Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3741] trunk/components-ext/src/main
|
From: <chr...@us...> - 2012-06-14 15:30:10
|
Revision: 3741
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3741&view=rev
Author: christinaunger
Date: 2012-06-14 15:30:01 +0000 (Thu, 14 Jun 2012)
Log Message:
-----------
[tbsl] basic update for regex descriptions, more filtering conditions on templates, and ASK -> SELECT hack
Modified Paths:
--------------
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/converter/DRS2SPARQL_Converter.java
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/ltag/parser/Preprocessor.java
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/sparql/Template.java
trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/Templator.java
trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex
trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/converter/DRS2SPARQL_Converter.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/converter/DRS2SPARQL_Converter.java 2012-06-14 14:54:54 UTC (rev 3740)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/converter/DRS2SPARQL_Converter.java 2012-06-14 15:30:01 UTC (rev 3741)
@@ -1,9 +1,6 @@
package org.dllearner.algorithm.tbsl.converter;
-import java.util.ArrayList;
-import java.util.HashSet;
-import java.util.List;
-import java.util.Set;
+import java.util.*;
import org.dllearner.algorithm.tbsl.sem.drs.Complex_DRS_Condition;
import org.dllearner.algorithm.tbsl.sem.drs.DRS;
@@ -30,8 +27,8 @@
public class DRS2SPARQL_Converter {
- // suppresses console output
- private boolean silent = true;
+ private boolean silent = true; // suppresses console output
+ private boolean oxford = true;
List<Slot> slots;
Template template;
List<Integer> usedInts;
@@ -100,7 +97,8 @@
if (!restructureEmpty(drs)) {
return null;
}
-// System.out.println("--- DRS (after) : " + drs); // DEBUG
+ replaceRegextoken(drs);
+ // System.out.println("--- DRS (after) : " + drs); // DEBUG
for (DiscourseReferent referent : drs.collectDRs()) {
if (referent.isMarked()) {
@@ -318,6 +316,21 @@
new SPARQL_Term(simple.getArguments().get(1).getValue(),true),
SPARQL_PairType.REGEX)));
}
+ else if (predicate.equals("regextoken")) {
+ String arg = simple.getArguments().get(1).getValue();
+ String regex = null;
+ for (Slot slot : slots) {
+ if (slot.getAnchor().equals(arg)) {
+ if (!slot.getWords().isEmpty()) regex = slot.getWords().get(0);
+ }
+ }
+ if (regex != null) {
+ query.addFilter(new SPARQL_Filter(new SPARQL_Pair(
+ new SPARQL_Term(simple.getArguments().get(0).getValue(),false),
+ new SPARQL_Term("'"+regex+"'",false),
+ SPARQL_PairType.REGEX)));
+ }
+ }
else {
if (arity == 1) {
SPARQL_Term term = new SPARQL_Term(simple.getArguments().get(0).getValue(),false);term.setIsVariable(true);
@@ -333,6 +346,45 @@
}
}
}
+
+ // TODO this is a hack in order to avoid ASK queries if DP is parsed
+ if (oxford) {
+ Hashtable<String,Integer> vs = new Hashtable<String,Integer>();
+ String v1; String v2;
+ for (SPARQL_Triple c : query.getConditions()) {
+ v1 = c.getVariable().toString().replace("?","");
+ v2 = c.getValue().toString().replace("?","");
+ // is it a slot variable?
+ boolean v1isSlotVar = false;
+ boolean v2isSlotVar = false;
+ for (Slot s : slots) {
+ if (s.getAnchor().equals(v1)) v1isSlotVar = true;
+ if (s.getAnchor().equals(v2)) v2isSlotVar = true;
+ }
+ if (!v1isSlotVar && !v1.matches("[0..9]+") && !v1.contains("count")) {
+ if (vs.containsKey(v1)) vs.put(v1,vs.get(v1)+1);
+ else vs.put(v1,1);
+ }
+ if (!v2isSlotVar && !v2.matches("[0..9]+") && !v2.contains("count")) {
+ if (vs.containsKey(v2)) vs.put(v2,vs.get(v2)+1);
+ else vs.put(v2,1);
+ }
+ }
+
+ int max = 0; String maxvar = null;
+ for (String var : vs.keySet()) {
+ if (vs.get(var) > max) {
+ max = vs.get(var);
+ maxvar = var;
+ }
+ }
+ if (maxvar != null) {
+ SPARQL_Term term = new SPARQL_Term(maxvar);
+ term.setIsVariable(true);
+ query.addSelTerm(term);
+ }
+ }
+
return query;
}
@@ -340,41 +392,41 @@
Set<Simple_DRS_Condition> equalsConditions = new HashSet<Simple_DRS_Condition>();
for (Simple_DRS_Condition c : drs.getAllSimpleConditions()) {
- if(c.getPredicate().equals("equal")) {
- equalsConditions.add(c);
- }
+ if(c.getPredicate().equals("equal")) equalsConditions.add(c);
}
DiscourseReferent firstArg;
DiscourseReferent secondArg;
boolean firstIsURI;
boolean secondIsURI;
+ boolean firstIsInt;
+ boolean secondIsInt;
for (Simple_DRS_Condition c : equalsConditions) {
- firstArg = c.getArguments().get(0);
+ firstArg = c.getArguments().get(0);
secondArg = c.getArguments().get(1);
firstIsURI = isUri(firstArg.getValue());
secondIsURI = isUri(secondArg.getValue());
+ firstIsInt = firstArg.getValue().matches("[0..9]+");
+ secondIsInt = secondArg.getValue().matches("[0..9]+");
- boolean oneArgIsInt = firstArg.toString().matches("[0..9]") || secondArg.toString().matches("[0..9]");
-
drs.removeCondition(c);
- if (firstIsURI) {
- drs.replaceEqualRef(secondArg, firstArg, false);
+ if (firstIsURI || firstIsInt) {
+ drs.replaceEqualRef(secondArg, firstArg, true);
for (Slot s : slots) {
if (s.getAnchor().equals(secondArg.getValue())) {
s.setAnchor(firstArg.getValue());
}
}
- } else if (secondIsURI) {
- drs.replaceEqualRef(firstArg, secondArg, false);
+ } else if (secondIsURI || secondIsInt) {
+ drs.replaceEqualRef(firstArg, secondArg, true);
for (Slot s : slots) {
if (s.getAnchor().equals(firstArg.getValue())) {
s.setAnchor(secondArg.getValue());
}
}
- } else if (!oneArgIsInt) {
+ } else {
drs.replaceEqualRef(firstArg, secondArg, false);
for (Slot s : slots) {
if (s.getAnchor().equals(firstArg.getValue())) {
@@ -396,6 +448,55 @@
}
}
+ private void replaceRegextoken(DRS drs) {
+
+ Set<Simple_DRS_Condition> cs = new HashSet<Simple_DRS_Condition>();
+ for (Simple_DRS_Condition c : drs.getAllSimpleConditions()) {
+ if(c.getPredicate().equals("regextoken")) cs.add(c);
+ }
+
+ String var;
+ String newvar;
+ String regex = "";
+ String[] forbidden = {"regextoken","regex","count","minimum","maximum","greater","less","greaterorequal","lessorequal","equal","sum"};
+ Set<Simple_DRS_Condition> used = new HashSet<Simple_DRS_Condition>();
+
+ for (Simple_DRS_Condition c : cs) {
+ var = c.getArguments().get(1).getValue();
+ newvar = c.getArguments().get(0).getValue();
+ for (Simple_DRS_Condition cond : drs.getAllSimpleConditions()) {
+ boolean takeit = false;
+ for (DiscourseReferent dr : cond.getArguments()) {
+ if (dr.getValue().equals(var)) {
+ takeit = true;
+ for (String f : forbidden) if (f.equals(cond.getPredicate())) takeit= false;
+ }
+ }
+ if (takeit) {
+ regex += cond.getPredicate().replace("SLOT","") + " ";
+ used.add(cond);
+ }
+ else {
+ for (DiscourseReferent dr : cond.getArguments()) {
+ if (dr.getValue().equals(var)) dr.setValue(newvar);
+ }
+ }
+ }
+ if (!regex.isEmpty()) {
+ c.getArguments().remove(1);
+ c.getArguments().add(new DiscourseReferent("'"+regex.trim()+"'"));
+ c.setPredicate("regex");
+ }
+ for (Slot s : slots) {
+ if (s.getWords().contains(var)) {
+ s.getWords().remove(var);
+ s.getWords().add(newvar);
+ }
+ }
+ }
+ for (Simple_DRS_Condition cond : used) drs.removeCondition(cond);
+ }
+
private boolean restructureEmpty(DRS drs) {
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/ltag/parser/Preprocessor.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/ltag/parser/Preprocessor.java 2012-06-14 14:54:54 UTC (rev 3740)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/ltag/parser/Preprocessor.java 2012-06-14 15:30:01 UTC (rev 3741)
@@ -18,6 +18,9 @@
static final String[] genericReplacements = { "[!?.,;]", "" };
static final String[] englishReplacements = { "don't", "do not", "doesn't", "does not" };
+ static final String[] hackReplacements = { " 1 "," one "," 2 "," two "," 3 "," three "," 4 "," four "," 5 "," five "," 6 "," six "," 7 "," seven ",
+ " 8 "," eight "," 9 "," nine "," 10 "," ten "," 11 "," eleven "," 12 "," twelve "," 13 "," thirteen "," 14 "," fourteen "," 15 "," fifteen ",
+ " 16 "," sixteen "," 17 "," seventeen "," 18 "," eighteen "," 19 "," nineteen "," 20 "," twenty "};
static boolean USE_NER;
static boolean VERBOSE;
static NER ner;
@@ -49,7 +52,9 @@
replacements.addAll(Arrays.asList(repl));
replacements.addAll(Arrays.asList(englishReplacements));
replacements.addAll(Arrays.asList(genericReplacements));
+ replacements.addAll(Arrays.asList(hackReplacements));
+ s = s.replaceAll(",\\s"," and ");
for (int i = 0; i < replacements.size(); i += 2) {
s = s.replaceAll(replacements.get(i), replacements.get(i + 1));
}
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/sparql/Template.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/sparql/Template.java 2012-06-14 14:54:54 UTC (rev 3740)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/sparql/Template.java 2012-06-14 15:30:01 UTC (rev 3741)
@@ -33,7 +33,28 @@
public Template checkandrefine() {
Set<Slot> argslots = new HashSet<Slot>();
- for (Slot slot : slots) if (slot.anchor.equals("SLOT_arg")) argslots.add(slot);
+ for (Slot slot : slots) if (slot.anchor.equals("SLOT_arg")) {
+ String var = slot.words.get(0);
+ // check for clash (v=LITERAL && v=RESOURCE)
+ for (Slot s : argslots) {
+ if (s.words.get(0).equals(slot.words.get(0)) && !s.type.equals(slot.type))
+ return null;
+ }
+ // check for clash (v=LITERAL && p(...,v)=OBJECTPROPERTY) || (v=RESOURCE && p(...,v)=DATATYPEPROPERTY)
+ SlotType clashing = null;
+ if (slot.type.equals(SlotType.LITERAL)) clashing = SlotType.OBJECTPROPERTY;
+ else if (slot.type.equals(SlotType.RESOURCE)) clashing = SlotType.DATATYPEPROPERTY;
+ for (Slot s : slots) {
+ if (clashing != null && s.type.equals(clashing)) {
+ for (SPARQL_Triple triple : query.conditions) {
+ if (triple.property.toString().equals("?"+s.anchor)) {
+ if (triple.value.toString().equals("?"+var)) return null;
+ }
+ }
+ }
+ }
+ argslots.add(slot);
+ }
for (Slot slot : slots) {
// check for clashes
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/Templator.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/Templator.java 2012-06-14 14:54:54 UTC (rev 3740)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/Templator.java 2012-06-14 15:30:01 UTC (rev 3741)
@@ -32,9 +32,9 @@
public class Templator {
private static final Logger logger = Logger.getLogger(Templator.class);
+
+ String[] GRAMMAR_FILES = {"tbsl/lexicon/english.lex","tbsl/lexicon/english_oxford.lex"};
- String[] GRAMMAR_FILES = {"tbsl/lexicon/english.lex","tbsl/lexicon/english_oxford.lex"};
-
private String[] noun = {"NN","NNS","NNP","NNPS","NPREP","JJNN","JJNPREP"};
private String[] adjective = {"JJ","JJR","JJS","JJH"};
private String[] verb = {"VB","VBD","VBG","VBN","VBP","VBZ","PASSIVE","PASSPART","VPASS","VPASSIN","GERUNDIN","VPREP","WHEN","WHERE"};
@@ -49,8 +49,8 @@
WordNet wordnet;
LingPipeLemmatizer lem = new LingPipeLemmatizer();
- DUDE2UDRS_Converter d2u = new DUDE2UDRS_Converter();
- DRS2SPARQL_Converter d2s = new DRS2SPARQL_Converter();
+ DUDE2UDRS_Converter d2u = new DUDE2UDRS_Converter();
+ DRS2SPARQL_Converter d2s = new DRS2SPARQL_Converter();
boolean ONE_SCOPE_ONLY = true;
boolean UNTAGGED_INPUT = true;
Modified: trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex
===================================================================
--- trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex 2012-06-14 14:54:54 UTC (rev 3740)
+++ trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex 2012-06-14 15:30:01 UTC (rev 3741)
@@ -172,6 +172,16 @@
eight || (NP NUM:'eight' NP*) || <x,l1,<e,t>,[l1:[x|count(x,8)]],[],[],[ SLOT_arg/RESOURCE/x ]>
nine || (NP NUM:'nine' NP*) || <x,l1,<e,t>,[l1:[x|count(x,9)]],[],[],[ SLOT_arg/RESOURCE/x ]>
ten || (NP NUM:'ten' NP*) || <x,l1,<e,t>,[l1:[x|count(x,10)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ eleven || (NP NUM:'one' NP*) || <x,l1,<e,t>,[l1:[x|count(x,11)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ twelve || (NP NUM:'two' NP*) || <x,l1,<e,t>,[l1:[x|count(x,12)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ thirteen || (NP NUM:'thirteen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,13)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ fourteen || (NP NUM:'fourteen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,14)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ fifteen || (NP NUM:'fifteen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,15)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ sixteen || (NP NUM:'sixteen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,16)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ seventeen || (NP NUM:'seventeen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,17)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ eighteen || (NP NUM:'eighteen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,18)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ nineteen || (NP NUM:'nineteen' NP*) || <x,l1,<e,t>,[l1:[x|count(x,19)]],[],[],[ SLOT_arg/RESOURCE/x ]>
+ twenty || (NP NUM:'twenty' NP*) || <x,l1,<e,t>,[l1:[x|count(x,20)]],[],[],[ SLOT_arg/RESOURCE/x ]>
one || (NUM NUM:'one') || <x,l1,e,[l1:[x|equal(x,1)]],[],[],[ SLOT_arg/LITERAL/x ]>
two || (NUM NUM:'two') || <x,l1,e,[l1:[x|equal(x,2)]],[],[],[ SLOT_arg/LITERAL/x ]>
@@ -183,4 +193,13 @@
eight || (NUM NUM:'eight') || <x,l1,e,[l1:[x|equal(x,8)]],[],[],[ SLOT_arg/LITERAL/x ]>
nine || (NUM NUM:'nine') || <x,l1,e,[l1:[x|equal(x,9)]],[],[],[ SLOT_arg/LITERAL/x ]>
ten || (NUM NUM:'ten') || <x,l1,e,[l1:[x|equal(x,10)]],[],[],[ SLOT_arg/LITERAL/x ]>
-
+ eleven || (NUM NUM:'eleven') || <x,l1,e,[l1:[x|equal(x,11)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ twelve || (NUM NUM:'twelve') || <x,l1,e,[l1:[x|equal(x,12)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ thirteen || (NUM NUM:'thirteen') || <x,l1,e,[l1:[x|equal(x,13)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ fourteen || (NUM NUM:'fourteen') || <x,l1,e,[l1:[x|equal(x,14)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ fifteen || (NUM NUM:'fifteen') || <x,l1,e,[l1:[x|equal(x,15)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ sixteen || (NUM NUM:'sixteen') || <x,l1,e,[l1:[x|equal(x,16)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ seventeen || (NUM NUM:'seventeen') || <x,l1,e,[l1:[x|equal(x,17)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ eighteen || (NUM NUM:'eighteen') || <x,l1,e,[l1:[x|equal(x,18)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ nineteen || (NUM NUM:'nineteen') || <x,l1,e,[l1:[x|equal(x,19)]],[],[],[ SLOT_arg/LITERAL/x ]>
+ twenty || (NUM NUM:'twenty') || <x,l1,e,[l1:[x|equal(x,20)]],[],[],[ SLOT_arg/LITERAL/x ]>
Modified: trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex
===================================================================
--- trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex 2012-06-14 14:54:54 UTC (rev 3740)
+++ trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex 2012-06-14 15:30:01 UTC (rev 3741)
@@ -10,7 +10,7 @@
for less than . pounds || (NP NP* (PP P:'for' (NP NUM[num] N:'pounds'))) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,y,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
from . to . pounds || (NP NP* (PP P:'from' NUM[num1] P:'to' NUM[num2] N:'pounds')) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), greaterorequal(y,n1), lessorequal(y,n2) ] ], [ (l2,n1,num1,e),(l3,n2,num2,e) ], [ l2=l1,l3=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
- with || (NP NP* (PP P:'with' DP[dp])) || <x,l1,<e,t>, [ l1:[ | empty(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[]>
+ with || (NP NP* (PP P:'with' DP[dp])) || <x,l1,<e,t>, [ l1:[ | empty(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[]> ;; <x,l1,<e,t>, [ l1:[ | SLOT_description(x,z), regextoken(z,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_description/DATATYPEPROPERTY/description, SLOT_arg/LITERAL/z ]>
square meters || (DP N:'square' N:'meters') || <x,l1,<<e,t>,t>>, [l1:[ | SLOT_size(x,y) ]], [],[],[SLOT_size/DATATYPEPROPERTY/size ]>
// MONTHS
@@ -26,4 +26,4 @@
september || (DP DP:'september') || <x,l1,<<e,t>,t>, [ l1:[ x | xsd:month(x,9) ]], [],[],[]>
october || (DP DP:'october') || <x,l1,<<e,t>,t>, [ l1:[ x | xsd:month(x,10) ]], [],[],[]>
november || (DP DP:'november') || <x,l1,<<e,t>,t>, [ l1:[ x | xsd:month(x,11) ]], [],[],[]>
- december || (DP DP:'december') || <x,l1,<<e,t>,t>, [ l1:[ x | xsd:month(x,12) ]], [],[],[]>
\ No newline at end of file
+ december || (DP DP:'december') || <x,l1,<<e,t>,t>, [ l1:[ x | xsd:month(x,12) ]], [],[],[]>
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
