function getLabelDistribution($crisisID)
{
$crisisID = intval($crisisID);
$sql = "select\n mf.nominalAttributeID,\n nl.nominalLabelCode,\n count(distinct d.documentID) as labelCountTotal,\n sum(!isEvaluationSet) as labelCountTraining,\n sum(isEvaluationSet) as labelCountEvaluation\n from model_family mf\n left join nominal_label nl on nl.nominalAttributeID=mf.nominalAttributeID\n left join document_nominal_label dnl on dnl.nominalLabelID=nl.nominalLabelID\n left join document d on d.documentID=dnl.documentID\n where mf.crisisID={$crisisID}\n group by nominalAttributeID, nominalLabelCode\n order by nominalAttributeID, nominalLabelCode!='null', labelCountTraining desc";
$attributes = array();
$con = getMySqlConnection();
$result = $con->query($sql);
if ($result) {
$a = null;
while ($row = $result->fetch_assoc()) {
$attributeID = intval($row["nominalAttributeID"]);
if (is_null($a) || $a->attributeID != $attributeID) {
$a = new AttributeStatistics();
$a->attributeID = $attributeID;
$a->rows = array();
$attributes[$attributeID] = $a;
}
$d = new LabelDistribution();
$d->labelCode = $row["nominalLabelCode"];
$d->labelCountTotal = intval($row["labelCountTotal"]);
$d->labelCountTraining = floatval($row["labelCountTraining"]);
$d->labelCountEvaluation = intval($row["labelCountEvaluation"]);
$a->rows[] = $d;
}
}
return $attributes;
}
function aggregateTaskAnswers($taskID)
{
global $minAgreement;
global $dontknow;
$taskID = intval($taskID);
$con = getMySqlConnection();
//Get all answers for this task
if ($result = $con->query("select answer from task_answer where documentID = {$taskID} and answer!='null'")) {
//Find the most frequent label per attribute
$answers = array();
while ($row = $result->fetch_assoc()) {
$answerSet = json_decode($row["answer"]);
foreach ($answerSet as $answer) {
$answers[$answer->attributeID][] = $answer->labelID;
}
}
$result->close();
$finalAnswers = array();
foreach ($answers as $attributeID => $labelIDs) {
$c = array_count_values($labelIDs);
if (max($c) < $minAgreement) {
//More answers are needed to reach agreement. Clean up and return.
$con->close();
return;
}
$val = array_search(max($c), $c);
$finalAnswers[$attributeID] = intval($val);
}
//Save the final answers
foreach ($finalAnswers as $attributeID => $label) {
if ($label == $dontknow) {
continue;
}
$con->query("insert ignore into document_nominal_label (documentID, nominalLabelID, timestamp) values " . "({$taskID}, {$label}, utc_timestamp())");
}
//Mark task as completed
$con->query("update document set hasHumanLabels=1 where documentID={$taskID}");
//Send notification that a new training sample has arrived
$crisisID = intval(selectSingleValue($con, "select crisisid from document where documentid={$taskID}"));
notifyRedis($crisisID, array_keys($answers));
}
$con->close();
}
