";
/***************************
* 현재 검색된 문서들 각각에서 term frequency vector를 가져오고 이를 이용해서
* 각 term들의 문서 내에서의 빈도수, 검색 결과내에서 텀이 등장하는 문서 개수를 계산
* output: docFreq[term_id]: 검색 결과에서 해당 텀을 가지고 있는 문서 개수
****************************/
$f1 = floatMicrotime();
// $pubmedID = '19890743'; // 19627182
$query = "SELECT r.docid, r.fvector FROM `cache_data` c,`raw_data` r WHERE c.docid=r.docid and c.cacheid=" . $cacheId; //. " and c.docid <= $pubmedID";
if($mysql_result = mysqli_query($mysqlLink, $query))
{
$dCummTime = 0;
while($row = mysqli_fetch_assoc($mysql_result))
{
$docId = $row["docid"];
$numOfDocuments++;
$curTerm = strtok($row["fvector"], ' :');
$curFreq = strtok(' :');
// term occurrence, document frequency 계산
// - 존재하지 않는 term이면 추가하고 있으면 curFreq를 더한다.
if(array_key_exists($curTerm, $termOccur))
{
$termOccur[$curTerm] += $curFreq;
$docFreq[$curTerm] += 1;
}
else
{
$termOccur[$curTerm] = $curFreq;
$docFreq[$curTerm] = 1;
}
$dTime = floatMicrotime();
while($curFreq != '')
{
$curTerm = strtok(' :');
$curFreq = strtok(' :');
if(strlen($curTerm) != 0)
{
// term occurrence, document frequency 계산
// - 존재하지 않는 term이면 추가하고 있으면 curFreq를 더한다.
if(array_key_exists($curTerm, $termOccur))
{
$termOccur[$curTerm] += $curFreq;
$docFreq[$curTerm] += 1;
}
else
{
$termOccur[$curTerm] = $curFreq;
$docFreq[$curTerm] = 1;
}
}
}
$dCummTime += (floatMicrotime() - $dTime);
} // end while(mysql)
//echo "dCummTime: $dCummTime ";
$tempDocFreq = $docFreq;
$docFreq = array();
foreach($tempDocFreq as $key => $val)
{
if($key != "null")
{
$docFreq[$key] = $val;
}
}
// 각 텀을 빈도수 순으로 정렬합니다. 오름차순 정렬입니다.
asort($docFreq);
// print_r($docFreq);
$globalDocFreq = array();
mysqli_free_result($mysql_result);
unset($mysql_result);
if(@$_REQUEST["debug"] == 1) echo 'f1_검색결과에서 문서별 단어개수, 단어등장 문서 개수 카운트:'. (floatMicrotime()-$f1).' ';
/***************************
* global idf 값 가져오는 부분
* output: $globalDocFreq[term_id]: 전체 문서 셋에서 해당 텀을 가지고 있는 문서 개수
****************************/
$f2 = floatMicrotime();
/////////////////////////////
$query = "SELECT * FROM idf WHERE fnum in (";
$termIdList = array_keys($docFreq);
for($i=0; $i";
if(isset($_REQUEST["globalIdfRatio"])) $threshold1_globalIdf_ratio = $_REQUEST["globalIdfRatio"];
//echo "GlobalIdfRatio: ". $threshold1_globalIdf_ratio . " ";
if(isset($_REQUEST["localGlobal"])) $local_global_threshold = $_REQUEST["localGlobal"];
//echo "localGlobal: ". $local_global_threshold . " ";
if(isset($_REQUEST["siblingThreshold"])) $siblingThreshold = $_REQUEST["siblingThreshold"];
//echo "Sibling Thresdhold: ". $siblingThreshold . " ";
if(isset($_REQUEST["parentThreshold"])) $parentThreshold = $_REQUEST["parentThreshold"];
//echo "Parent-Child Thredhold: ". $parentThreshold . " ";
// $threshold1_globalIdf = 18000000 * $threshold1_globalIdf_ratio;
// ---끝--- 외부에서 Threadhold를 수정할 수 있게 합니다. 논문 실험용 - 김태훈 2010/07/10
/////////
// echo "ternId\tlocal_idf\tglobal_idf\n";
//////////////////////////////////////////////////////////
// 2010 DTMBIO 논문 때문에 특정 키워드일때 Threshold 값을 다르게 지정하도록 함 ---시작 ---
// 2010/7/14 김태훈
if(@$_REQUEST["debug"] == 1) echo 'f2 전체 데이터베이스에서 해당 텀을 가지고있는 문서 개수 카운트:'. (floatMicrotime()-$f2).' ';
$f3 = floatMicrotime();
if($searchOption->getKeyword() === "mrsazzz"
|| $searchOption->getKeyword() === "mrsa strainzzz"
|| $searchOption->getKeyword() === "mrsa resistancezzz")
{ // 논문 구현때문에 하드 코딩한 부분
foreach($tempDocFreq as $key => $val)
{
$global = $globalDocFreq[$key];
$p_x = $val/$numOfDocuments;
if($global < 18000000 * 0.05) // global idf < # of all documents * 0.05 만 써요
{
if($p_x > 0.05) //sebset idf > # of result documents * 0.05
{
$dfRatio = $val / $global;
if($dfRatio < 0.05)
{
$docFreq[$key] = $val;
$count++;
}
}
}
$termEntropy[$key] = number_format((-1) * $p_x * log($p_x, 10), 4);
}
}
else // 아래가 정상적인 코드
{
foreach($tempDocFreq as $key => $val)
{
if($termTypeMapping[$key] == "abstract")
{
$global = $globalDocFreq[$key] / $numOfGlobalDocuments;
$p_x = $val/$numOfDocuments;
if(@$_REQUEST["debug"] == 1) {
if(true /*$global < $threshold1_globalIdf_ratio && $globalDocFreq[$key] > 100*/)
{
$dfRatio = $p_x / $global;
if($p_x > $threshold1_localIdf_ratio && $dfRatio > $local_global_threshold) //subset idf > # of result documents * 0.05
{
$docFreq[$key] = $val;
$count++;
}
}
} else
{
if($global < $threshold1_globalIdf_ratio && $globalDocFreq[$key] > 100)
{
$dfRatio = $p_x / $global;
if($p_x > $threshold1_localIdf_ratio) //subset idf > # of result documents * 0.05
{
$docFreq[$key] = $val;
$count++;
}
else if($dfRatio > $local_global_threshold) // (local idf/# of local documents) / (global idf/# of global documents) < 9.2
{
$docFreq[$key] = $val;
$count++;
}
}
}
/* 플래시 안에서의 똥그라미 싸이즈 변경을 위한 새로운 공식 */
// $termEntropy[$key] = $p_x/$global;
$termEntropy[$key] = number_format((-1) * $p_x * log($p_x,10), 4);
}
}
/* 똥그라미 사이즈 변경을 위한 parameter */
$addPara = 0.1;
$maxEntropy = max($termEntropy);
$entropyList = $termEntropy;
$termEntropy = array();
foreach($entropyList as $key => $val)
{
$termEntropy[$key] = ($entropyList[$key]/$maxEntropy) * (1 - $addPara) + $addPara;
}
}
unset($tempDocFreq);
unset($globalDocFreq);
// -----끝 ----- 2010 DTMBIO 논문 때문에 특정 키워드일때 Threshold 값을 다르게 지정하도록 함 ------
if($count<=0)
{
$endFlag = 1;
}
//echo "# of result documents : $numOfDocuments\n";
//echo "# of useful terms: $count\n";
//echo "# of terms: ".sizeof($termOccur)."\n";
unset($docId);
unset($curTerm);
unset($curFreq);
} // end if
// echo "";
unset($termOccur);
if(@$_REQUEST["debug"] == 1) echo 'f3 concept 찾기:'. (floatMicrotime()-$f3).' ';
$f4 = floatMicrotime();
if($endFlag != 1)
{
$i = 0;
$j = 0;
$termIdx2Id = array();
//useful termTerm matrix 생성 init = 0
foreach($docFreq as $key1 => $val1)
{
$termIdx2Id[$i++] = $key1;
foreach($docFreq as $key2 => $val2)
{
if(!isset($termTerm[$key1][$key2]))
{
$termTerm[$key1][$key2] = 0;
}
}
}
// 두 번째 db scan. 첫 번째 step에서 추출 된 term list에 속한 term들만 term-term table을 만든다(co-occurrence)
// termterm 행렬도 DOCUMENT frequency 임 (텀프리퀀시 아님)
$query = "SELECT r.docid, r.fvector FROM `cache_data` c,`raw_data` r WHERE c.docid=r.docid and c.cacheid=" . $cacheId ;//. " and r.docid <= $pubmedID";
if($mysql_result = mysqli_query($mysqlLink, $query))
{
while($row = mysqli_fetch_assoc($mysql_result))
{
$docId = $row["docid"];
$numTermsInThisDoc = 0;
// 현재 row(document)의 term list 추출
$curDocTerm = array();
$curTerm = strtok($row["fvector"], ' :');
$curFreq = strtok(' :');
//useful termList에 속한 term만 curDocTerm에 저장
if(array_key_exists($curTerm, $docFreq))
{
$curDocTerm[$numTermsInThisDoc++] = $curTerm;
}
while($curFreq != '')
{
$curTerm = strtok(' :');
$curFreq = strtok(' :');
if(strlen($curTerm) != 0)
{
if(array_key_exists($curTerm, $docFreq))
{
$curDocTerm[$numTermsInThisDoc++] = $curTerm;
}
}
}//end while`````
// useful한 term list에 존재하는 term들간의 co-occurrence 계산
for($i=0; $i<$numTermsInThisDoc-1; $i++)
{
for($j=$i+1; $j<$numTermsInThisDoc; $j++)
{
$termTerm[$curDocTerm[$i]][$curDocTerm[$j]] += 1;
}
}
}//end while
unset($curDocTerm);
}//end if
if(@$_REQUEST["debug"] == 1) echo 'f4 현재 검색결과에서 텀과 텀간의 빈도수 행렬 생성:'. (floatMicrotime()-$f4).' ';
// term간의 relation 찾기
$numTerm = sizeof($docFreq);
$RELATION_SIBLING = "Sibling";
$parentChildRelation = "Parent-Child";
$overTreeRelation = "Over Tree Parent-Child";
/*****************************************
* term-term matrix와 df matrix를 보면서 확률 계산 - sibling/parent relation detect
****************************************/
// 비교를 위한 나이브 메소드 - POST
// ------------------
if(@$_REQUEST["debug"] == 1)
{
$f5_naive = floatMicrotime();
$termRelation = array();
$termRalationSibling = array();
$whoIsMyRoot = array();
$termHavingRelation = array();
$myGraph = array();
for($i=0; $i<$numTerm; $i++)
{
$whoIsMyRoot[$termIdx2Id[$i]] = $termIdx2Id[$i]; // 이거 두개는
$edgeFromRoot[$termIdx2Id[$i]] = 0; // 나중에 저장된 relation으로 tree 구조 visualize 해줄 때 필요할거 같아서
$myGraph[$termIdx2Id[$i]] = $i;
}// 각 term의 graph상 root값을 자기 자신으로 초기화, root로부터의 거리는 0
for($i=0; $i<$numTerm; $i++)
{
for($j=0; $j<$numTerm; $j++)
{
$proiGivenj = $projGiveni = 0;
if($termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]]!=0) // term_i와 term_j가 같이 나오는 document가 있어요
{
$proiGivenj = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$j]]; // P(Term_i | Term_j)
$projGiveni = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$i]]; // P(Term_j | Term_i)
if(($proiGivenj > $siblingThreshold) && ($projGiveni > $siblingThreshold)) // Term_i와 Term_j 는 sibling
{
$termRalationSibling[$termIdx2Id[$j]][$termIdx2Id[$i]] = $RELATION_SIBLING;
$tempGraph = $myGraph[$termIdx2Id[$i]];
for($r=0; $r<$numTerm; $r++)
{
if($myGraph[$termIdx2Id[$r]]==$tempGraph)
{
$myGraph[$termIdx2Id[$r]]=$myGraph[$termIdx2Id[$j]];
}
}
$termHavingRelation[$termIdx2Id[$i]] = $whoIsMyRoot[$termIdx2Id[$i]];
$termHavingRelation[$termIdx2Id[$j]] = $whoIsMyRoot[$termIdx2Id[$j]];
}
if($projGiveni > $parentThreshold && $proiGivenj < $parentThreshold) // Term_j는 Term_i의 parent
{
$termRelation[$termIdx2Id[$j]][$termIdx2Id[$i]] = $parentChildRelation;
$tempRoot = $whoIsMyRoot[$termIdx2Id[$i]];
$tempEdge = $edgeFromRoot[$termIdx2Id[$i]];
$tempGraph = $myGraph[$termIdx2Id[$i]];
// 기존 term_i를 root로 저장했던 term들에 대해서 term_j의 root로 지정
for($r=0; $r<$numTerm; $r++)
{
if($whoIsMyRoot[$termIdx2Id[$r]] == $tempRoot)
{
$whoIsMyRoot[$termIdx2Id[$r]] = $whoIsMyRoot[$termIdx2Id[$j]];
$edgeFromRoot[$termIdx2Id[$r]] += $edgeFromRoot[$termIdx2Id[$j]] + 1; // root로 부터의 거리 재계산
}
if($myGraph[$termIdx2Id[$r]]==$tempGraph)
{
$myGraph[$termIdx2Id[$r]]=$myGraph[$termIdx2Id[$j]];
}
}
$termHavingRelation[$termIdx2Id[$i]] = $whoIsMyRoot[$termIdx2Id[$i]];
$termHavingRelation[$termIdx2Id[$j]] = $whoIsMyRoot[$termIdx2Id[$j]];
// break;
}
}
else // 같이 나오는 document가 하나도 없어요. relation 계산도 안함.
{
$proiGivenj = $projGiveni = 0;
}
}
}
// 중복된 관계 제거
foreach ($termRelation as $key1 => $subTermRelation1) {
foreach ($subTermRelation1 as $key2 => $value1) {
if($value1 == $parentChildRelation)
{
foreach ($termRelation as $key3 => $subTermRelation2) {
foreach ($subTermRelation2 as $key4 => $value2) {
if($value2 == $parentChildRelation
&& ($key2 == $key4 && $key1 != $key3))
{
$termRelation[$key3][$key4] = null;
}
}
}
}
}
}
foreach ($termRelation as $key1 => $subTermRelation1) {
foreach ($subTermRelation1 as $key2 => $value1) {
if(is_null($termRelation[$key1][$key2]))
{
unset($termRelation[$key1][$key2]);
}
}
}
if(@$_REQUEST["debug"] == 1) echo 'f5_POST:'. (floatMicrotime()-$f5_naive).' ';
}
// KBS 저널 논문에 제안된 COMPACT 메소드
// -----------------------------
$f5_compact = floatMicrotime();
$termRelation = array();
$termRalationSibling = array();
$whoIsMyRoot = array();
$termHavingRelation = array();
$myGraph = array();
for($i=0; $i<$numTerm; $i++)
{
$whoIsMyRoot[$termIdx2Id[$i]] = $termIdx2Id[$i]; // 이거 두개는
$edgeFromRoot[$termIdx2Id[$i]] = 0; // 나중에 저장된 relation으로 tree 구조 visualize 해줄 때 필요할거 같아서
$myGraph[$termIdx2Id[$i]] = $i;
}// 각 term의 graph상 root값을 자기 자신으로 초기화, root로부터의 거리는 0
for($i=0; $i<$numTerm-1; $i++)
{
for($j=$i+1; $j<$numTerm; $j++)
{
$proiGivenj = $projGiveni = 0;
if($termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]]!=0) // term_i와 term_j가 같이 나오는 document가 있어요
{
$proiGivenj = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$j]]; // P(Term_i | Term_j)
$projGiveni = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$i]]; // P(Term_j | Term_i)
if($projGiveni > $parentThreshold && $proiGivenj < $parentThreshold) // Term_j는 Term_i의 parent
{
// echo "** $proiGivenj({$termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]]}/{$docFreq[$termIdx2Id[$j]]}), ".
// "$projGiveni({$termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]]}/{$docFreq[$termIdx2Id[$i]]}) ";
$termRelation[$termIdx2Id[$j]][$termIdx2Id[$i]] = $parentChildRelation;
$tempRoot = $whoIsMyRoot[$termIdx2Id[$i]];
$tempEdge = $edgeFromRoot[$termIdx2Id[$i]];
$tempGraph = $myGraph[$termIdx2Id[$i]];
for($r=0; $r<$numTerm; $r++)
{
if($whoIsMyRoot[$termIdx2Id[$r]] == $tempRoot)
{
$whoIsMyRoot[$termIdx2Id[$r]] = $whoIsMyRoot[$termIdx2Id[$j]];
$edgeFromRoot[$termIdx2Id[$r]] += $edgeFromRoot[$termIdx2Id[$j]] + 1; // root로 부터의 거리 재계산
}
if($myGraph[$termIdx2Id[$r]]==$tempGraph)
{
$myGraph[$termIdx2Id[$r]]=$myGraph[$termIdx2Id[$j]];
}
}
$termHavingRelation[$termIdx2Id[$i]] = $whoIsMyRoot[$termIdx2Id[$i]];
$termHavingRelation[$termIdx2Id[$j]] = $whoIsMyRoot[$termIdx2Id[$j]];
break;
}
// else {
// echo "$proiGivenj, $projGiveni ";
// }
}
else // 같이 나오는 document가 하나도 없어요. relation 계산도 안함.
{
$proiGivenj = $projGiveni = 0;
}
}
}
if(@$_REQUEST["debug"] == 1) echo 'f5_COMPACT_minimal:'. (floatMicrotime()-$f5_compact).' ';
$f6 = floatMicrotime();
$numTerm = sizeof($termIdx2Id);
$candidateList = array();
// additional relation detect
for($i=0; $i<$numTerm-1; $i++)
{
foreach($termIdx2Id as $key => $val)
{
$candidateList[$val] = true;
}
$candidateList[$termIdx2Id[$i]] = false;
// find ancestors of $termIdx2Id[$i]
$ancestorList = array();
$ancestorList = findAncestors($i, $ancestorList, $termIdx2Id, $termRelation, $parentChildRelation, $overTreeRelation, $numTerm);
foreach($ancestorList as $key => $val)
{
$candidateList[$key] = false;
}
for($j=$i+1; $j<$numTerm; $j++)
{
$proiGivenj = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$j]]; // P(Term_i | Term_j)
$projGiveni = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$i]]; // P(Term_j | Term_i)
// $projGiveni = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$i]];
// $proiGivenj = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$j]];
if(($candidateList[$termIdx2Id[$j]] == true)
&& ($projGiveni > $parentThreshold) && ($proiGivenj < $parentThreshold)) // term_j가 속하는 tree와 relation이 존재하지 않고, threshold를 만족할 때
{
$termRelation[$termIdx2Id[$j]][$termIdx2Id[$i]] = $overTreeRelation;
$tempGraph = $myGraph[$termIdx2Id[$i]];
for($r=0; $r<$numTerm; $r++)
{
if($myGraph[$termIdx2Id[$r]]==$tempGraph)
{
$myGraph[$termIdx2Id[$r]]=$myGraph[$termIdx2Id[$j]];
}
}
$treeIndex[$myGraph[$termIdx2Id[$i]]] = 1;
$termHavingRelation[$termIdx2Id[$i]] = $whoIsMyRoot[$termIdx2Id[$i]];
$termHavingRelation[$termIdx2Id[$j]] = $whoIsMyRoot[$termIdx2Id[$j]];
// find ancestors of $termIdx2Id[$j]
$ancestorList = array();
$candidateList[$termIdx2Id[$j]] = false;
$ancestorList = findAncestors($j, $ancestorList, $termIdx2Id, $termRelation, $parentChildRelation, $overTreeRelation, $numTerm);
foreach($ancestorList as $key => $val)
{
$candidateList[$key] = false;
}
}
}
}
$currentTime = floatMicrotime();
if(@$_REQUEST["debug"] == 1) echo 'f6_COMPACT_compact:'. ($currentTime-$f6).' ';
if(@$_REQUEST["debug"] == 1) echo 'f5+f6 = COMPACT:'. ($currentTime - $f5_compact).' ';
/*
Pairwise로 sibling 찾기
*/
$f6_sibling = floatMicrotime();
for($i=0; $i<$numTerm-1; $i++)
{
for($j=$i+1; $j<$numTerm; $j++)
{
$proiGivenj = $projGiveni = 0;
if($termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]]!=0) // term_i와 term_j가 같이 나오는 document가 있어요
{
$proiGivenj = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$j]]; // P(Term_i | Term_j)
$projGiveni = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$i]]; // P(Term_j | Term_i)
// $proiGivenj = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$j]]; // P(Term_i | Term_j)
// $projGiveni = $termTerm[$termIdx2Id[$i]][$termIdx2Id[$j]] / $docFreq[$termIdx2Id[$i]]; // P(Term_j | Term_i)
if(($proiGivenj > $siblingThreshold) && ($projGiveni > $siblingThreshold)) // Term_i와 Term_j 는 sibling
{
$termRalationSibling[$termIdx2Id[$j]][$termIdx2Id[$i]] = $RELATION_SIBLING;
$tempGraph = $myGraph[$termIdx2Id[$i]];
for($r=0; $r<$numTerm; $r++)
{
if($myGraph[$termIdx2Id[$r]]==$tempGraph)
{
$myGraph[$termIdx2Id[$r]]=$myGraph[$termIdx2Id[$j]];
}
}
$termHavingRelation[$termIdx2Id[$i]] = $whoIsMyRoot[$termIdx2Id[$i]];
$termHavingRelation[$termIdx2Id[$j]] = $whoIsMyRoot[$termIdx2Id[$j]];
}
}
}
}
if(@$_REQUEST["debug"] == 1) echo 'f6_sibling detection:'. (floatMicrotime()-$f6_sibling).' ';
$f7 = floatMicrotime();
//////////////////////////////////////////
// term index와 실제 문자열과 맵핑 리스트 가져오기
// --------------------------
// 쿼리 생성: term index를 실제 문자열과 맵핑하는 배열 생성.
$query = "SELECT fnum, termname, type FROM idf WHERE fnum in (";
$count = 0;
foreach($termIdx2Id as $termId)
{
if($count != 0) $query .=",";
$query .= "$termId";
$count++;
}
$query .=");";
$termNameMapping = array();
if($mysql_result = mysqli_query($mysqlLink, $query))
{
while($row = mysqli_fetch_assoc($mysql_result))
{
$termNameMapping[$row["fnum"]] = $row["termname"];
}
}
mysqli_free_result($mysql_result);
unset($mysql_result);
if(@$_REQUEST["debug"] == 1) echo 'f7 term index와 실제 문자열과 맵핑 리스트 가져오기:'. (floatMicrotime()-$f7).' ';
$f8 = floatMicrotime();
// Sibiling 처리
// =============================
// sibling 끼리 그룹으로 묶습니다. sibling 끼리는 undirected graph
$siblingGroup = array();
$remainSiblingCheck = $termRalationSibling;
// print_r($remainSiblingCheck);
// echo " ";
while(sizeof($remainSiblingCheck)>0)
{
$currentSiblingGroup = array();
$remainKeys = array_keys($remainSiblingCheck);
// 현재 텀과, 현재 텀의 sibling을 sibling 그룹에 추가
$currentSiblingGroup[] = $remainKeys[0];
$currentSiblingGroup = array_merge($currentSiblingGroup, array_keys($remainSiblingCheck[$remainKeys[0]]));
unset($remainSiblingCheck[$remainKeys[0]]);
// var_dump($remainSiblingCheck);
$siblingGroup[] = $currentSiblingGroup;
}
// echo " ";
// print_r($siblingGroup);
// echo " ";
$deleteIndex = array();
// 각각 그룹별로 겹치는게 있으면 하나의 그룹으로 통합
for($i=0;$i=0;$i--)
{
unset($siblingGroup[$deleteIndex[$i]]);
}
$siblingGroup = array_values($siblingGroup);
// print_r($siblingGroup);
// echo " ";
// sibling 관계에 자손들과 parents-child 관계인 경우 sibling을 가리키도록 변경
// var_dump($siblingGroup);
for($i=0;$i";
print_r($siblingGroup[$i]);
echo " ";
}
for($j=0;$j{$termNameMapping[$termIdJ]}, {$termNameMapping[$termIdK]} ";
$overTreeDirection = 0;
// var_dump($result_intersect);
foreach($result_intersect as $key => $value)
{
// echo "__{$overTreeDirection},i:{$i},j:{$j},k:{$k},value:{$value},{$termRelation[$termIdJ][$value]}, {$termRelation[$termIdK][$value]}" ;
if($termRelation[ $termIdJ ][$value] ==$overTreeRelation
&& (isset($termRelation[ $termIdK ][$value])
&& $termRelation[ $termIdK ][$value] == $parentChildRelation))
{
unset($termRelation[ $termIdJ ][$value]);
// if(empty($termRelation[ $siblingGroup[$i][$j] ]))
// {
// unset($termRelation[ $siblingGroup[$i][$j] ]);
// }
$overTreeDirection |= 1;
}
if((isset($termRelation[ $termIdJ ][$value])
&& $termRelation[ $termIdJ ][$value] == $parentChildRelation)
&& ($termRelation[ $termIdK ][$value] == $overTreeRelation ))
{
unset($termRelation[ $termIdK ][$value]);
// if(empty($termRelation[ $siblingGroup[$i][$k] ]))
// {
// unset($termRelation[ $siblingGroup[$i][$k] ]);
// }
$overTreeDirection |= 2;
}
// echo "{$overTreeDirection}({$termNameMapping[$value]}) ";
}
// echo " ".$overTreeDirection;
if($overTreeDirection & 1)
{
$termRelation[ $siblingGroup[$i][$j]][$siblingGroup[$i][$k]] = $overTreeRelation;
// echo "->:{$termNameMapping[$siblingGroup[$i][$j]]}, {$termNameMapping[$siblingGroup[$i][$k]]} ";
}
if($overTreeDirection & 2)
{
$termRelation[ $siblingGroup[$i][$k]][$siblingGroup[$i][$j]] = $overTreeRelation;
// echo "<-:{$termNameMapping[$siblingGroup[$i][$j]]}, {$termNameMapping[$siblingGroup[$i][$k]]} ";
}
}
}
}
}
}
// print_r($termRelation);
// print_r($termNameMapping);
if(@$_REQUEST["debug"] == 1) echo 'f8:'. (floatMicrotime()-$f8).' ';
//소요 시간 출력
if(@$_REQUEST["debug"] == 1) echo "time: ".(floatMicrotime()-$startTime)." ";
$numOfParentChild = 0;
$numOfSibling = 0;
$xml = "";
$xml .= "getKeyword()}\" totalDoc=\"{$numOfDocuments}\" >\n";
// perent-child relation
foreach($termRelation as $key1 => $val1)
{
foreach($val1 as $key2 => $val2) // $key1 : term1, $key2 : term2, $val2 : relation
{
if($val2==$parentChildRelation)
{
$xml .= "\n";
//parent 정보 출력
$xml .= "";
$xml .= $termNameMapping[$key1]."\n";
//child 정보 출력
$xml .= "";
$xml .= $termNameMapping[$key2]."\n";
$xml .= "\n";
$numOfParentChild++;
}
}
}
// sibling relation
foreach($termRalationSibling as $key1 => $val1)
{
foreach($val1 as $key2 => $val2) // $key1 : term1, $key2 : term2, $val2 : relation
{
if($val2==$RELATION_SIBLING)
{
$xml .= "\n";
//sibling1 정보 출력
$xml .= "";
$xml .= $termNameMapping[$key1]."\n";
//sibling2 정보 출력
$xml .= "";
$xml .= $termNameMapping[$key2]."\n";
$xml .= "\n";
$numOfSibling++;
}
}
}
// over-tree relation
foreach($termRelation as $key1 => $val1)
{
foreach($val1 as $key2 => $val2) // $key1 : term1, $key2 : term2, $val2 : relation
{
if($val2==$overTreeRelation)
{
$xml .= "\n";
//parent 정보 출력
$xml .= "";
$xml .= $termNameMapping[$key1]."\n";
//child 정보 출력
$xml .= "";
$xml .= $termNameMapping[$key2]."\n";
$xml .= "\n";
$numOfParentChild++;
}
}
}
// relation을 가지지 않은 node
$xml .= "\n";
foreach($docFreq as $key => $val)
{
if(array_key_exists($key, $termHavingRelation)==false)
{
$xml .= "";
$xml .= $termNameMapping[$key]."\n";
}
}
$xml .= "\n";
$xml .= "\n";
if(@$_REQUEST["debug"] == 1)
{
echo "# of retrieved docs:".$numOfDocuments." ";
echo "# of concepts".sizeof($termHavingRelation)." ";
echo "# of parent-child relation".$numOfParentChild." ";
echo "# of sibling relation".$numOfSibling." ";
echo $xml;
}
$ontolotyMapBefore='
";
/* // term1 (term1의 root, term1이 속한 그래프번호) term2 (term2의 root, term2가 속한 그래프 번호) : term1과 term2의 relation
foreach($termRelation as $key1 => $val1)
{
foreach($val1 as $key2 => $val2)
{
echo $key1."(R: ".$whoIsMyRoot[$key1].", G: ".$myGraph[$key1].", F: ".$docFreq[$key1]." ) ".$key2."(R: ".$whoIsMyRoot[$key2].", G: ".$myGraph[$key2].", F: ".$docFreq[$key2]." ) : ".$val2." ";
}
}
*/
unset($termRelation);
unset($termRalationSibling);
unset($termIdx2Id);
unset($termEntropy);
unset($myGraph);
unset($treeIndex);
unset($whoIsMyRoot);
unset($edgeFromRoot);
unset($termTerm);
unset($treeIndex);
unset($docFreq);
unset($termNameMapping);
unset($termTypeMapping);
}
}// endif 내꺼 몽땅 !
return;
/*********************************************************/
$tagCloudBefore='
";
for($i=0;$i<$searchOption->getMeshCount();$i++){
echo"$i -> $meshNameList[$i], the number of occurence: $frequency[$i]";
echo" ";
}
for($i=0;$i<$searchOption->getMeshCount();$i++){
$frequency[$i]=pow($frequency[$i]/$localTotalCount, 2)*pow(-($frequency[$i]/$localTotalCount)*log($frequency[$i]/$localTotalCount, 10), 2);
echo"$i -> $meshNameList[$i], the number of occurence: $frequency[$i]";
echo" ";
}
*/
}
Warning: session_start(): Cannot send session cookie - headers already sent by (output started at /var/www/html/refmed/phpspeedy/class.compressor.php:722) in /var/www/html/refmed/index.php on line 43
Warning: session_start(): Cannot send session cache limiter - headers already sent (output started at /var/www/html/refmed/phpspeedy/class.compressor.php:722) in /var/www/html/refmed/index.php on line 43
Fatal error: Class 'compressor' not found in /var/www/html/refmed/index.php on line 50