trjhtr

My question is about my Ontology which retrieves synonyms
from the ontology by using isSynonymOf object property. (Note that
my Ontology file is 260 kb and expected to become 500kb). I am using
following code to retrieve the synonyms. The performance in terms of
response time is very low. It takes a lot of time(minutes) to
display synonyms. How can i improve it?
Note: The problem looks like in similarity algorithm and iterator.

Here is My full Code

import java.io.InputStream;
import java.util.ArrayList;
import java.util.Iterator;

import org.apache.jena.ontology.Individual;
import org.apache.jena.ontology.OntModel;
import org.apache.jena.ontology.OntModelSpec;
import org.apache.jena.ontology.SymmetricProperty;
import org.apache.jena.rdf.model.ModelFactory;
import org.apache.jena.rdf.model.NodeIterator;
import org.apache.jena.util.FileManager;

import xyzWordAnalyzer;
import cde.model.SimilarityModel;
public class ontologyConnector

 private static ontologyConnector instance;
 private static OntModel ontModel;
 protected static final Sring SOURCE_FILE = "http://abc.owl";
 protected static final String NS = SOURCE_FILE + "abc";

 public static synchronized ontologyConnector getInstance()
 
 if (instance == null)
 
 instance = new ontologyConnector();
 

 return instance;
 

 public ontologyConnector()
 
 ontModel = ModelFactory.createOntologyModel(OntModelSpec.OWL_DL_MEM_RULE_INF);
 InputStream in = FileManager.get().open(SOURCE_FILE);
 ontModel.read(in, "");
 

 public ArrayList<SimilarityModel> getRelatums(String keyword)
 
 ArrayList<String> list = new ArrayList<String>();

 keyword = keyword.replaceAll(" ", "_");

 SymmetricProperty isSynonymOf = ontModel.getSymmetricProperty("http://www.semanticweb.org/abc#isSynonymOf");

 Iterator<Individual> iterInd = ontModel.listIndividuals();
 while (iterInd.hasNext())
 
 Individual ind = iterInd.next();
 if (ind.getLocalName().equalsIgnoreCase(keyword))
 
 NodeIterator iterVal = ind.listPropertyValues(isSynonymOf);
 while (iterVal.hasNext())
 
 list.add(iterVal.nextNode().asResource().getLocalName().toString().replace("_", " "));
 
 
 

 list = removeDuplicationfromList(keyword, list);

 // list.remove(keyword);

 ArrayList<SimilarityModel> simList = new ArrayList<SimilarityModel>();

 Double tot = (double) 0;
 WordAnalyzer wa = new WordAnalyzer();

 for (int i = 0; i < list.size(); i++) 
 Double tmpDouble = wa.getSimilarity(keyword.toLowerCase(), 
 list.get(i));

 int tmp = 0;
 if (Double.isNaN(tmpDouble)) 
 // tmpDouble = (double) -200;
 tmp = -200;
 else 
 tmpDouble *= 100;
 tot += tmpDouble;
 tmp = tmpDouble.intValue();
 

 SimilarityModel simModel = new SimilarityModel(list.get(i), tmp);
 simList.add(simModel);
 

 Double avg = tot / simList.size();
 if (!Double.isNaN(avg)) 
 for (int i = 0; i < simList.size(); i++) 
 if (simList.get(i).getSimilarity() == -200) 
 simList.get(i).setSimilarity(avg.intValue());
 
 
 

 return simList;


 private ArrayList<String> removeDuplicationfromList(String str, 
 ArrayList<String> list) 
 for (int i = list.size() - 1; i >= 0; i--) 
 if (list.get(i).toString().equalsIgnoreCase(str)) 
 list.remove(i);
 
 
 return list;

Design

You have implemented the class as a singleton, but made the constructor public. That makes the use of the singleton pattern kind of useless. Since you want the class to only be instantiated once, which is inside the class itself, you should make it private.

public ArrayList<SimilarityModel> getRelatums(String keyword)

Especially in the API of a class, i. e. its public members like this method, you should use interfaces or abstract types, rather than specific implementations. If you replace ArrayList<SimilarityModel> with List<SimilarityModel>, you can use any implementation of List that you like, without having to change many parts of your code.

Style

The Java convention for class names is UpperCamelCase. ontologyConnector should be OntologyConnector.

Your names are not only inconsistent in capitalization, but also abbreviation. OntModel should be written out as OntologyModel, to fit with OntologyConnector but also because abbreviations sometimes force the reader to guess what it means and thus destroy the code's readability.

public ArrayList<SimilarityModel> getRelatums(String keyword)

The return type suggests that the method returns a collection of SimilarityModels, but is named getRelatums. It should be called getSimilarityModels, if that's what it is actually doing. If not, then the name of the type SimilarityModel might be chosen badly.

As others have already pointed out in the comments, in order to get help with the performance problem you need to add more information and code (e. g. the classes OntModel and SymmetricProperty). However in my opinion there is still a lot you can improve in the already available code to make it worth a review. Please add more code to get further suggestions as performance improvements.