am 13864787: Merge "Rename probability to unigramProbability."
* commit '138647876becc41c401087037013e781142753b6': Rename probability to unigramProbability.main
commit
e53c39b43c
|
@ -136,14 +136,14 @@ class DicNode {
|
||||||
}
|
}
|
||||||
|
|
||||||
void initAsChild(const DicNode *const dicNode, const int childrenPtNodeArrayPos,
|
void initAsChild(const DicNode *const dicNode, const int childrenPtNodeArrayPos,
|
||||||
const int probability, const int wordId, const bool isBlacklistedOrNotAWord,
|
const int unigramProbability, const int wordId, const bool isBlacklistedOrNotAWord,
|
||||||
const uint16_t mergedNodeCodePointCount, const int *const mergedNodeCodePoints) {
|
const uint16_t mergedNodeCodePointCount, const int *const mergedNodeCodePoints) {
|
||||||
uint16_t newDepth = static_cast<uint16_t>(dicNode->getNodeCodePointCount() + 1);
|
uint16_t newDepth = static_cast<uint16_t>(dicNode->getNodeCodePointCount() + 1);
|
||||||
mIsCachedForNextSuggestion = dicNode->mIsCachedForNextSuggestion;
|
mIsCachedForNextSuggestion = dicNode->mIsCachedForNextSuggestion;
|
||||||
const uint16_t newLeavingDepth = static_cast<uint16_t>(
|
const uint16_t newLeavingDepth = static_cast<uint16_t>(
|
||||||
dicNode->mDicNodeProperties.getLeavingDepth() + mergedNodeCodePointCount);
|
dicNode->mDicNodeProperties.getLeavingDepth() + mergedNodeCodePointCount);
|
||||||
mDicNodeProperties.init(childrenPtNodeArrayPos, mergedNodeCodePoints[0],
|
mDicNodeProperties.init(childrenPtNodeArrayPos, mergedNodeCodePoints[0],
|
||||||
probability, wordId, isBlacklistedOrNotAWord, newDepth, newLeavingDepth,
|
unigramProbability, wordId, isBlacklistedOrNotAWord, newDepth, newLeavingDepth,
|
||||||
dicNode->mDicNodeProperties.getPrevWordIds());
|
dicNode->mDicNodeProperties.getPrevWordIds());
|
||||||
mDicNodeState.init(&dicNode->mDicNodeState, mergedNodeCodePointCount,
|
mDicNodeState.init(&dicNode->mDicNodeState, mergedNodeCodePointCount,
|
||||||
mergedNodeCodePoints);
|
mergedNodeCodePoints);
|
||||||
|
@ -217,8 +217,9 @@ class DicNode {
|
||||||
return mDicNodeProperties.getChildrenPtNodeArrayPos();
|
return mDicNodeProperties.getChildrenPtNodeArrayPos();
|
||||||
}
|
}
|
||||||
|
|
||||||
int getProbability() const {
|
// TODO: Remove
|
||||||
return mDicNodeProperties.getProbability();
|
int getUnigramProbability() const {
|
||||||
|
return mDicNodeProperties.getUnigramProbability();
|
||||||
}
|
}
|
||||||
|
|
||||||
AK_FORCE_INLINE bool isTerminalDicNode() const {
|
AK_FORCE_INLINE bool isTerminalDicNode() const {
|
||||||
|
|
|
@ -84,7 +84,7 @@ namespace latinime {
|
||||||
/* static */ int DicNodeUtils::getBigramNodeProbability(
|
/* static */ int DicNodeUtils::getBigramNodeProbability(
|
||||||
const DictionaryStructureWithBufferPolicy *const dictionaryStructurePolicy,
|
const DictionaryStructureWithBufferPolicy *const dictionaryStructurePolicy,
|
||||||
const DicNode *const dicNode, MultiBigramMap *const multiBigramMap) {
|
const DicNode *const dicNode, MultiBigramMap *const multiBigramMap) {
|
||||||
const int unigramProbability = dicNode->getProbability();
|
const int unigramProbability = dicNode->getUnigramProbability();
|
||||||
if (multiBigramMap) {
|
if (multiBigramMap) {
|
||||||
const int *const prevWordIds = dicNode->getPrevWordIds();
|
const int *const prevWordIds = dicNode->getPrevWordIds();
|
||||||
return multiBigramMap->getBigramProbability(dictionaryStructurePolicy,
|
return multiBigramMap->getBigramProbability(dictionaryStructurePolicy,
|
||||||
|
|
|
@ -59,11 +59,11 @@ class DicNodeVector {
|
||||||
}
|
}
|
||||||
|
|
||||||
void pushLeavingChild(const DicNode *const dicNode, const int childrenPtNodeArrayPos,
|
void pushLeavingChild(const DicNode *const dicNode, const int childrenPtNodeArrayPos,
|
||||||
const int probability, const int wordId, const bool isBlacklistedOrNotAWord,
|
const int unigramProbability, const int wordId, const bool isBlacklistedOrNotAWord,
|
||||||
const uint16_t mergedNodeCodePointCount, const int *const mergedNodeCodePoints) {
|
const uint16_t mergedNodeCodePointCount, const int *const mergedNodeCodePoints) {
|
||||||
ASSERT(!mLock);
|
ASSERT(!mLock);
|
||||||
mDicNodes.emplace_back();
|
mDicNodes.emplace_back();
|
||||||
mDicNodes.back().initAsChild(dicNode, childrenPtNodeArrayPos, probability,
|
mDicNodes.back().initAsChild(dicNode, childrenPtNodeArrayPos, unigramProbability,
|
||||||
wordId, isBlacklistedOrNotAWord, mergedNodeCodePointCount, mergedNodeCodePoints);
|
wordId, isBlacklistedOrNotAWord, mergedNodeCodePointCount, mergedNodeCodePoints);
|
||||||
}
|
}
|
||||||
|
|
||||||
|
|
|
@ -29,19 +29,19 @@ namespace latinime {
|
||||||
class DicNodeProperties {
|
class DicNodeProperties {
|
||||||
public:
|
public:
|
||||||
AK_FORCE_INLINE DicNodeProperties()
|
AK_FORCE_INLINE DicNodeProperties()
|
||||||
: mChildrenPtNodeArrayPos(NOT_A_DICT_POS), mProbability(NOT_A_PROBABILITY),
|
: mChildrenPtNodeArrayPos(NOT_A_DICT_POS), mUnigramProbability(NOT_A_PROBABILITY),
|
||||||
mDicNodeCodePoint(NOT_A_CODE_POINT), mWordId(NOT_A_WORD_ID),
|
mDicNodeCodePoint(NOT_A_CODE_POINT), mWordId(NOT_A_WORD_ID),
|
||||||
mIsBlacklistedOrNotAWord(false), mDepth(0), mLeavingDepth(0) {}
|
mIsBlacklistedOrNotAWord(false), mDepth(0), mLeavingDepth(0) {}
|
||||||
|
|
||||||
~DicNodeProperties() {}
|
~DicNodeProperties() {}
|
||||||
|
|
||||||
// Should be called only once per DicNode is initialized.
|
// Should be called only once per DicNode is initialized.
|
||||||
void init(const int childrenPos, const int nodeCodePoint, const int probability,
|
void init(const int childrenPos, const int nodeCodePoint, const int unigramProbability,
|
||||||
const int wordId, const bool isBlacklistedOrNotAWord, const uint16_t depth,
|
const int wordId, const bool isBlacklistedOrNotAWord, const uint16_t depth,
|
||||||
const uint16_t leavingDepth, const int *const prevWordIds) {
|
const uint16_t leavingDepth, const int *const prevWordIds) {
|
||||||
mChildrenPtNodeArrayPos = childrenPos;
|
mChildrenPtNodeArrayPos = childrenPos;
|
||||||
mDicNodeCodePoint = nodeCodePoint;
|
mDicNodeCodePoint = nodeCodePoint;
|
||||||
mProbability = probability;
|
mUnigramProbability = unigramProbability;
|
||||||
mWordId = wordId;
|
mWordId = wordId;
|
||||||
mIsBlacklistedOrNotAWord = isBlacklistedOrNotAWord;
|
mIsBlacklistedOrNotAWord = isBlacklistedOrNotAWord;
|
||||||
mDepth = depth;
|
mDepth = depth;
|
||||||
|
@ -53,7 +53,7 @@ class DicNodeProperties {
|
||||||
void init(const int rootPtNodeArrayPos, const int *const prevWordIds) {
|
void init(const int rootPtNodeArrayPos, const int *const prevWordIds) {
|
||||||
mChildrenPtNodeArrayPos = rootPtNodeArrayPos;
|
mChildrenPtNodeArrayPos = rootPtNodeArrayPos;
|
||||||
mDicNodeCodePoint = NOT_A_CODE_POINT;
|
mDicNodeCodePoint = NOT_A_CODE_POINT;
|
||||||
mProbability = NOT_A_PROBABILITY;
|
mUnigramProbability = NOT_A_PROBABILITY;
|
||||||
mWordId = NOT_A_WORD_ID;
|
mWordId = NOT_A_WORD_ID;
|
||||||
mIsBlacklistedOrNotAWord = false;
|
mIsBlacklistedOrNotAWord = false;
|
||||||
mDepth = 0;
|
mDepth = 0;
|
||||||
|
@ -64,7 +64,7 @@ class DicNodeProperties {
|
||||||
void initByCopy(const DicNodeProperties *const dicNodeProp) {
|
void initByCopy(const DicNodeProperties *const dicNodeProp) {
|
||||||
mChildrenPtNodeArrayPos = dicNodeProp->mChildrenPtNodeArrayPos;
|
mChildrenPtNodeArrayPos = dicNodeProp->mChildrenPtNodeArrayPos;
|
||||||
mDicNodeCodePoint = dicNodeProp->mDicNodeCodePoint;
|
mDicNodeCodePoint = dicNodeProp->mDicNodeCodePoint;
|
||||||
mProbability = dicNodeProp->mProbability;
|
mUnigramProbability = dicNodeProp->mUnigramProbability;
|
||||||
mWordId = dicNodeProp->mWordId;
|
mWordId = dicNodeProp->mWordId;
|
||||||
mIsBlacklistedOrNotAWord = dicNodeProp->mIsBlacklistedOrNotAWord;
|
mIsBlacklistedOrNotAWord = dicNodeProp->mIsBlacklistedOrNotAWord;
|
||||||
mDepth = dicNodeProp->mDepth;
|
mDepth = dicNodeProp->mDepth;
|
||||||
|
@ -76,7 +76,7 @@ class DicNodeProperties {
|
||||||
void init(const DicNodeProperties *const dicNodeProp, const int codePoint) {
|
void init(const DicNodeProperties *const dicNodeProp, const int codePoint) {
|
||||||
mChildrenPtNodeArrayPos = dicNodeProp->mChildrenPtNodeArrayPos;
|
mChildrenPtNodeArrayPos = dicNodeProp->mChildrenPtNodeArrayPos;
|
||||||
mDicNodeCodePoint = codePoint; // Overwrite the node char of a passing child
|
mDicNodeCodePoint = codePoint; // Overwrite the node char of a passing child
|
||||||
mProbability = dicNodeProp->mProbability;
|
mUnigramProbability = dicNodeProp->mUnigramProbability;
|
||||||
mWordId = dicNodeProp->mWordId;
|
mWordId = dicNodeProp->mWordId;
|
||||||
mIsBlacklistedOrNotAWord = dicNodeProp->mIsBlacklistedOrNotAWord;
|
mIsBlacklistedOrNotAWord = dicNodeProp->mIsBlacklistedOrNotAWord;
|
||||||
mDepth = dicNodeProp->mDepth + 1; // Increment the depth of a passing child
|
mDepth = dicNodeProp->mDepth + 1; // Increment the depth of a passing child
|
||||||
|
@ -88,8 +88,8 @@ class DicNodeProperties {
|
||||||
return mChildrenPtNodeArrayPos;
|
return mChildrenPtNodeArrayPos;
|
||||||
}
|
}
|
||||||
|
|
||||||
int getProbability() const {
|
int getUnigramProbability() const {
|
||||||
return mProbability;
|
return mUnigramProbability;
|
||||||
}
|
}
|
||||||
|
|
||||||
int getDicNodeCodePoint() const {
|
int getDicNodeCodePoint() const {
|
||||||
|
@ -130,9 +130,11 @@ class DicNodeProperties {
|
||||||
// Use a default copy constructor and an assign operator because shallow copies are ok
|
// Use a default copy constructor and an assign operator because shallow copies are ok
|
||||||
// for this class
|
// for this class
|
||||||
int mChildrenPtNodeArrayPos;
|
int mChildrenPtNodeArrayPos;
|
||||||
int mProbability;
|
// TODO: Remove
|
||||||
|
int mUnigramProbability;
|
||||||
int mDicNodeCodePoint;
|
int mDicNodeCodePoint;
|
||||||
int mWordId;
|
int mWordId;
|
||||||
|
// TODO: Remove
|
||||||
bool mIsBlacklistedOrNotAWord;
|
bool mIsBlacklistedOrNotAWord;
|
||||||
uint16_t mDepth;
|
uint16_t mDepth;
|
||||||
uint16_t mLeavingDepth;
|
uint16_t mLeavingDepth;
|
||||||
|
|
|
@ -54,15 +54,15 @@ namespace latinime {
|
||||||
current.swap(next);
|
current.swap(next);
|
||||||
}
|
}
|
||||||
|
|
||||||
int maxProbability = NOT_A_PROBABILITY;
|
int maxUnigramProbability = NOT_A_PROBABILITY;
|
||||||
for (const DicNode &dicNode : current) {
|
for (const DicNode &dicNode : current) {
|
||||||
if (!dicNode.isTerminalDicNode()) {
|
if (!dicNode.isTerminalDicNode()) {
|
||||||
continue;
|
continue;
|
||||||
}
|
}
|
||||||
// dicNode can contain case errors, accent errors, intentional omissions or digraphs.
|
// dicNode can contain case errors, accent errors, intentional omissions or digraphs.
|
||||||
maxProbability = std::max(maxProbability, dicNode.getProbability());
|
maxUnigramProbability = std::max(maxUnigramProbability, dicNode.getUnigramProbability());
|
||||||
}
|
}
|
||||||
return maxProbability;
|
return maxUnigramProbability;
|
||||||
}
|
}
|
||||||
|
|
||||||
/* static */ void DictionaryUtils::processChildDicNodes(
|
/* static */ void DictionaryUtils::processChildDicNodes(
|
||||||
|
|
|
@ -87,7 +87,7 @@ const int SuggestionsOutputUtils::MIN_LEN_FOR_MULTI_WORD_AUTOCORRECT = 16;
|
||||||
+ doubleLetterCost;
|
+ doubleLetterCost;
|
||||||
const bool isPossiblyOffensiveWord =
|
const bool isPossiblyOffensiveWord =
|
||||||
traverseSession->getDictionaryStructurePolicy()->getProbability(
|
traverseSession->getDictionaryStructurePolicy()->getProbability(
|
||||||
terminalDicNode->getProbability(), NOT_A_PROBABILITY) <= 0;
|
terminalDicNode->getUnigramProbability(), NOT_A_PROBABILITY) <= 0;
|
||||||
const bool isExactMatch =
|
const bool isExactMatch =
|
||||||
ErrorTypeUtils::isExactMatch(terminalDicNode->getContainedErrorTypes());
|
ErrorTypeUtils::isExactMatch(terminalDicNode->getContainedErrorTypes());
|
||||||
const bool isExactMatchWithIntentionalOmission =
|
const bool isExactMatchWithIntentionalOmission =
|
||||||
|
|
|
@ -162,7 +162,8 @@ class TypingTraversal : public Traversal {
|
||||||
}
|
}
|
||||||
|
|
||||||
AK_FORCE_INLINE bool isGoodToTraverseNextWord(const DicNode *const dicNode) const {
|
AK_FORCE_INLINE bool isGoodToTraverseNextWord(const DicNode *const dicNode) const {
|
||||||
const int probability = dicNode->getProbability();
|
// TODO: Quit using unigram probability and use probability in the context.
|
||||||
|
const int probability = dicNode->getUnigramProbability();
|
||||||
if (probability < ScoringParams::THRESHOLD_NEXT_WORD_PROBABILITY) {
|
if (probability < ScoringParams::THRESHOLD_NEXT_WORD_PROBABILITY) {
|
||||||
return false;
|
return false;
|
||||||
}
|
}
|
||||||
|
|
Loading…
Reference in New Issue