Give the best score to the most probable exact match.

Bug: 9490364
Change-Id: If7fe5da190516d4b5f7a41b752ef5960205a5899
main
Keisuke Kuroynagi 2013-06-21 19:11:39 +09:00
parent 9380b56db6
commit f0dc0b1b74
3 changed files with 27 additions and 1 deletions

View File

@ -36,6 +36,7 @@ namespace latinime {
const int Suggest::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16; const int Suggest::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16;
const int Suggest::MIN_CONTINUOUS_SUGGESTION_INPUT_SIZE = 2; const int Suggest::MIN_CONTINUOUS_SUGGESTION_INPUT_SIZE = 2;
const float Suggest::AUTOCORRECT_CLASSIFICATION_THRESHOLD = 0.33f; const float Suggest::AUTOCORRECT_CLASSIFICATION_THRESHOLD = 0.33f;
const int Suggest::FINAL_SCORE_PENALTY_FOR_NOT_BEST_EXACT_MATCHED_WORD = 1;
/** /**
* Returns a set of suggestions for the given input touch points. The commitPoint argument indicates * Returns a set of suggestions for the given input touch points. The commitPoint argument indicates
@ -148,6 +149,8 @@ int Suggest::outputSuggestions(DicTraverseSession *traverseSession, int *frequen
&doubleLetterTerminalIndex, &doubleLetterLevel); &doubleLetterTerminalIndex, &doubleLetterLevel);
int maxScore = S_INT_MIN; int maxScore = S_INT_MIN;
int bestExactMatchedNodeTerminalIndex = -1;
int bestExactMatchedNodeOutputWordIndex = -1;
// Output suggestion results here // Output suggestion results here
for (int terminalIndex = 0; terminalIndex < terminalSize && outputWordIndex < MAX_RESULTS; for (int terminalIndex = 0; terminalIndex < terminalSize && outputWordIndex < MAX_RESULTS;
++terminalIndex) { ++terminalIndex) {
@ -186,7 +189,6 @@ int Suggest::outputSuggestions(DicTraverseSession *traverseSession, int *frequen
const int finalScore = SCORING->calculateFinalScore( const int finalScore = SCORING->calculateFinalScore(
compoundDistance, traverseSession->getInputSize(), compoundDistance, traverseSession->getInputSize(),
isForceCommitMultiWords || (isValidWord && SCORING->doesAutoCorrectValidWord())); isForceCommitMultiWords || (isValidWord && SCORING->doesAutoCorrectValidWord()));
maxScore = max(maxScore, finalScore); maxScore = max(maxScore, finalScore);
if (TRAVERSAL->allowPartialCommit()) { if (TRAVERSAL->allowPartialCommit()) {
@ -200,6 +202,25 @@ int Suggest::outputSuggestions(DicTraverseSession *traverseSession, int *frequen
if (isValidWord) { if (isValidWord) {
outputTypes[outputWordIndex] = Dictionary::KIND_CORRECTION | outputTypeFlags; outputTypes[outputWordIndex] = Dictionary::KIND_CORRECTION | outputTypeFlags;
frequencies[outputWordIndex] = finalScore; frequencies[outputWordIndex] = finalScore;
if (isSafeExactMatch) {
// Demote exact matches that are not the highest probable node among all exact
// matches.
const bool isBestTerminal = bestExactMatchedNodeTerminalIndex < 0
|| terminals[bestExactMatchedNodeTerminalIndex].getProbability()
< terminalDicNode->getProbability();
const int outputWordIndexToBeDemoted = isBestTerminal ?
bestExactMatchedNodeOutputWordIndex : outputWordIndex;
if (outputWordIndexToBeDemoted >= 0) {
frequencies[outputWordIndexToBeDemoted] -=
FINAL_SCORE_PENALTY_FOR_NOT_BEST_EXACT_MATCHED_WORD;
}
if (isBestTerminal) {
// Updates the best exact matched node index.
bestExactMatchedNodeTerminalIndex = terminalIndex;
// Updates the best exact matched output word index.
bestExactMatchedNodeOutputWordIndex = outputWordIndex;
}
}
// Populate the outputChars array with the suggested word. // Populate the outputChars array with the suggested word.
const int startIndex = outputWordIndex * MAX_WORD_LENGTH; const int startIndex = outputWordIndex * MAX_WORD_LENGTH;
terminalDicNode->outputResult(&outputCodePoints[startIndex]); terminalDicNode->outputResult(&outputCodePoints[startIndex]);

View File

@ -82,6 +82,8 @@ class Suggest : public SuggestInterface {
// Threshold for autocorrection classifier // Threshold for autocorrection classifier
static const float AUTOCORRECT_CLASSIFICATION_THRESHOLD; static const float AUTOCORRECT_CLASSIFICATION_THRESHOLD;
// Final score penalty to exact match words that are not the most probable exact match.
static const int FINAL_SCORE_PENALTY_FOR_NOT_BEST_EXACT_MATCHED_WORD;
const Traversal *const TRAVERSAL; const Traversal *const TRAVERSAL;
const Scoring *const SCORING; const Scoring *const SCORING;

View File

@ -163,6 +163,9 @@ class TypingWeighting : public Weighting {
float getTerminalLanguageCost(const DicTraverseSession *const traverseSession, float getTerminalLanguageCost(const DicTraverseSession *const traverseSession,
const DicNode *const dicNode, const float dicNodeLanguageImprobability) const { const DicNode *const dicNode, const float dicNodeLanguageImprobability) const {
// We promote exact matches here to prevent them from being pruned. The final score of
// exact match nodes might be demoted later in Suggest::outputSuggestions if there are
// multiple exact matches.
const float languageImprobability = (dicNode->isExactMatch()) ? const float languageImprobability = (dicNode->isExactMatch()) ?
0.0f : dicNodeLanguageImprobability; 0.0f : dicNodeLanguageImprobability;
return languageImprobability * ScoringParams::DISTANCE_WEIGHT_LANGUAGE; return languageImprobability * ScoringParams::DISTANCE_WEIGHT_LANGUAGE;