Menu
▾
▴
[DL-Learner SVN] SF.net SVN: dl-learner:[3750] trunk/components-ext/src/main
|
From: <chr...@us...> - 2012-06-15 14:46:56
|
Revision: 3750
http://dl-learner.svn.sourceforge.net/dl-learner/?rev=3750&view=rev
Author: christinaunger
Date: 2012-06-15 14:46:46 +0000 (Fri, 15 Jun 2012)
Log Message:
-----------
[tbsl] a few more lexical extensions
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/templator/SlotBuilder.java
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 2012-06-15 14:20:33 UTC (rev 3749)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/converter/DRS2SPARQL_Converter.java 2012-06-15 14:46:46 UTC (rev 3750)
@@ -470,7 +470,7 @@
for (DiscourseReferent dr : cond.getArguments()) {
if (dr.getValue().equals(var)) {
takeit = true;
- for (String f : forbidden) if (f.equals(cond.getPredicate())) takeit= false;
+ for (String f : forbidden) if (cond.getPredicate().contains(f)) takeit= false;
}
}
if (takeit) {
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 2012-06-15 14:20:33 UTC (rev 3749)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/ltag/parser/Preprocessor.java 2012-06-15 14:46:46 UTC (rev 3750)
@@ -96,6 +96,7 @@
Pattern whenPattern = Pattern.compile("\\A(when/WRB\\s(.+\\s)(\\w+)/((V[A-Z]+)|(PASS[A-Z]+)))");
Pattern wherePattern = Pattern.compile("\\A(where/WRB\\s(.+\\s)(\\w+)/((V[A-Z]+)|(PASS[A-Z]+)))");
Pattern adjsPattern = Pattern.compile("((\\w+)/JJ.(\\w+)/JJ)");
+ Pattern adjnnpPattern = Pattern.compile("((\\w+)(?<!many)/JJ.(\\w+)/NNP(S)?)");
Pattern adjnounPattern = Pattern.compile("((\\w+)(?<!many)/JJ.(\\w+)/NN(S)?)");
Pattern adjnprepPattern = Pattern.compile("((\\w+)(?<!many)/JJ.(\\w+)/NPREP)");
@@ -217,7 +218,12 @@
while (m.find()) {
if (VERBOSE) logger.trace("Replacing " + m.group(1) + " by " + m.group(2)+"_"+m.group(3)+"/JJ");
condensedstring = condensedstring.replaceFirst(m.group(1),m.group(2)+"_"+m.group(3)+"/JJ");
- }
+ }
+ m = adjnnpPattern.matcher(condensedstring);
+ while (m.find()) {
+ if (VERBOSE) logger.trace("Replacing " + m.group(1) + " by " + m.group(2)+"_"+m.group(3)+"/NNP");
+ condensedstring = condensedstring.replaceFirst(m.group(1),m.group(2)+"_"+m.group(3)+"/NNP");
+ }
m = adjnounPattern.matcher(condensedstring);
while (m.find()) {
if (VERBOSE) logger.trace("Replacing " + m.group(1) + " by " + m.group(2)+"_"+m.group(3)+"/JJNN");
Modified: trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/SlotBuilder.java
===================================================================
--- trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/SlotBuilder.java 2012-06-15 14:20:33 UTC (rev 3749)
+++ trunk/components-ext/src/main/java/org/dllearner/algorithm/tbsl/templator/SlotBuilder.java 2012-06-15 14:46:46 UTC (rev 3750)
@@ -79,14 +79,16 @@
}
// treetoken
String treetoken = "N:'" + token.toLowerCase() + "'";
- if (token.trim().contains(" ")) {
- String[] tokenParts = token.split(" ");
- treetoken = "";
- for (String t : tokenParts) {
- treetoken += " N:'" + t.toLowerCase() + "'";
- }
- treetoken = treetoken.trim();
+ String[] tokenParts;
+ if (token.trim().contains(" ")) tokenParts = token.split(" ");
+// else if (token.contains("_")) tokenParts = token.split("_");
+ else { tokenParts = new String[1]; tokenParts[0] = token; }
+ treetoken = "";
+ for (String t : tokenParts) {
+ treetoken += " N:'" + t.toLowerCase() + "'";
}
+ treetoken = treetoken.trim();
+
//
if (pos.equals("NN") || pos.equals("NNS")) {
/* DP */
@@ -165,7 +167,7 @@
else if(pos.equals("JJNN") && token.contains("_")) {
String[] tokens = token.split("_");
String nntoken = tokens[tokens.length-1];
- String jjtoken = token.replace("SLOT_","").replace(nntoken,"").replace("_"," ").trim();
+ String jjtoken = token.replace("SLOT_","").replace(nntoken,"").trim();
String slotfluent = "SLOT_" + tokenfluent + "/CLASS/" + token;
String slotnn = "SLOT_" + nntoken + "/CLASS/" + nntoken;
// String semantics = "<x,l1,<e,t>,[ l1:[ | SLOT_" + tokenfluent + "(x) ] ],[],[],[" + slotfluent + "]> " +
@@ -302,6 +304,10 @@
String[] wasGerEntry = {token,
"(S DP[comp] (VP V:'was' DP[subject] V:'" + token + "'))",
"<y,l1,t,[ l1:[ | SLOT_" + token + "(y,z) ] ],[(l2,y,comp,<<e,t>,t>), (l3,z,subject,<<e,t>,t>) ],[ l2=l1, l3=l1 ],[" + symslot + "]>"};
+ String[] adjEntry = {token,
+ "(NP ADJ:'"+token+"' NP*)",
+ "<x,l1,<e,t>,[ l1:[ | SLOT_description(x,y), regex(y,'"+ token +"')] ],[],[],[ SLOT_description/DATATYPEPROPERTY/description ]>"};
+ result.add(adjEntry);
result.add(gerEntry);
result.add(wasGerEntry);
}
@@ -331,6 +337,8 @@
result.add(whereEntry1);
result.add(whereEntry2);
}
+
+ // TODO relative clauses missing!
}
/* ADJECTIVES */
Modified: trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex
===================================================================
--- trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex 2012-06-15 14:20:33 UTC (rev 3749)
+++ trunk/components-ext/src/main/resources/tbsl/lexicon/english_oxford.lex 2012-06-15 14:46:46 UTC (rev 3750)
@@ -2,12 +2,18 @@
// 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 ]>
+ 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 ]>
+ 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 ]>
+ 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/near ]>
+ 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/near ]>
+ 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/near ]>
+ 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/near ]>
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:'more' 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 less 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), 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 ]>
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 ]>
This was sent by the SourceForge.net collaborative development platform, the world's largest Open Source development site.
|
