Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[1610] trunk
|
From: <jen...@us...> - 2009-02-18 08:54:20
|
Revision: 1610
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=1610&view=rev
Author: jenslehmann
Date: 2009-02-18 08:54:16 +0000 (Wed, 18 Feb 2009)
Log Message:
-----------
accuracy of solutions displayed; some cleanup
Modified Paths:
--------------
trunk/doc/manual/manual.tex
trunk/src/dl-learner/org/dllearner/Info.java
trunk/src/dl-learner/org/dllearner/algorithms/refinement2/ROLearner2.java
trunk/src/dl-learner/org/dllearner/learningproblems/PosNegLPStandard.java
trunk/src/dl-learner/org/dllearner/learningproblems/PosOnlyLP.java
Modified: trunk/doc/manual/manual.tex
===================================================================
--- trunk/doc/manual/manual.tex 2009-02-17 19:23:47 UTC (rev 1609)
+++ trunk/doc/manual/manual.tex 2009-02-18 08:54:16 UTC (rev 1610)
@@ -70,30 +70,27 @@
\begin{verbatim}
DL-Learner 2008-10-13 command line interface
-starting component manager ... OK (81ms)
+starting component manager ... OK (157ms)
initialising component "OWL file" ... OK (0ms)
-initialising component "fast instance checker" ... OK (861ms)
-initialising component "two valued definition learning
- problem" ... OK (0ms)
-initialising component "refinement operator based learning
- algorithm II" ... OK (14ms)
+initialising component "fast instance checker" ... OK (842ms)
+initialising component "pos neg learning problem" ... OK (0ms)
+initialising component "refinement operator based
+ learning algorithm II" ... OK (14ms)
starting top down refinement with: Thing (50% accuracy)
more accurate (83,33%) class expression found: male
solutions (at most 20 are shown):
-1: (male and hasChild some Thing) (length 5, depth 3)
+1: (male and hasChild some Thing) (accuracy 100%, length 5, depth 3)
Algorithm terminated succesfully.
number of retrievals: 4
retrieval reasoning time: 0ms ( 0ms per retrieval)
number of instance checks: 93 (0 multiple)
-instance check reasoning time: 0ms ( 0ms per instance check)
-overall reasoning time: 0ms (1,931% of overall runtime)
-overall algorithm runtime: 28ms
+instance check reasoning time: 1ms ( 0ms per instance check)
+overall reasoning time: 1ms (11,016% of overall runtime)
+overall algorithm runtime: 17ms
\end{verbatim}
-\todo{display accuracy of solutions}
-
The first part of the output tells you which components are used (more on this in Section \ref{sec:components}). In the second part you see output coming from the used learning algorithm, i.e.~it can print information while running (``more accurate (83,33\%) class description found'') and the final solutions, it computed. The results are displayed in Manchester OWL Syntax\footnote{\mos}. There can be several solutions, in which case they are ordered with the most promising one in the first position. In this case the only solution is \verb|male and hasChild some Thing| defining the class father. The last part of the output contains some runtime statistics.
\section{DL-Learner Architecture}
@@ -126,7 +123,7 @@
Knowledge sources have a URI and can be included in conf files using \verb|import("$url");|, e.g.~\verb|import("ontology.owl")|. Depending on the file ending, DL-Learner will guess the correct type of knowledge source. If you want to overwrite this, you can use a second parameter with value \verb|OWL|, \verb|KB|, or \verb|SPARQL|, e.g.~\verb|import("ontology.owl","OWL")|.
\begin{description}
- \item[OWL File] DL-Learner supports OWL files in different formats, e.g. RDF/XML or N-Triples. If there is a standard OWL format, you want to use, but is not supported by DL-Learner please let us know. Note: As we use the OWL API for parsing, all formats supported by it are possible\footnote{ for a list see \url{http://owlapi.sourceforge.net/}}.
+ \item[OWL File] DL-Learner supports OWL files in different formats, e.g. RDF/XML or N-Triples. If there is a standard OWL format, you want to use, but is not supported by DL-Learner please let us know. We use the OWL API for parsing, so all formats supported by it can be used\footnote{ for a list see \url{http://owlapi.sourceforge.net/}}.
\item[KB File] KB files are an internal non-standardised knowledge representation format, which corresponds to description logic syntax except that the special symbols have been replaced by ASCII strings, e.g.~\verb|AND| instead of $\sqcap$. You can find several KB files in the examples folder.
\item[SPARQL Endpoint] DL-Learner allows to use SPARQL endpoints as background knowledge source, which enables the incorporation of very large knowledge bases, e.g. DBpedia\cite{2008_dbpedia}, in DL-Learner. This works by using a set of start instances, which usually correspond to the examples in a learning problem, and then retrieving knowledge about these instances via SPARQL queries. The obtained knowledge base fragment can be converted to OWL and consumed by a reasoner later since it is now sufficiently small to be processed in reasonable time. Please see \cite{2009_ijswis} for details, the parameters for the extraction algorithm are similar (different names only) to the ones here. Some options of the SPARQL component are:
\begin{itemize}
Modified: trunk/src/dl-learner/org/dllearner/Info.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/Info.java 2009-02-17 19:23:47 UTC (rev 1609)
+++ trunk/src/dl-learner/org/dllearner/Info.java 2009-02-18 08:54:16 UTC (rev 1610)
@@ -3,6 +3,6 @@
package org.dllearner;
public class Info {
- public static final String build = "2008-10-13";
+ public static final String build = "2009-02-17";
}
\ No newline at end of file
Modified: trunk/src/dl-learner/org/dllearner/algorithms/refinement2/ROLearner2.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/algorithms/refinement2/ROLearner2.java 2009-02-17 19:23:47 UTC (rev 1609)
+++ trunk/src/dl-learner/org/dllearner/algorithms/refinement2/ROLearner2.java 2009-02-18 08:54:16 UTC (rev 1610)
@@ -149,10 +149,7 @@
private ExampleBasedNode startNode;
// solution protocol
- // TODO isn't solution found already true if solutions.size()>=0 ???
- @Deprecated
- private boolean solutionFound = false;
- private List<Description> solutions = new LinkedList<Description>();
+ private List<ExampleBasedNode> solutions = new LinkedList<ExampleBasedNode>();
// used refinement operator and heuristic (exchangeable)
private RhoDRDown operator;
@@ -300,7 +297,6 @@
candidates.clear();
candidatesStable.clear();
newCandidates.clear();
- solutionFound = false;
solutions.clear();
maxExecutionTimeAlreadyReached = false;
minExecutionTimeAlreadyReached = false;
@@ -493,7 +489,7 @@
boolean solution = checkSubtreePosOnly(bestNode);
if (solution) {
- solutions.add(bestNode.getConcept());
+ solutions.add(bestNode);
ExampleBasedNode bestChild = bestNode.getChildren().last();
System.out.println("solution: " + bestNode.getConcept());
System.out.println("maxPosOnlyExpansion: " + maxPosOnlyExpansion);
@@ -513,14 +509,11 @@
}
- // handle termination criteria
- handleStoppingConditions();
-
// Anzahl Schleifendurchläufe
loop++;
}// end while
- if (isSolutionFound()) {
+ if (solutions.size()>0) {
//if (solutionFound) {
int solutionLimit = 20;
// we do not need to print the best node if we display the top 20 solutions below anyway
@@ -531,11 +524,11 @@
int show = 1;
String manchester = "MANCHESTER:\n";
String kbSyntax = "KBSyntax:\n";
- for (Description c : solutions) {
- logger.info(show + ": " + c.toManchesterSyntaxString(baseURI, prefixes) + " (length " + c.getLength()
- + ", depth " + c.getDepth() + ")");
+ for (ExampleBasedNode c : solutions) {
+ logger.info(show + ": " + c.getConcept().toManchesterSyntaxString(baseURI, prefixes) + " (accuracy " + df.format(100*c.getAccuracy(nrOfPositiveExamples, nrOfNegativeExamples)) + "%, length " + c.getConcept().getLength()
+ + ", depth " + c.getConcept().getDepth() + ")");
// manchester += show + ": " + c.toManchesterSyntaxString(baseURI, prefixes) + "\n";
- kbSyntax += show + ": " + c.toKBSyntaxString() + "\n";
+ kbSyntax += show + ": " + c.getConcept().toKBSyntaxString() + "\n";
if (show >= solutionLimit) {
break;
}
@@ -862,8 +855,7 @@
} else {
// Lösung gefunden
if (quality >= 0 && quality <= allowedMisclassifications && !posOnly) {
- solutionFound = true;
- solutions.add(refinement);
+ solutions.add(newNode);
}
newCandidates.add(newNode);
@@ -1294,9 +1286,10 @@
logger.trace("ordered by generality (most special solutions first):");
SubsumptionComparator sc = new SubsumptionComparator(rs);
TreeSet<Description> solutionsOrderedBySubsumption = new TreeSet<Description>(sc);
- solutionsOrderedBySubsumption.addAll(solutions);
+// solutionsOrderedBySubsumption.addAll(solutions);
for (Description d : solutionsOrderedBySubsumption)
logger.trace("special: " + d);
+ throw new Error("implementation needs to be updated to show ordered solutions");
}
/*
* for (int j = 0; j < solutions.size(); j++) { Description d =
@@ -1438,94 +1431,6 @@
}
- /**
- * shows if a solution is found
- * @return
- */
- private boolean isSolutionFound(){
- if(this.solutionFound){
- return true;
- }else if(solutions.size()>0){
- return true;
- }
-
- return false;
-
-
- }
-
- @Deprecated
- private void handleStoppingConditions() {
- solutionFound = (guaranteeXgoodDescriptions());
- solutionFound = (minExecutionTimeReached() && solutionFound);
- if (maxExecutionTimeReached()) {
- stop();
- if (solutions.size() > 0)
- solutionFound = true;
- }
- if(!solutionFound && maxClassDescriptionTests != 0) {
- int conceptTests = conceptTestsReasoner + conceptTestsTooWeakList + conceptTestsOverlyGeneralList;
- solutionFound = (conceptTests >= maxClassDescriptionTests);
- }
- }
-
- @Deprecated
- private boolean guaranteeXgoodDescriptions() {
- if (guaranteeXgoodAlreadyReached)
- return true;
- if (solutions.size() > guaranteeXgoodDescriptions) {
- if (guaranteeXgoodDescriptions != 1) {
- logger.info("Minimum number (" + guaranteeXgoodDescriptions
- + ") of good descriptions reached, stopping now.");
- }
- guaranteeXgoodAlreadyReached = true;
- return true;
- } else
- return false;
-
- }
-
- @Deprecated
- private boolean maxExecutionTimeReached() {
- if (maxExecutionTimeInSeconds == 0)
- return false;
- if (maxExecutionTimeAlreadyReached)
- return true;
- long needed = System.currentTimeMillis() - this.runtime;
- long maxMilliSeconds = maxExecutionTimeInSeconds * 1000;
- if (maxMilliSeconds < needed) {
- logger.info("Maximum time (" + maxExecutionTimeInSeconds
- + " seconds) reached, stopping now...");
- maxExecutionTimeAlreadyReached = true;
- return true;
- } else
- return false;
-
- }
-
- /**
- * true if minExecutionTime reached
- *
- * @return true
- */
- @Deprecated
- private boolean minExecutionTimeReached() {
- if (minExecutionTimeAlreadyReached)
- return true;
- long needed = System.currentTimeMillis() - this.runtime;
- long minMilliSeconds = minExecutionTimeInSeconds * 1000;
- if (minMilliSeconds < needed) {
- if (minExecutionTimeInSeconds != 0) {
- logger.info("Minimum time (" + minExecutionTimeInSeconds
- + " seconds) reached, stopping when next solution is found");
- }
- minExecutionTimeAlreadyReached = true;
- return true;
- } else
- return false;
-
- }
-
public boolean isRunning() {
return isRunning;
}
Modified: trunk/src/dl-learner/org/dllearner/learningproblems/PosNegLPStandard.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/learningproblems/PosNegLPStandard.java 2009-02-17 19:23:47 UTC (rev 1609)
+++ trunk/src/dl-learner/org/dllearner/learningproblems/PosNegLPStandard.java 2009-02-18 08:54:16 UTC (rev 1610)
@@ -71,7 +71,7 @@
* @see org.dllearner.core.Component#getName()
*/
public static String getName() {
- return "two valued definition learning problem";
+ return "pos neg learning problem";
}
Modified: trunk/src/dl-learner/org/dllearner/learningproblems/PosOnlyLP.java
===================================================================
--- trunk/src/dl-learner/org/dllearner/learningproblems/PosOnlyLP.java 2009-02-17 19:23:47 UTC (rev 1609)
+++ trunk/src/dl-learner/org/dllearner/learningproblems/PosOnlyLP.java 2009-02-18 08:54:16 UTC (rev 1610)
@@ -83,7 +83,7 @@
}
public static String getName() {
- return "positive only definition learning problem";
+ return "pos only learning problem";
}
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
