Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3361] trunk
|
From: <jen...@us...> - 2011-11-02 14:07:37
|
Revision: 3361
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3361&view=rev
Author: jenslehmann
Date: 2011-11-02 14:07:30 +0000 (Wed, 02 Nov 2011)
Log Message:
-----------
- fixed a difficult to find bug in the refinement operator for string datatypes (related to an optimisation using property ranges)
- added support for single quoted strings in manchester parser (allows to use strings in conf files)
Modified Paths:
--------------
trunk/components-core/src/main/java/org/dllearner/algorithms/celoe/CELOE.java
trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParser.java
trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParserTokenManager.java
trunk/components-core/src/main/java/org/dllearner/parser/manchester.jj
trunk/components-core/src/main/java/org/dllearner/refinementoperators/RhoDRDown.java
trunk/components-core/src/test/java/org/dllearner/test/junit/ParserTests.java
trunk/examples/nlp2rdf/reuters_gold_vs_copper/learn.conf
Modified: trunk/components-core/src/main/java/org/dllearner/algorithms/celoe/CELOE.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/algorithms/celoe/CELOE.java 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/components-core/src/main/java/org/dllearner/algorithms/celoe/CELOE.java 2011-11-02 14:07:30 UTC (rev 3361)
@@ -453,7 +453,7 @@
// for(Description refinement : refinements) {
// System.out.println("refinement: " + refinement);
// }
-// if(loop % 100 == 0) {
+// if((loop+1) % 500 == 0) {
// System.out.println(getMinimumHorizontalExpansion() + " - " + getMaximumHorizontalExpansion());
// System.exit(0);
// }
Modified: trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParser.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParser.java 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParser.java 2011-11-02 14:07:30 UTC (rev 3361)
@@ -190,11 +190,6 @@
finally { jj_save(5, xla); }
}
- private boolean jj_3R_3() {
- if (jj_scan_token(STRING)) return true;
- return false;
- }
-
private boolean jj_3R_8() {
if (jj_scan_token(21)) return true;
if (jj_3R_2()) return true;
@@ -366,6 +361,11 @@
return false;
}
+ private boolean jj_3R_3() {
+ if (jj_scan_token(STRING)) return true;
+ return false;
+ }
+
/** Generated Token Manager. */
public ManchesterSyntaxParserTokenManager token_source;
SimpleCharStream jj_input_stream;
Modified: trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParserTokenManager.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParserTokenManager.java 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/components-core/src/main/java/org/dllearner/parser/ManchesterSyntaxParserTokenManager.java 2011-11-02 14:07:30 UTC (rev 3361)
@@ -18,11 +18,11 @@
{
case 0:
if ((active0 & 0x40000L) != 0L)
- return 31;
+ return 34;
return -1;
case 1:
if ((active0 & 0x40000L) != 0L)
- return 31;
+ return 34;
return -1;
default :
return -1;
@@ -83,7 +83,7 @@
if ((active0 & 0x20000L) != 0L)
return jjStopAtPos(1, 17);
else if ((active0 & 0x40000L) != 0L)
- return jjStartNfaWithStates_0(1, 18, 31);
+ return jjStartNfaWithStates_0(1, 18, 34);
break;
case 79:
return jjMoveStringLiteralDfa2_0(active0, 0xc00L);
@@ -220,7 +220,7 @@
private int jjMoveNfa_0(int startState, int curPos)
{
int startsAt = 0;
- jjnewStateCnt = 31;
+ jjnewStateCnt = 34;
int i = 1;
jjstateSet[0] = startState;
int kind = 0x7fffffff;
@@ -235,15 +235,6 @@
{
switch(jjstateSet[--i])
{
- case 31:
- if ((0xbfffffffffffdbffL & l) != 0L)
- jjCheckNAddTwoStates(4, 5);
- else if (curChar == 62)
- {
- if (kind > 20)
- kind = 20;
- }
- break;
case 0:
if ((0x3fe000000000000L & l) != 0L)
{
@@ -255,15 +246,26 @@
{
if (kind > 8)
kind = 8;
- jjCheckNAdd(28);
+ jjCheckNAdd(31);
}
else if (curChar == 47)
jjAddStates(3, 5);
else if (curChar == 60)
+ jjCheckNAddTwoStates(7, 8);
+ else if (curChar == 39)
jjCheckNAddTwoStates(4, 5);
else if (curChar == 34)
jjCheckNAddTwoStates(1, 2);
break;
+ case 34:
+ if ((0xbfffffffffffdbffL & l) != 0L)
+ jjCheckNAddTwoStates(7, 8);
+ else if (curChar == 62)
+ {
+ if (kind > 20)
+ kind = 20;
+ }
+ break;
case 1:
if ((0xfffffffbffffdbffL & l) != 0L)
jjCheckNAddTwoStates(1, 2);
@@ -273,131 +275,143 @@
kind = 19;
break;
case 3:
- if (curChar == 60)
+ if (curChar == 39)
jjCheckNAddTwoStates(4, 5);
break;
case 4:
- if ((0xbfffffffffffdbffL & l) != 0L)
+ if ((0xfffffffbffffdbffL & l) != 0L)
jjCheckNAddTwoStates(4, 5);
break;
case 5:
+ if (curChar == 39 && kind > 19)
+ kind = 19;
+ break;
+ case 6:
+ if (curChar == 60)
+ jjCheckNAddTwoStates(7, 8);
+ break;
+ case 7:
+ if ((0xbfffffffffffdbffL & l) != 0L)
+ jjCheckNAddTwoStates(7, 8);
+ break;
+ case 8:
if (curChar == 62 && kind > 20)
kind = 20;
break;
- case 6:
+ case 9:
if (curChar == 47)
jjAddStates(3, 5);
break;
- case 7:
+ case 10:
if (curChar == 47)
jjCheckNAddStates(6, 8);
break;
- case 8:
+ case 11:
if ((0xffffffffffffdbffL & l) != 0L)
jjCheckNAddStates(6, 8);
break;
- case 9:
+ case 12:
if ((0x2400L & l) != 0L && kind > 5)
kind = 5;
break;
- case 10:
+ case 13:
if (curChar == 10 && kind > 5)
kind = 5;
break;
- case 11:
+ case 14:
if (curChar == 13)
- jjstateSet[jjnewStateCnt++] = 10;
+ jjstateSet[jjnewStateCnt++] = 13;
break;
- case 12:
+ case 15:
if (curChar == 42)
- jjCheckNAddTwoStates(13, 14);
+ jjCheckNAddTwoStates(16, 17);
break;
- case 13:
+ case 16:
if ((0xfffffbffffffffffL & l) != 0L)
- jjCheckNAddTwoStates(13, 14);
+ jjCheckNAddTwoStates(16, 17);
break;
- case 14:
+ case 17:
if (curChar == 42)
jjCheckNAddStates(9, 11);
break;
- case 15:
+ case 18:
if ((0xffff7bffffffffffL & l) != 0L)
- jjCheckNAddTwoStates(16, 14);
+ jjCheckNAddTwoStates(19, 17);
break;
- case 16:
+ case 19:
if ((0xfffffbffffffffffL & l) != 0L)
- jjCheckNAddTwoStates(16, 14);
+ jjCheckNAddTwoStates(19, 17);
break;
- case 17:
+ case 20:
if (curChar == 47 && kind > 6)
kind = 6;
break;
- case 18:
+ case 21:
if (curChar == 42)
- jjstateSet[jjnewStateCnt++] = 12;
+ jjstateSet[jjnewStateCnt++] = 15;
break;
- case 19:
+ case 22:
if (curChar == 42)
- jjCheckNAddTwoStates(20, 21);
+ jjCheckNAddTwoStates(23, 24);
break;
- case 20:
+ case 23:
if ((0xfffffbffffffffffL & l) != 0L)
- jjCheckNAddTwoStates(20, 21);
+ jjCheckNAddTwoStates(23, 24);
break;
- case 21:
+ case 24:
if (curChar == 42)
jjCheckNAddStates(12, 14);
break;
- case 22:
+ case 25:
if ((0xffff7bffffffffffL & l) != 0L)
- jjCheckNAddTwoStates(23, 21);
+ jjCheckNAddTwoStates(26, 24);
break;
- case 23:
+ case 26:
if ((0xfffffbffffffffffL & l) != 0L)
- jjCheckNAddTwoStates(23, 21);
+ jjCheckNAddTwoStates(26, 24);
break;
- case 24:
+ case 27:
if (curChar == 47 && kind > 7)
kind = 7;
break;
- case 25:
+ case 28:
if ((0x3fe000000000000L & l) == 0L)
break;
if (kind > 8)
kind = 8;
jjCheckNAddStates(0, 2);
break;
- case 26:
+ case 29:
if ((0x3ff000000000000L & l) == 0L)
break;
if (kind > 8)
kind = 8;
- jjCheckNAdd(26);
+ jjCheckNAdd(29);
break;
- case 27:
+ case 30:
if ((0x3ff000000000000L & l) != 0L)
- jjCheckNAddTwoStates(27, 28);
+ jjCheckNAddTwoStates(30, 31);
break;
- case 28:
+ case 31:
if (curChar != 46)
break;
if (kind > 9)
kind = 9;
- jjCheckNAdd(29);
+ jjCheckNAdd(32);
break;
- case 29:
+ case 32:
if ((0x3ff000000000000L & l) == 0L)
break;
if (kind > 9)
kind = 9;
- jjCheckNAdd(29);
+ jjCheckNAdd(32);
break;
- case 30:
+ case 33:
if (curChar != 48)
break;
if (kind > 8)
kind = 8;
- jjCheckNAdd(28);
+ jjCheckNAdd(31);
break;
default : break;
}
@@ -410,32 +424,36 @@
{
switch(jjstateSet[--i])
{
- case 31:
- case 4:
+ case 34:
+ case 7:
if ((0xffffffffefffffffL & l) != 0L)
- jjCheckNAddTwoStates(4, 5);
+ jjCheckNAddTwoStates(7, 8);
break;
case 1:
if ((0xffffffffefffffffL & l) != 0L)
jjAddStates(15, 16);
break;
- case 8:
+ case 4:
+ if ((0xffffffffefffffffL & l) != 0L)
+ jjAddStates(17, 18);
+ break;
+ case 11:
jjAddStates(6, 8);
break;
- case 13:
- jjCheckNAddTwoStates(13, 14);
- break;
- case 15:
case 16:
- jjCheckNAddTwoStates(16, 14);
+ jjCheckNAddTwoStates(16, 17);
break;
- case 20:
- jjCheckNAddTwoStates(20, 21);
+ case 18:
+ case 19:
+ jjCheckNAddTwoStates(19, 17);
break;
- case 22:
case 23:
- jjCheckNAddTwoStates(23, 21);
+ jjCheckNAddTwoStates(23, 24);
break;
+ case 25:
+ case 26:
+ jjCheckNAddTwoStates(26, 24);
+ break;
default : break;
}
} while(i != startsAt);
@@ -448,37 +466,41 @@
{
switch(jjstateSet[--i])
{
- case 31:
- case 4:
+ case 34:
+ case 7:
if ((jjbitVec0[i2] & l2) != 0L)
- jjCheckNAddTwoStates(4, 5);
+ jjCheckNAddTwoStates(7, 8);
break;
case 1:
if ((jjbitVec0[i2] & l2) != 0L)
jjAddStates(15, 16);
break;
- case 8:
+ case 4:
if ((jjbitVec0[i2] & l2) != 0L)
- jjAddStates(6, 8);
+ jjAddStates(17, 18);
break;
- case 13:
+ case 11:
if ((jjbitVec0[i2] & l2) != 0L)
- jjCheckNAddTwoStates(13, 14);
+ jjAddStates(6, 8);
break;
- case 15:
case 16:
if ((jjbitVec0[i2] & l2) != 0L)
- jjCheckNAddTwoStates(16, 14);
+ jjCheckNAddTwoStates(16, 17);
break;
- case 20:
+ case 18:
+ case 19:
if ((jjbitVec0[i2] & l2) != 0L)
- jjCheckNAddTwoStates(20, 21);
+ jjCheckNAddTwoStates(19, 17);
break;
- case 22:
case 23:
if ((jjbitVec0[i2] & l2) != 0L)
- jjCheckNAddTwoStates(23, 21);
+ jjCheckNAddTwoStates(23, 24);
break;
+ case 25:
+ case 26:
+ if ((jjbitVec0[i2] & l2) != 0L)
+ jjCheckNAddTwoStates(26, 24);
+ break;
default : break;
}
} while(i != startsAt);
@@ -490,15 +512,15 @@
kind = 0x7fffffff;
}
++curPos;
- if ((i = jjnewStateCnt) == (startsAt = 31 - (jjnewStateCnt = startsAt)))
+ if ((i = jjnewStateCnt) == (startsAt = 34 - (jjnewStateCnt = startsAt)))
return curPos;
try { curChar = input_stream.readChar(); }
catch(java.io.IOException e) { return curPos; }
}
}
static final int[] jjnextStates = {
- 26, 27, 28, 7, 18, 19, 8, 9, 11, 14, 15, 17, 21, 22, 24, 1,
- 2,
+ 29, 30, 31, 10, 21, 22, 11, 12, 14, 17, 18, 20, 24, 25, 27, 1,
+ 2, 4, 5,
};
/** Token literal values. */
@@ -518,8 +540,8 @@
0xfeL,
};
protected SimpleCharStream input_stream;
-private final int[] jjrounds = new int[31];
-private final int[] jjstateSet = new int[62];
+private final int[] jjrounds = new int[34];
+private final int[] jjstateSet = new int[68];
protected char curChar;
/** Constructor. */
public ManchesterSyntaxParserTokenManager(SimpleCharStream stream){
@@ -546,7 +568,7 @@
{
int i;
jjround = 0x80000001;
- for (i = 31; i-- > 0;)
+ for (i = 34; i-- > 0;)
jjrounds[i] = 0x80000000;
}
Modified: trunk/components-core/src/main/java/org/dllearner/parser/manchester.jj
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/parser/manchester.jj 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/components-core/src/main/java/org/dllearner/parser/manchester.jj 2011-11-02 14:07:30 UTC (rev 3361)
@@ -64,7 +64,8 @@
| < NOT: "not" >
| < GE: ">=" >
| < LE: "<=" >
- | < STRING: "\"" (~["\"","\\","\n","\r"])* "\"" >
+ // support single quotes and double quotes
+ | < STRING: "\"" (~["\"","\\","\n","\r"])* "\"" | "'" (~["\"","\\","\n","\r"])* "'">
| < URI: "<" (~[">","\\","\n","\r"])* ">" >
}
Modified: trunk/components-core/src/main/java/org/dllearner/refinementoperators/RhoDRDown.java
===================================================================
--- trunk/components-core/src/main/java/org/dllearner/refinementoperators/RhoDRDown.java 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/components-core/src/main/java/org/dllearner/refinementoperators/RhoDRDown.java 2011-11-02 14:07:30 UTC (rev 3361)
@@ -145,6 +145,7 @@
private Map<NamedClass, Set<ObjectProperty>> appOP = new TreeMap<NamedClass, Set<ObjectProperty>>();
private Map<NamedClass, Set<DatatypeProperty>> appBD = new TreeMap<NamedClass, Set<DatatypeProperty>>();
private Map<NamedClass, Set<DatatypeProperty>> appDD = new TreeMap<NamedClass, Set<DatatypeProperty>>();
+ private Map<NamedClass, Set<DatatypeProperty>> appSD = new TreeMap<NamedClass, Set<DatatypeProperty>>();
// most general applicable properties
private Map<NamedClass,Set<ObjectProperty>> mgr = new TreeMap<NamedClass,Set<ObjectProperty>>();
@@ -838,6 +839,7 @@
private void computeTopRefinements(int maxLength, NamedClass domain) {
long topComputationTimeStartNs = System.nanoTime();
+// System.out.println("computing top refinements for " + domain + " up to length " + maxLength);
if(domain == null && m.size() == 0)
computeM();
@@ -957,6 +959,12 @@
topARefinementsLength.put(domain,maxLength);
topComputationTimeNs += System.nanoTime() - topComputationTimeStartNs;
+
+// if(domain == null) {
+// System.out.println("computed top refinements: " + topRefinementsCumulative.get(maxLength));
+// } else {
+// System.out.println("computed top refinements: " + topARefinementsCumulative.get(domain).get(maxLength));
+// }
}
// compute M_\top
@@ -1046,6 +1054,8 @@
}
m.put(4,m4);
+// System.out.println("m: " + m);
+
mComputationTimeNs += System.nanoTime() - mComputationTimeStartNs;
}
@@ -1183,7 +1193,7 @@
}
mA.get(nc).put(4,m4);
-// System.out.println(mA.get(nc));
+// System.out.println("m for " + nc + ": " + mA.get(nc));
mComputationTimeNs += System.nanoTime() - mComputationTimeStartNs;
}
@@ -1330,7 +1340,7 @@
private void computeMgsdRecursive(NamedClass domain, Set<DatatypeProperty> currProperties, Set<DatatypeProperty> mgsdTmp) {
for(DatatypeProperty prop : currProperties) {
- if(appDD.get(domain).contains(prop))
+ if(appSD.get(domain).contains(prop))
mgsdTmp.add(prop);
else
computeMgsdRecursive(domain, reasoner.getSubProperties(prop), mgsdTmp);
@@ -1372,7 +1382,19 @@
if(!isDisjoint(domain,d))
applicableDDPs.add(role);
}
- appDD.put(domain, applicableDDPs);
+ appDD.put(domain, applicableDDPs);
+
+ // string datatype properties
+ Set<DatatypeProperty> mostGeneralSDPs = reasoner.getStringDatatypeProperties();
+ Set<DatatypeProperty> applicableSDPs = new TreeSet<DatatypeProperty>();
+ for(DatatypeProperty role : mostGeneralSDPs) {
+// Description d = (NamedClass) rs.getDomain(role);
+ Description d = reasoner.getDomain(role);
+// System.out.println("domain: " + d);
+ if(!isDisjoint(domain,d))
+ applicableSDPs.add(role);
+ }
+ appSD.put(domain, applicableSDPs);
}
// returns true of the intersection contains elements disjoint
Modified: trunk/components-core/src/test/java/org/dllearner/test/junit/ParserTests.java
===================================================================
--- trunk/components-core/src/test/java/org/dllearner/test/junit/ParserTests.java 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/components-core/src/test/java/org/dllearner/test/junit/ParserTests.java 2011-11-02 14:07:30 UTC (rev 3361)
@@ -68,6 +68,8 @@
"(<http://example.com/prop> some (<http://example.com/class1> and <http://example.com/class2>))",
// another nested expression
"(<http://nlp2rdf.lod2.eu/schema/string/Document> and (<http://nlp2rdf.lod2.eu/schema/string/subStringTrans> some <http://www.w3.org/2002/07/owl#Thing>))",
+ // a test with a single quoted string
+ "(<http://nlp2rdf.lod2.eu/schema/string/Document> and (<http://nlp2rdf.lod2.eu/schema/string/subStringTrans> some ( <http://nlp2rdf.lod2.eu/schema/sso/lemma> value 'copper')))"
};
// loop through all test cases
Modified: trunk/examples/nlp2rdf/reuters_gold_vs_copper/learn.conf
===================================================================
--- trunk/examples/nlp2rdf/reuters_gold_vs_copper/learn.conf 2011-11-02 12:46:04 UTC (rev 3360)
+++ trunk/examples/nlp2rdf/reuters_gold_vs_copper/learn.conf 2011-11-02 14:07:30 UTC (rev 3361)
@@ -36,9 +36,10 @@
alg.type = "celoe"
// alg.writeSearchTree = true
alg.replaceSearchTree = true
-alg.maxExecutionTimeInSeconds = 100
+alg.maxExecutionTimeInSeconds = 10
alg.maxNrOfResults = 10
alg.startClass = "(<http://nlp2rdf.lod2.eu/schema/string/Document> and (<http://nlp2rdf.lod2.eu/schema/string/subStringTrans> some <http://www.w3.org/2002/07/owl#Thing>))"
+// solution:
+// alg.startClass = "(<http://nlp2rdf.lod2.eu/schema/string/Document> and (<http://nlp2rdf.lod2.eu/schema/string/subStringTrans> some ( <http://nlp2rdf.lod2.eu/schema/sso/lemma> value 'copper')))"
-
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
