push($document_ids, $row['document_id']);
	}
	$number_of_sentences = count($sentence_ids);
	$sentence_id_string = join(",", $sentence_ids);
	// 	COUNT(word_id)*LOG(".$number_of_sentences."/COUNT(DISTINCT(sentence_id)))
	$sql = "SELECT count(distinct sentence_id) as score, word, word_id, pos
			FROM word, sentence_xref_word
			WHERE word.id = word_id
			AND sentence_id IN ($sentence_id_string)
			AND word_id NOT IN ($idString)
			AND ((pos like 'N%') OR (pos like 'V%') OR (pos like 'J%'))
			GROUP BY word
			ORDER BY score DESC
			LIMIT 500;";
	$result = mysql_query($sql) or die("get-associated-words.php L82 <b>A fatal MySQL error occured</b>.
				<br/> Query: " . $sql . "
				<br/> Error: (" . mysql_errno() . ") " . mysql_error());
	$words = array("Adverbs"=>array(), "Adjectives"=>array(), "Nouns"=>array(), "Verbs"=>array(), ""=>array());
	$word_set_memberships = getWordSetMemberships();
	while($row = mysql_fetch_array($result)){
		if ($row['score'] > 0 && !in_array($row['word'], $STOPWORDS)) {
			$pos = "";
			if(substr($row['pos'], 0, 1) == "V"){
				$pos = "Verbs";
			}
			if(substr($row['pos'], 0, 1) == "J"){
				$pos = "Adjectives";
			}
			if(substr($row['pos'], 0, 1) == "R"){
				$pos = "Adverbs";
			}
			if(substr($row['pos'], 0, 1) == "N"){
				$pos = "Nouns";
			}
			$word = array(
					"id"=>$row['word_id'], 
					"word"=>$row['word'], 
					"score"=>$row['score']);
			if (array_key_exists($row['word_id'], $word_set_memberships)) {
				$word['word_set'] = join(" ", 
					$word_set_memberships[$row['word_id']]);
			}
			array_push(
				$words[$pos], $word);
			}
	}

	// Get synonyms.

	$sql = "SELECT *, sum(score) as total from synsets 
	WHERE word1_id IN ($idString)
	GROUP BY word2
	ORDER BY total DESC;";
	$result = $result = mysql_query($sql) or die("get-associated-words.php l.157 
		<b>A fatal MySQL error occured</b>.
				<br/> Query: " . $sql . "
				<br/> Error: (" . mysql_errno() . ") " . mysql_error());
	$words["Synsets"] = array();
	while ($row = mysql_fetch_assoc($result)) {
		$word = array(
			"id"=>$row['word2_id'],
			"word"=>$row['word2'],
			"score"=>$row['total']
		);
		if (array_key_exists($row['word2_id'], $word_set_memberships)) {
			$word['word_set'] = join(" ", 
				$word_set_memberships[$row['word2_id']]);
		}
		array_push($words["Synsets"], $word);
	}

	// Get frequent phrases containing that word in this context

	// $id_strings = array();
	// foreach ($ids as $id) {
	// 	array_push($id_strings, ".$id.");
	// }
	// $id_strings = join(" ", $id_strings);
	// $sql = "SELECT id from sequence 
	// where match(id) against(\"$id_strings\" IN BOOLEAN MODE) ;";
	// $result = $result = mysql_query($sql) or die("get-associated-words.php l.169 
	// 		<b>A fatal MySQL error occured</b>.
	// 				<br/> Query: " . $sql . "
	// 				<br/> Error: (" . mysql_errno() . ") " . mysql_error());
	// $sequence_ids = array();
	// while ($row = mysql_fetch_assoc($result)) {
	// 	array_push($sequence_ids, "'".$row['id']."'");
	// }
	// $sequence_id_string = join(",", $sequence_ids);
	// $sql = "SELECT sequence as word, sequence_id, count(sequence_id) as score 
	// FROM sequence_xref_sentence, sequence
	// WHERE sequence.id = sequence_id
	// AND length > 1
	// AND lemmatized = 0
	// AND sentence_count > 1
	// AND sequence.id in ($sequence_id_string)
	// AND sentence_id in ($sentence_id_string)
	// GROUP BY sequence_id
	// ORDER BY score DESC
	// LIMIT 150;";
	// $result = $result = mysql_query($sql) or die("get-associated-words.php l.185 
	// 		<b>A fatal MySQL error occured</b>.
	// 				<br/> Query: " . $sql . "
	// 				<br/> Error: (" . mysql_errno() . ") " . mysql_error());
	// $words["Sequences"] = array();
	// while ($row = mysql_fetch_assoc($result)) {
	// 	array_push($words["Sequences"], $row);
	// }
	return $words;
}

?>