From a161a4afd6145d7ed6ba7f67106f5fb5c9887903 Mon Sep 17 00:00:00 2001
From: satok <satok@google.com>
Date: Mon, 16 Jan 2012 18:38:32 +0900
Subject: [PATCH] Use edit distance for transposing correction
+1 73
-1 4
+2 0
-2 0
+3 0
-3 0
+4 11
-4 19
+5 9
-5 3
+6 2
-6 63
+7 2
-7 8
Change-Id: I269cd2386f451f8932e4e0ae66223e794fdfa862
---
native/src/correction.cpp | 26 ++++++++++++++------------
native/src/defines.h | 2 +-
2 files changed, 15 insertions(+), 13 deletions(-)
diff --git a/native/src/correction.cpp b/native/src/correction.cpp
index 75831b69d..5dc6f8737 100644
--- a/native/src/correction.cpp
+++ b/native/src/correction.cpp
@@ -653,9 +653,10 @@ int Correction::RankingAlgorithm::calculateFinalFreq(const int inputIndex, const
int finalFreq = freq;
// TODO: Optimize this.
- // TODO: Ignoring edit distance for transposed char, for now
- if (transposedCount == 0 && (proximityMatchedCount > 0 || skipped || excessiveCount > 0)) {
- ed = getCurrentEditDistance(editDistanceTable, inputLength, outputIndex + 1);
+ if (transposedCount > 0 || proximityMatchedCount > 0 || skipped || excessiveCount > 0) {
+ ed = getCurrentEditDistance(editDistanceTable, inputLength, outputIndex + 1)
+ - transposedCount;
+
const int matchWeight = powerIntCapped(typedLetterMultiplier,
max(inputLength, outputIndex + 1) - ed);
multiplyIntCapped(matchWeight, &finalFreq);
@@ -667,21 +668,22 @@ int Correction::RankingAlgorithm::calculateFinalFreq(const int inputIndex, const
ed = max(0, ed - quoteDiffCount);
- if (ed == 1 && (inputLength == outputIndex || inputLength == outputIndex + 2)) {
- // Promote a word with just one skipped or excessive char
- if (sameLength) {
- multiplyRate(WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_RATE, &finalFreq);
- } else {
+ if (transposedCount < 1) {
+ if (ed == 1 && (inputLength == outputIndex || inputLength == outputIndex + 2)) {
+ // Promote a word with just one skipped or excessive char
+ if (sameLength) {
+ multiplyRate(WORDS_WITH_JUST_ONE_CORRECTION_PROMOTION_RATE, &finalFreq);
+ } else {
+ multiplyIntCapped(typedLetterMultiplier, &finalFreq);
+ }
+ } else if (ed == 0) {
multiplyIntCapped(typedLetterMultiplier, &finalFreq);
+ sameLength = true;
}
- } else if (ed == 0) {
- multiplyIntCapped(typedLetterMultiplier, &finalFreq);
- sameLength = true;
}
adjustedProximityMatchedCount = min(max(0, ed - (outputIndex + 1 - inputLength)),
proximityMatchedCount);
} else {
- // TODO: Calculate the edit distance for transposed char
const int matchWeight = powerIntCapped(typedLetterMultiplier, matchCount);
multiplyIntCapped(matchWeight, &finalFreq);
}
diff --git a/native/src/defines.h b/native/src/defines.h
index ce3f85acd..31175c369 100644
--- a/native/src/defines.h
+++ b/native/src/defines.h
@@ -190,7 +190,7 @@ static void prof_out(void) {
#define WORDS_WITH_MISSING_SPACE_CHARACTER_DEMOTION_RATE 67
#define WORDS_WITH_EXCESSIVE_CHARACTER_DEMOTION_RATE 75
#define WORDS_WITH_EXCESSIVE_CHARACTER_OUT_OF_PROXIMITY_DEMOTION_RATE 75
-#define WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE 60
+#define WORDS_WITH_TRANSPOSED_CHARACTERS_DEMOTION_RATE 70
#define FULL_MATCHED_WORDS_PROMOTION_RATE 120
#define WORDS_WITH_PROXIMITY_CHARACTER_DEMOTION_RATE 90
#define WORDS_WITH_MATCH_SKIP_PROMOTION_RATE 105