Partial fix for '(QP up to ...)' as reported in #1363

stanfordnlp · Jul 8, 2023 · 8c46648 · 8c46648
1 parent 5c57fde
commit 8c46648
Show file tree

Hide file tree

Showing 3 changed files with 12 additions and 2 deletions.
diff --git a/src/edu/stanford/nlp/trees/CoordinationTransformer.java b/src/edu/stanford/nlp/trees/CoordinationTransformer.java
@@ -686,7 +686,7 @@ private static Tree findCCparent(Tree t, Tree root) {
     TregexPattern.compile("@WHADVP < ((WRB=node1 < /^(?i:how)$/) $+ (VB=node2 < /^(?i)come$/))"), //how come
     TregexPattern.compile("@VP < ((VBD=node1 < had|'d) $+ (@PRT|ADVP=node2 <: (RBR < /^(?i)better$/)))"), //had better
     TregexPattern.compile("@QP|XS < ((JJR|RBR|IN=node1 < /^(?i)(more|less)$/) $+ (IN=node2 < /^(?i)than$/))"), //more/less than
-    TregexPattern.compile("@QP < ((JJR|RBR|IN=node1 < /^(?i)up$/) $+ (IN|TO=node2 < /^(?i)to$/))"), //up to
+    TregexPattern.compile("@QP < ((JJR|RBR|RP|IN=node1 < /^(?i)up$/) $+ (IN|TO=node2 < /^(?i)to$/))"), //up to
     TregexPattern.compile("@S|SQ|VP|ADVP|PP < (@ADVP < ((IN|RB=node1 < /^(?i)at$/) $+ (JJS|RBS=node2 < /^(?i)least$/)) !$+ (RB < /(?i)(once|twice)/))"), //at least
 
   };

diff --git a/src/edu/stanford/nlp/trees/EnglishGrammaticalRelations.java b/src/edu/stanford/nlp/trees/EnglishGrammaticalRelations.java
@@ -980,7 +980,15 @@ private EnglishGrammaticalRelations() {}
   public static final GrammaticalRelation QUANTIFIER_MODIFIER =
     new GrammaticalRelation(Language.English, "quantmod", "quantifier modifier",
         MODIFIER, "QP", tregexCompiler,
-            "QP < IN|RB|RBR|RBS|PDT|DT|JJ|JJR|JJS|XS=target");
+            // RP is because sometimes "up" in "up to ___" gets tagged RP in PTB
+            // this is probably a mistake - generally it is tagged IN
+            // but sometimes the tagger follows suit
+            // there are no conflicts elsewhere in the targets of a QP,
+            // so there should be no need to specifically check for the phrase "up to" for `up_RP`
+            "QP < IN|RB|RBR|RBS|PDT|DT|JJ|JJR|JJS|XS|RP=target",
+            // TO is for the "to" in "up to ___"
+            // TODO: but currently not working for up_IN to_IN foo_CD, since it wants to make TO the head of IN!
+            "(QP < (TO=target < /^(?i:to)$/) < (__=up < /^(?i:up)$/)) : (=up $++ =target)");
 
 
   /**

diff --git a/src/edu/stanford/nlp/trees/UniversalEnglishGrammaticalRelations.java b/src/edu/stanford/nlp/trees/UniversalEnglishGrammaticalRelations.java
@@ -1066,6 +1066,8 @@ private UniversalEnglishGrammaticalRelations() {}
             "/^(?:VP|ADJP|JJP|WHADJP|SQ?|SBARQ?|SINV|XS|RRC|(?:WH)?NP(?:-TMP|-ADV)?)$/ < (RB|RBR|RBS|WRB|ADVP|WHADVP=target !< " + ETC_PAT + " [!<+(/ADVP/) (@ADVP < (IN < /(?i:at)/)) |  !<+(/ADVP/) (@ADVP < NP)] )",
             "QP < IN|RB|RBR|RBS|PDT|DT|JJ|JJR|JJS|XS=target", //quantmod relation in original SD
             "QP < (MWE=target < (JJR|RBR|IN < /^(?i)(more|less)$/) < (IN < /^(?i)than$/))", //more than / less than
+            // TODO: could condense with the previous rule to save a little time
+            "QP < (MWE=target < (__ < /^(?i)up$/) < (__ < /^(?i)to$/))", // up to
             // avoids adverb conjunctions matching as advmod; added JJ to catch How long
             // "!< no" so we can get neg instead for "no foo" when no is tagged as RB
             // we allow CC|CONJP as long as it is not between the target and the head