Generalized Quoted-ISA rules to compensate parsing mistakes.

Author: Mischn

pom.xml

@@ -26,7 +26,7 @@
 
     <groupId>org.lambda3.text.simplification</groupId>
     <artifactId>discourse-simplification</artifactId>
-    <version>7.1.0-SNAPSHOT</version>
+    <version>7.1.0</version>
     <packaging>jar</packaging>
 
     <name>Discourse Simplification</name>

src/main/java/org/lambda3/text/simplification/discourse/tree/extraction/rules/QuotedISAPostExtractor.java

@@ -48,23 +48,19 @@ public class QuotedISAPostExtractor extends ExtractionRule {
 
     @Override
     public Optional<Extraction> extract(Tree parseTree) {
-        TregexPattern p = TregexPattern.compile("ROOT <<: (S < (S=s < (NP $.. VP) ?,, (/``/=startOut) ?<<, (/``/=startIn) ?<<- (/''/=endIn) ?.. (/''/=endOut) $.. (NP $.. VP)))");
+        TregexPattern p = TregexPattern.compile("ROOT <<: (S << (/``/=start .. (NP .. (/VB./ .. (/''/=end .. (NP .. VP))))))");
         TregexMatcher matcher = p.matcher(parseTree);
 
         while (matcher.findAt(parseTree)) {
             Tree quoteStart;
-            if (matcher.getNode("startOut") != null) {
-                quoteStart = matcher.getNode("startOut");
-            } else if (matcher.getNode("startIn") != null) {
-                quoteStart = matcher.getNode("startIn");
+            if (matcher.getNode("start") != null) {
+                quoteStart = (matcher.getNode("start"));
             } else {
                 continue;
             }
             Tree quoteEnd;
-            if (matcher.getNode("endOut") != null) {
-                quoteEnd = matcher.getNode("endOut");
-            } else if (matcher.getNode("endIn") != null) {
-                quoteEnd = matcher.getNode("endIn");
+            if (matcher.getNode("end") != null) {
+                quoteEnd = matcher.getNode("end");
             } else {
                 continue;
             }

src/main/java/org/lambda3/text/simplification/discourse/tree/extraction/rules/QuotedISAPreExtractor.java

@@ -49,23 +49,19 @@ public class QuotedISAPreExtractor extends ExtractionRule {
 
     @Override
     public Optional<Extraction> extract(Tree parseTree) {
-        TregexPattern p = TregexPattern.compile("ROOT <<: (S < (NP $.. (VP <+(VP) (S=s < (NP $.. VP) ?,, (/``/=startOut) ?<<, (/``/=startIn) ?<<- (/''/=endIn) ?.. (/''/=endOut)))))");
+        TregexPattern p = TregexPattern.compile("ROOT <<: (S << (NP .. (VP .. (/``/=start .. (NP .. (/VB./ .. (/''/=end)))))))");
         TregexMatcher matcher = p.matcher(parseTree);
 
         while (matcher.findAt(parseTree)) {
             Tree quoteStart;
-            if (matcher.getNode("startOut") != null) {
-                quoteStart = matcher.getNode("startOut");
-            } else if (matcher.getNode("startIn") != null) {
-                quoteStart = matcher.getNode("startIn");
+            if (matcher.getNode("start") != null) {
+                quoteStart = (matcher.getNode("start"));
             } else {
                 continue;
             }
             Tree quoteEnd;
-            if (matcher.getNode("endOut") != null) {
-                quoteEnd = matcher.getNode("endOut");
-            } else if (matcher.getNode("endIn") != null) {
-                quoteEnd = matcher.getNode("endIn");
+            if (matcher.getNode("end") != null) {
+                quoteEnd = matcher.getNode("end");
             } else {
                 continue;
             }