Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3759] trunk/components-ext/src/main
|
From: <chr...@us...> - 2012-06-18 10:50:56
|
Revision: 3759
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3759&view=rev
Author: christinaunger
Date: 2012-06-18 10:50:46 +0000 (Mon, 18 Jun 2012)
Log Message:
-----------
[tbsl] added proper support for "[PP] and [PP]"
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/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 2012-06-18 07:44:24 UTC (rev 3758)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/converter/DRS2SPARQL_Converter.java 2012-06-18 10:50:46 UTC (rev 3759)
@@ -76,9 +76,7 @@
slots = ls;
Query q = convert(drs, new Query(), false);
- if (q == null) {
- return null;
- }
+ if (q == null) return null;
q.setPrefixes(prefixes);
template.setQuery(q);
@@ -87,12 +85,13 @@
System.out.println("... done");
}
+ usedInts = new ArrayList<Integer>();
return template;
}
private Query convert(DRS drs, Query query, boolean negate) {
-// System.out.println("--- DRS (before): " + drs); // DEBUG
+// System.out.println("\n--- DRS (before): " + drs); // DEBUG
redundantEqualRenaming(drs);
if (!restructureEmpty(drs)) {
return null;
@@ -474,10 +473,10 @@
}
}
if (takeit) {
- regex += cond.getPredicate().replace("SLOT_","") + " ";
+ regex += cond.getPredicate().replace("SLOT","").replaceAll("_"," ");
used.add(cond);
}
- else {
+ else if (!cond.getPredicate().equals("regextoken")) {
for (DiscourseReferent dr : cond.getArguments()) {
if (dr.getValue().equals(var)) dr.setValue(newvar);
}
@@ -488,7 +487,7 @@
c.getArguments().add(new DiscourseReferent("'"+regex.trim()+"'"));
c.setPredicate("regex");
}
- else {} // TODO should not happen!
+ else { used.add(c); } // TODO should not happen!
for (Slot s : slots) {
if (s.getWords().contains(var)) {
s.getWords().remove(var);
@@ -497,6 +496,41 @@
}
}
for (Simple_DRS_Condition cond : used) drs.removeCondition(cond);
+
+ // postprocessing
+ boolean success = false;
+ Set<Simple_DRS_Condition> oldconds = new HashSet<Simple_DRS_Condition>();
+ for (Simple_DRS_Condition c : drs.getAllSimpleConditions()) {
+ String d = "";
+ String d2 = "";
+ String newregex = "";
+ if (c.getPredicate().equals("SLOT_description")) {
+ d = c.getArguments().get(0).getValue();
+ d2 = c.getArguments().get(1).getValue();
+ for (Simple_DRS_Condition cond : drs.getAllSimpleConditions()) {
+ if (cond.getPredicate().equals("SLOT_description") && cond.getArguments().get(1).getValue().equals(d)) {
+ oldconds.add(c);
+ success = true;
+ break;
+ }
+ }
+ }
+ if (success) {
+ for (Simple_DRS_Condition cond : drs.getAllSimpleConditions()) {
+ if (cond.getPredicate().equals("regex") &&
+ (cond.getArguments().get(0).getValue().equals(d) || cond.getArguments().get(0).getValue().equals(d2))) {
+ newregex += cond.getArguments().get(1).getValue().replaceAll("'","").replaceAll("_"," ").trim() + " ";
+ oldconds.add(cond);
+ }
+ }
+ for (Simple_DRS_Condition cond : oldconds) drs.removeCondition(cond);
+ List<DiscourseReferent> newrefs = new ArrayList<DiscourseReferent>();
+ newrefs.add(new DiscourseReferent(d));
+ newrefs.add(new DiscourseReferent("'"+newregex.trim()+"'"));
+ drs.addCondition(new Simple_DRS_Condition("regex",newrefs));
+ break;
+ }
+ }
}
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 2012-06-18 07:44:24 UTC (rev 3758)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/Templator.java 2012-06-18 10:50:46 UTC (rev 3759)
@@ -57,7 +57,7 @@
boolean UNTAGGED_INPUT = true;
boolean USE_NER = false;
boolean USE_WORDNET = true;
- boolean VERBOSE = true;
+ boolean VERBOSE = false;
private String taggedInput;
Modified: trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex
===================================================================
--- trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex 2012-06-18 07:44:24 UTC (rev 3758)
+++ trunk/components-ext/src/main/resources/tbsl/lexicon/english.lex 2012-06-18 10:50:46 UTC (rev 3759)
@@ -142,10 +142,14 @@
and || (S S* CC:'and' S[s]) || <x,l1,t,[l1:[|]],[(l2,y,s,t)],[l1=l2],[]>
and || (DP DP* CC:'and' DP[dp]) || <x,l1,<<e,t>,t>,[l1:[|]],[(l2,y,dp,<<e,t>,t>)],[l1=l2],[]>
and || (NP NP* CC:'and' NP[np]) || <x,l1,<e,t>,[l1:[|x=y]],[(l2,y,np,<e,t>)],[l1=l2],[]>
+ and || (PP PP* CC:'and' PP[pp]) || <x,l1,<e,t>,[l1:[|x=y]],[(l2,y,pp,<e,t>)],[l1=l2],[]>
+ and || (ADJ ADJ* CC:'and' ADJ[adj]) || <x,l1,<e,t>,[l1:[|]],[(l2,y,adj,<e,t>)],[l1=l2],[]>
but || (S S* CC:'but' S[s]) || <x,l1,t,[l1:[|]],[(l2,y,s,t)],[l1=l2],[]>
but || (DP DP* CC:'but' DP[dp]) || <x,l1,<<e,t>,t>,[l1:[|]],[(l2,y,dp,<<e,t>,t>)],[l1=l2],[]>
but || (NP NP* CC:'but' NP[np]) || <x,l1,<e,t>,[l1:[|x=y]],[(l2,y,np,<e,t>)],[l1=l2],[]>
+ but || (PP PP* CC:'but' PP[pp]) || <x,l1,<e,t>,[l1:[|x=y]],[(l2,y,pp,<e,t>)],[l1=l2],[]>
+ but || (ADJ ADJ* CC:'but' ADJ[adj]) || <x,l1,<e,t>,[l1:[|]],[(l2,y,adj,<e,t>)],[l1=l2],[]>
as well as || (NP NP* CC:'as' CC:'well' CC:'as' NP[np]) || <x,l1,<e,t>,[l1:[|]],[(l2,y,np,<e,t>)],[l1=l2],[]>
Modified: trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex
===================================================================
--- trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex 2012-06-18 07:44:24 UTC (rev 3758)
+++ trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex 2012-06-18 10:50:46 UTC (rev 3759)
@@ -2,26 +2,37 @@
// PREPOSITIONS
close to || (NP NP* (PP P:'close' P:'to' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_closeto(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_closeto/OBJECTPROPERTY/near ]>
+ close to || (PP P:'close' P:'to' DP[dp]) || <x,l1,<e,t>, [ l1:[ | SLOT_closeto(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_closeto/OBJECTPROPERTY/near ]>
near || (NP NP* (PP P:'near' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/near ]>
+ near || (PP P:'near' DP[dp]) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/near ]>
nearby || (NP NP* (PP P:'nearby' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_nearby(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_nearby/OBJECTPROPERTY/near ]>
+ nearby || (PP P:'nearby' DP[dp]) || <x,l1,<e,t>, [ l1:[ | SLOT_nearby(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_nearby/OBJECTPROPERTY/near ]>
within walking distance from || (NP NP* (PP P:'within' (NP N:'walking' N:'distance' P:'from' DP[dp]))) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
+ within walking distance from || (PP P:'within' (NP N:'walking' N:'distance' P:'from' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
within minutes of || (NP NP* (PP P:'within' (NP N:'minutes' P:'of' DP[dp]))) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
+ within minutes of || (PP P:'within' (NP N:'minutes' P:'of' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
in walking distance from || (NP NP* (PP P:'in' (NP N:'walking' N:'distance' P:'from' DP[dp]))) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
+ in walking distance from || (PP P:'in' (NP N:'walking' N:'distance' P:'from' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
at walking distance from || (NP NP* (PP P:'at' (NP N:'walking' N:'distance' P:'from' DP[dp]))) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
- in || (NP NP* (PP P:'in' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_location(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_location/PROPERTY/location^city^postal_code^address^street ]>
- since || (NP NP* (PP P:'since' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_since(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_since/PROPERTY/since ]>
+ at walking distance from || (PP P:'at' (NP N:'walking' N:'distance' P:'from' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_near(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_near/OBJECTPROPERTY/at_walking_distance ]>
+ in || (NP NP* (PP P:'in' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_location(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_location/PROPERTY/location^city^postal_code^address^street ]>
+ since || (NP NP* (PP P:'since' DP[dp])) || <x,l1,<e,t>, [ l1:[ | SLOT_since(x,y) ] ], [ (l2,y,dp,<<e,t>,t>) ], [ l2=l1 ],[ SLOT_since/PROPERTY/since ]>
+
for .+ pounds || (NP NP* (PP P:'for' (NP NUM[num] N:'pounds'))) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y) ] ], [ (l2,y,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
for more than .+ pounds || (NP NP* (PP P:'for' DET:'more' DET:'than' (NP NUM[num] N:'pounds'))) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), greater(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
for less than .+ pounds || (NP NP* (PP P:'for' DET:'less' DET:'than' (NP NUM[num] N:'pounds'))) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
for less than || (NP NP* (PP P:'for' DET:'less' DET:'than' NUM[num])) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
for more than || (NP NP* (PP P:'for' DET:'more' DET:'than' NUM[num])) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), greater(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
cheaper than .+ pounds || (NP NP* (ADJ ADJ:'cheaper' DET:'than' (NP NUM[num] N:'pounds'))) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
+ cheaper than .+ pounds || (ADJ ADJ:'cheaper' DET:'than' (NP NUM[num] N:'pounds')) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
below .+ pounds || (NP NP* (PP P:'below' (NP NUM[num] N:'pounds'))) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,z,num,e) ], [ l2=l1 ],[ SLOT_price/DATATYPEPROPERTY/price ]>
+ below .+ pounds || (PP P:'below' (NP NUM[num] N:'pounds')) || <x,l1,<e,t>, [ l1:[ | SLOT_price(x,y), less(y,z) ] ], [ (l2,z,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 ]>
between .+ and .+ pounds || (NP NP* (PP P:'between' NUM[num1] P:'and' 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 ],[]> ;; <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 ]>
+ with || (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
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
