Java Code Examples for org.semanticweb.owlapi.model.OWLClass#compareTo()

public void makeAllByAllLowestCommonSubsumer() {
	LOG.info("all x all...");		

	Set<OWLClass> atts = getAllAttributeClasses();
	for (OWLClass a : atts) {
		LOG.info("  "+a+" vs ALL");		
		for (OWLClass b : atts) {
			// LCS operation is symmetric, only pre-compute one way
			if (a.compareTo(b) > 0) {
				OWLClass lcs = this.getLowestCommonSubsumerClass(a, b);
				//System.out.println("LCS( "+pp(a)+" , "+pp(b)+" ) = "+pp(lcs));
			}
		}
	}
	LOG.info("DONE all x all");		
	reason();
}
 

/**
 * generates new LCS grouping classes
 * 
 * @param leafClasses
 */
public void generateLeastCommonSubsumers(Set<OWLClass> leafClasses) {

	LOG.info("Generating LCSs en masse; leaf classes "+leafClasses.size());
	LOG.info("Num materialized class expressions (prior) "+materializedClassExpressionMap.size());
	for (OWLClass a : leafClasses) {
		LOG.info("  ALL vs: "+a+" '"+getAnyLabel(a)+"'");		
		for (OWLClass b : leafClasses) {
			// LCS operation is symmetric, only pre-compute one way
			if (a.compareTo(b) > 0) {
				OWLClass lcs = getLowestCommonSubsumerClass(a, b, leafClasses);
			}
		}
	}
	LOG.info("DONE all x all");		
	LOG.info("Num materialized class expressions (post) "+materializedClassExpressionMap.size());
}
 

@CLIMethod("--sim-save-lcs-cache")
public void simSaveLCSCache(Opts opts) throws Exception {
	opts.info("[-m ICTHRESHOLD] OUTFILE", "saves a LCS cache to a file. This should be called AFTER --sim-compare-atts");
	Double thresh = null;
	OWLClass debugClass = null;
	while (opts.hasOpts()) {
		if (opts.nextEq("-m|--min-ic")) {
			opts.info("ICTHRESHOLD", "If the IC of the LCS is less than this value, an entry is not written.\n" +
					"After subsequent loading of the cache, pairs with no entry are equivalent to pairs with a LCS with IC=0");
			thresh = Double.valueOf(opts.nextOpt());
		}
		else if (opts.nextEq("--debug-class")) {
			debugClass = this.resolveClass(opts.nextOpt());
		}
		else {
			break;
		}
	}

	// No Sim object, so all by all has not yet been calculated
	if (owlsim == null) {
		owlsim = getOwlSimFactory().createOwlSim(g.getSourceOntology());
		if (debugClass != null) {
			((FastOwlSim)owlsim).debugClass = debugClass;
		}
		owlsim.createElementAttributeMapFromOntology();

		// we have just created a sim object, so the cache is not populated;
		// therefore we can be more efficient by avoiding lookups
		owlsim.setNoLookupForLCSCache(true);
		Set<OWLClass> atts = owlsim.getAllAttributeClasses();
		LOG.info("Number of attribute classes: "+atts.size());
		for (OWLClass i : atts) {
			//LOG.info("Comparing "+i+" to all attributes");
			for (OWLClass j : atts) {
				// metric is symmetrical
				if (i.compareTo(j) < 0)
					continue;
				owlsim.getLowestCommonSubsumerWithIC(i, j, thresh);					
			}
		}

	}

	owlsim.saveLCSCache(opts.nextOpt(), thresh);
	LOG.info("Saved cache");
}
 

/**
 * gets the LCS. If multiple named LCSs exist in the ontology, then
 * a class expression is generated from the filtered intersection
 * 
 * @param a
 * @param b
 * @return OWLClassExpression
 */
public OWLClassExpression getLowestCommonSubsumer(OWLClassExpression a, OWLClassExpression b) {
	if (a.equals(b)) {
		return a;
	}
	if (a instanceof OWLClass && b instanceof OWLClass) {
		if (getReasoner().getSuperClasses(a, false).getFlattened().contains(b)) {
			return b;
		}
		if (getReasoner().getSuperClasses(b, false).getFlattened().contains(a)) {
			return a;
		}
	}

	// returns the set of named LCSs that already exist
	Set<Node<OWLClass>> ccs = getNamedLowestCommonSubsumers(a,b);

	// single LCS - return this directly
	if (ccs.size() == 1) {
		return ccs.iterator().next().getRepresentativeElement();
	}

	// make a class expression from the filtered intersection of LCSs

	Set<OWLClass> ops = new HashSet<OWLClass>();
	Set<OWLClass> skips = new HashSet<OWLClass>();
	OWLClass best = null;
	double bestIC = 0.0;
	for (Node<OWLClass> n : ccs) {
		OWLClass c = n.getRepresentativeElement();
		// allows custom filtering; e.g. upper-level classes
		// note: should be removed at view stage			
		if (this.getVerbotenClasses().contains(c))
			continue;
		// TODO: custom filtering
		boolean skip = false;
		Double ic = getInformationContentForAttribute(c);
		if (ic == null) {
			// if the attributes classes have not been filtered, then null values
			// (ie classes with no instances) are possible
			continue;
		}
		if (ic > bestIC) {
			bestIC = ic;
			best = c;
		}
		if (ic < 2.5) { // TODO: configure
			//LOG.info("SKIPPING: "+c+" IC="+ic);
			continue;
		}
		if (this.getNamedReflexiveSubsumers(c).size() < 2) {
			LOG.info("SKIPPING: "+c+" no parents");
			// non-grouping attribute
			continue;
		}
		// don't make intersections of similar elements
		for (Node<OWLClass> n2 : ccs) {
			OWLClass c2 = n2.getRepresentativeElement();
			double ic2 = getInformationContentForAttribute(c2); 
			// prioritize the one with highest IC
			if (ic < ic2 || (ic==ic2 && c.compareTo(c2) > 0)) {
				// TODO: configure simj thresh
				if (this.getAttributeJaccardSimilarity(c, c2) > 0.75) {
					LOG.info("SKIPPING: "+c+" too similar to "+n2);
					skip = true;
				}
			}
		}
		if (skip)
			continue;
		// not filtered
		ops.add(c);
	}

	if (ops.size() == 1) {
		return ops.iterator().next();
	}
	if (ops.size() == 0) {
		// none pass: choose the best representative of the intersection
		return best;
	}
	return owlDataFactory.getOWLObjectIntersectionOf(ops);
}
 

public static void compareLCSWithEnrichmentGrouping(OWLOntology ont, Double pthresh, OWLPrettyPrinter owlpp) throws UnknownOWLClassException, MathException {
	OwlSimFactory osf = new FastOwlSimFactory();
	OwlSim sim = osf.createOwlSim(ont);
	sim.createElementAttributeMapFromOntology();
	OWLClass populationClass = null;
	if (pthresh == null)
		pthresh = 0.01;
	int nHypotheses = ((sim.getAllAttributeClasses().size()-1) * sim.getAllAttributeClasses().size())/2;
	for (OWLClass c : sim.getAllAttributeClasses()) {
		for (OWLClass d : sim.getAllAttributeClasses()) {
			if (c.compareTo(d) > 0) {
				ScoreAttributeSetPair lcsic = sim.getLowestCommonSubsumerWithIC(c, d);
				//System.out.println("TEST: "+c+" "+d);
				EnrichmentResult enr = sim.calculatePairwiseEnrichment(populationClass, c, d);
				if (enr == null) {
					//System.err.println("NULL: "+c+" "+d);
					continue;
				}
				double pvc = enr.pValue * nHypotheses;
				if (pvc < pthresh) {
					// TODO - place this in OwlSim
					Set<OWLNamedIndividual> inds = 
							new HashSet<OWLNamedIndividual>(sim.getElementsForAttribute(c));
					inds.addAll(sim.getElementsForAttribute(d));
					double pu = inds.size() / (double)sim.getCorpusSize();
					double ic = - Math.log(pu) / Math.log(2);
					double foldChange = ic / lcsic.score;
					if (foldChange < 1.5)
						continue;
					if (lcsic.attributeClassSet.size() == 0) {
						System.err.println("# no LCS for "+c+" "+d);
						continue;
					}
					System.out.println(owlpp.render(c)+"\t"+owlpp.render(d)+"\t"+pvc+"\t"+owlpp.render(lcsic.getArbitraryAttributeClass())+"\t"+lcsic.score+"\t"+ic+"\t"+foldChange);
				}
			}
		}
		
		
	}
}