/learn_one_category_5.php

#!/usr/bin/php
<?php
$folder = $argv[1] ;
$count  = $argv[2] ;
$limit  = $argv[3] ;


include("functions.php");

echo "working on folder '".$folder."' \n\n";
$dllearnerurl = "http://139.18.2.57:8888/interfaces/rest" ;
	
	$ListInstancesOfCategory = file_array_outputer($folder."/instances.list");	// list of relevant examples (+)
	$rand = 5;
	$pos = get_random_uri($ListInstancesOfCategory, $rand);
	//$pos = array("http://dbpedia.org/resource/John_Skandalis","http://dbpedia.org/resource/Michael_Crockett","http://dbpedia.org/resource/Gareth_Ellis","http://dbpedia.org/resource/Ken_McGuinness","http://dbpedia.org/resource/Tim_Moltzen");
	var_dump($pos);
	$ListInstancesOfCategorysSister = file_array_outputer($folder."/instances_catsister.list");  // list of non-relevant examples (-)
	$neg = get_random_uri($ListInstancesOfCategorysSister, $rand);
	//$neg = array("http://dbpedia.org/resource/Aaron_Murphy","http://dbpedia.org/resource/Lee_Te_Maari","http://dbpedia.org/resource/Liam_Harrison_%28rugby_league%29","http://dbpedia.org/resource/Gavin_Dodd","http://dbpedia.org/resource/Dominic_Feau%27nati");
	var_dump($neg);

$allinstances  = file($folder."/instances.list");
if(count($allinstances)!= 100 ){
	echo "ERROR count was not 100: ". count($allinstances)."\n";
	};


$step = 1;
$sum_step = 5;									// LOOP counter
$maxfoundeditems = 0;
$collectmachteditems = array();
$allfoundeditems = 0;						// collection all founded instances
$lastsparql = "";
$instancesOfCategory = file_array_outputer($folder."/instances.list");
$index_position = 4;
$category_name = substr($folder,$index_position);

// INITIAL VARIABLES TO COLLECT DATA
$bestaccuracy 	= 0;
$lastaccuracy 	= 0;
$bestconcept 	= 0;
$lastconcept 	= "";
$lastsparql  	= "";  
$best_count_matching_in_all_steps = 0;		// collection the best number of positive results in any steps
  
while ($step <= $sum_step) {	
	$count_matching = 0;	
	$learned_results = learn($dllearnerurl, $limit, $pos, $neg);
	echo "########################     RESULTS after ".$step.".learning     ######################## \n\n";
	print_r($learned_results);
	echo "########################  [END] RESULTS after ".$step.".learning  ######################## \n\n";

		// STORE ALL RESULTS AFTER LEARNING IN AN ARRAY
		$SPARQL["success"]  = $learned_results ["learningresult"]["success"];
        $SPARQL["accuracy"] = $learned_results ["learningresult"]["accuracy"];
        $SPARQL["concept"]  = $learned_results ["learningresult"]["manchester"];
        $SPARQL["sparql"]   = $learned_results ["learningresult"]["sparql"];

        
        /* Checking for SUCCESSFUL */
        if($SPARQL["success"] == '1') {
                    /** collecting the best accuracy & his concept */
                    if ($bestaccuracy < $SPARQL["accuracy"]) {
                        $bestaccuracy = $SPARQL["accuracy"];
                        $bestconcept  = $SPARQL["concept"];
                    }
                    
                    /** collecting the last accuracy & his concept */
                    $lastaccuracy = $SPARQL["accuracy"];
                    $lastconcept  = $SPARQL["concept"];
                    $lastsparql   = $SPARQL["sparql"];
                      
                    /* get URI over live.dbpedia.org endpoint within the received query */
                    $defaultgraphURI = 'http://dbpedia.org';
                    $sparqlQueryString = $SPARQL["sparql"];
                    $sparqlQueryString = str_replace("SELECT", "SELECT DISTINCT", $sparqlQueryString);
                    $ep = "http://hanne.aksw.org:8892/sparql?query=";
                    
                /* EXECUTE THE RETURNED SPARQL QUERY */
				$ListOfResultbyLearning =  executeSparqlQuery_json ($sparqlQueryString, "subject");
				$ListOfResultPosbyLearning = array ();
                echo "============= results from executing the sparql query =============  \n";

                foreach($ListOfResultbyLearning as $ins_uri) {
                                      
                    foreach($ListInstancesOfCategory as $ins)
                    {
                        if($ins_uri === $ins) {
                            $collectmachteditems[$ins] = 1 ;
                            $ListOfResultPosbyLearning [] = $ins;   // collect all positive results
                            $allfoundeditems++;
                            $count_matching++;
                        }
                    }
                }

                echo "ListOfResultbyLearning \n";
                print_r($ListOfResultbyLearning);
				
				// Compare two arrays: ListOfResultbyLearning and ListOfResultPosbyLearning
				// to get an new arrays: ListOfResultNegbyLearning
				$ListOfResultNegbyLearning = array_diff($ListOfResultbyLearning,$ListOfResultPosbyLearning); // collect all negative results

                echo "* Total Matching: ".$count_matching."\n";
                
                /** Compare to get the best number of matching results */
                if($best_count_matching_in_all_steps < $count_matching) {
					$best_count_matching_in_all_steps = $count_matching;
				}
					
                
                echo "Positive examples \n\n";
                print_r($ListOfResultPosbyLearning);	// List of the returned items which are matched agains the items of the specified category 
                $pos = array_merge(get_random_uri($ListOfResultPosbyLearning, 5),$pos);
                echo "**** (+) Preparation a new list of postive examples for the next step **** \n";
                print_r($pos);
				
				echo "Negative examples \n\n";
				print_r($ListOfResultNegbyLearning);	// List of the returned items which are matched agains the items of the specified category 
            		
				// initial a new negative example list
				$neg = array_merge(get_random_uri($ListOfResultNegbyLearning, 5),$neg);
				echo "**** (-) Preparation a new list of negative examples for the next step *** \n";
				print_r($neg);
			}			
            /* END: Checking for SUCCESSFUL */
        

$step++;

}
			echo "** Totally collected items ".count($collectmachteditems);

			$maxfoundeditems = count(array_keys($collectmachteditems));
			$numberofinstancesOfCategory = count(array_keys($instancesOfCategory));

            $CategoryStatisticDetails ["categoryname"]  	= $category_name;
            $CategoryStatisticDetails ["bestaccuracy"]  	= $bestaccuracy;
            $CategoryStatisticDetails ["lastaccuracy"]  	= $lastaccuracy;
            $CategoryStatisticDetails ["bestconcept"]   	= $bestconcept;
            $CategoryStatisticDetails ["lastconcept"]  		= $lastconcept;
            $CategoryStatisticDetails ["maxfoundeditems"] 	= $maxfoundeditems;
            $CategoryStatisticDetails ["recallMAX"]     	= ($maxfoundeditems/$numberofinstancesOfCategory) ;            // 
            $CategoryStatisticDetails ["recallBEST"]     	= ($best_count_matching_in_all_steps / $numberofinstancesOfCategory) ;
            //$CategoryStatisticDetails ["recallBEST"]    	= ($bestaccuracy / 100);                // calculation follow the best accuracy
            $CategoryStatisticDetails ["recallAVG"]     	= ($allfoundeditems/($sum_step*$numberofinstancesOfCategory));
            $CategoryStatisticDetails ["precTOTAL"]     	= ($allfoundeditems/($sum_step*$limit));    // 
            $CategoryStatisticDetails ["precBEST"]      	= ($best_count_matching_in_all_steps /$limit);              // 
            $CategoryStatisticDetails ["precAVG"]       	= ($allfoundeditems/($sum_step*$limit));
            $CategoryStatisticDetails ["lastsparql"]    	= $lastsparql;                          //
        
            echo "CategoryStatisticDetails \n";
            print_r($CategoryStatisticDetails);
   /*  START: Adding data into sheet */
	include ("class.spreadsheet.php");
	$doc = new spreadsheet();
	$acc = array("dmlldHBsYWNlc0Bnb29nbGVtYWlsLmNvbQ==","ODIxNTIyODIxNTIy");
	$doc->authenticate(base64_decode($acc[0]), base64_decode($acc[1]));
	$doc->setSpreadsheet("MatchedResults");
	$doc->setWorksheet("Statistic-DL-Learner-$limit");
	// END initial a doc
    
             $my_data = array("Nr" => $count,
                             "Category Name" => $CategoryStatisticDetails ["categoryname"],
                             "Best Accuracy" => $CategoryStatisticDetails ["bestaccuracy"],
                             "Last Accuracy" => $CategoryStatisticDetails ["lastaccuracy"],
                             "Best Concept" => $CategoryStatisticDetails ["bestconcept"],
                             "Last Concept" => $CategoryStatisticDetails ["lastconcept"],
                             "#Max collected Members" => $CategoryStatisticDetails ["maxfoundeditems"],
                             "recallMAX" => round($CategoryStatisticDetails ["recallMAX"],4),
                             "recallBEST" => round($CategoryStatisticDetails ["recallBEST"],4),
                             "recallAVG" => round($CategoryStatisticDetails ["recallAVG"],4) ,
                             "precTOTAL" => round($CategoryStatisticDetails ["precTOTAL"],4),
                             "precBEST" => round($CategoryStatisticDetails ["precBEST"],4),
                             "precAVG" => round($CategoryStatisticDetails ["precAVG"],4),
                             "Last SPARQL query" => $CategoryStatisticDetails ["lastsparql"]);
            $doc->add($my_data);
   
   /*  END: Adding data into sheet */