From 6d41ff008fe8f65a1e65ef918c6d6d209be66a2a Mon Sep 17 00:00:00 2001
From: Adam Radziszewski <adam.radziszewski@pwr.wroc.pl>
Date: Wed, 5 Dec 2012 14:53:51 +0100
Subject: [PATCH] update noprep.ccl, WCCL rules for removing prepositions from
NP chunks: handle AgP and AdjP chunks as well
---
tools/noprep.ccl | 83 ++++++++++++++++++++++++++++++++++++++++++++----
1 file changed, 76 insertions(+), 7 deletions(-)
diff --git a/tools/noprep.ccl b/tools/noprep.ccl
index 583ef6b..ede0168 100644
--- a/tools/noprep.ccl
+++ b/tools/noprep.ccl
@@ -1,28 +1,97 @@
-// generates "real" NP chunks by removing chunk-initial prepositions
-// the new chunks are added as "cut_np"
+// generates "real" NP, AdjP and AgP chunks by removing
+// chunk-initial prepositions
+// the new chunks are added as cut_np, cut_adjp and cut_agp
+// the prepositions cut from np and adjp are marked as cut_prep
// to be used with wccl-rules
tag_rules(
- rule("noprep",
+ rule("noprep_np",
and(
isannbeg(0, "chunk_np"),
not(in({prep}, class[0])),
rlook(0, end, $End,
isannend($End, "chunk_np")
+ ),
+ or(
+ rlook(0, $End, $Hd, isannhead($Hd, "chunk_np")),
+ setvar($Hd, 0)
)
),
- mark(0, $End, "cut_np")
+ mark(0, $End, $Hd, "cut_np")
);
- rule("withprep",
+ rule("withprep_np",
and(
isannbeg(0, "chunk_np"),
in({prep}, class[0]),
rlook(0, end, $End,
isannend($End, "chunk_np")
+ ),
+ or(
+ rlook(1, $End, $Hd, isannhead($Hd, "chunk_np")),
+ setvar($Hd, 1)
)
),
- mark(1, $End, "cut_np")
+ mark(0, 0, "cut_prep"),
+ mark(1, $End, $Hd, "cut_np")
+ );
+
+ rule("noprep_adjp",
+ and(
+ isannbeg(0, "chunk_adjp"),
+ not(in({prep}, class[0])),
+ rlook(0, end, $End,
+ isannend($End, "chunk_adjp")
+ ),
+ or(
+ rlook(0, $End, $Hd, isannhead($Hd, "chunk_adjp")),
+ setvar($Hd, 0)
+ )
+ ),
+ mark(0, $End, $Hd, "cut_adjp")
+ );
+ rule("withprep_adjp",
+ and(
+ isannbeg(0, "chunk_adjp"),
+ in({prep}, class[0]),
+ rlook(0, end, $End,
+ isannend($End, "chunk_adjp")
+ ),
+ or(
+ rlook(1, $End, $Hd, isannhead($Hd, "chunk_adjp")),
+ setvar($Hd, 1)
+ )
+ ),
+ mark(0, 0, "cut_prep"),
+ mark(1, $End, $Hd, "cut_adjp")
+ );
+
+ rule("noprep_agp",
+ and(
+ isannbeg(0, "chunk_agp"),
+ not(in({prep}, class[0])),
+ rlook(0, end, $End,
+ isannend($End, "chunk_agp")
+ ),
+ or(
+ rlook(0, $End, $Hd, isannhead($Hd, "chunk_agp")),
+ setvar($Hd, 0)
+ )
+ ),
+ mark(0, $End, $Hd, "cut_agp")
+ );
+ rule("withprep_agp",
+ and(
+ isannbeg(0, "chunk_agp"),
+ in({prep}, class[0]),
+ rlook(0, end, $End,
+ isannend($End, "chunk_agp")
+ ),
+ or(
+ rlook(1, $End, $Hd, isannhead($Hd, "chunk_agp")),
+ setvar($Hd, 1)
+ )
+ ),
+ mark(1, $End, $Hd, "cut_agp")
)
)
-
--
GitLab