Start updating historical information.

Bug: 11073222

Change-Id: Ic74951f46af05d3edfe08fca37352c5d5c1047e0
main
Keisuke Kuroyanagi 2013-12-09 16:02:07 +09:00
parent 0a213fd625
commit 6c1382b4ae
5 changed files with 55 additions and 9 deletions

View File

@ -55,7 +55,7 @@ bool Ver4BigramListPolicy::addNewEntry(const int terminalId, const int newTarget
}
const BigramEntry newBigramEntry(false /* hasNext */, NOT_A_PROBABILITY,
newTargetTerminalId);
const BigramEntry bigramEntryToWrite = getUpdatedBigramEntry(&newBigramEntry,
const BigramEntry bigramEntryToWrite = createUpdatedBigramEntryFrom(&newBigramEntry,
newProbability, timestamp);
// Write an entry.
const int writingPos = mBigramDictContent->getBigramListHeadPos(terminalId);
@ -81,7 +81,7 @@ bool Ver4BigramListPolicy::addNewEntry(const int terminalId, const int newTarget
}
const BigramEntry updatedBigramEntry =
originalBigramEntry.updateTargetTerminalIdAndGetEntry(newTargetTerminalId);
const BigramEntry bigramEntryToWrite = getUpdatedBigramEntry(
const BigramEntry bigramEntryToWrite = createUpdatedBigramEntryFrom(
&updatedBigramEntry, newProbability, timestamp);
return mBigramDictContent->writeBigramEntry(&bigramEntryToWrite, entryPosToUpdate);
}
@ -94,7 +94,7 @@ bool Ver4BigramListPolicy::addNewEntry(const int terminalId, const int newTarget
// Write new entry at a head position of the bigram list.
int writingPos = mBigramDictContent->getBigramListHeadPos(terminalId);
const BigramEntry newBigramEntry(true /* hasNext */, NOT_A_PROBABILITY, newTargetTerminalId);
const BigramEntry bigramEntryToWrite = getUpdatedBigramEntry(
const BigramEntry bigramEntryToWrite = createUpdatedBigramEntryFrom(
&newBigramEntry, newProbability, timestamp);
if (!mBigramDictContent->writeBigramEntryAndAdvancePosition(&bigramEntryToWrite, &writingPos)) {
return false;
@ -218,14 +218,17 @@ int Ver4BigramListPolicy::getEntryPosToUpdate(const int targetTerminalIdToFind,
return invalidEntryPos;
}
const BigramEntry Ver4BigramListPolicy::getUpdatedBigramEntry(
const BigramEntry Ver4BigramListPolicy::createUpdatedBigramEntryFrom(
const BigramEntry *const originalBigramEntry, const int newProbability,
const int timestamp) const {
if (mNeedsToDecayWhenUpdating) {
// TODO: Update historical information.
const int probability = ForgettingCurveUtils::getUpdatedEncodedProbability(
originalBigramEntry->getProbability(), newProbability);
return originalBigramEntry->updateProbabilityAndGetEntry(probability);
const HistoricalInfo updatedHistoricalInfo =
ForgettingCurveUtils::createUpdatedHistoricalInfoFrom(
originalBigramEntry->getHistoricalInfo(), newProbability, timestamp);
return originalBigramEntry->updateProbabilityAndGetEntry(probability)
.updateHistoricalInfoAndGetEntry(&updatedHistoricalInfo);
} else {
return originalBigramEntry->updateProbabilityAndGetEntry(newProbability);
}

View File

@ -59,7 +59,7 @@ class Ver4BigramListPolicy : public DictionaryBigramsStructurePolicy {
int getEntryPosToUpdate(const int targetTerminalIdToFind, const int bigramListPos) const;
const BigramEntry getUpdatedBigramEntry(const BigramEntry *const originalBigramEntry,
const BigramEntry createUpdatedBigramEntryFrom(const BigramEntry *const originalBigramEntry,
const int newProbability, const int timestamp) const;
BigramDictContent *const mBigramDictContent;

View File

@ -292,10 +292,13 @@ const ProbabilityEntry Ver4PatriciaTrieNodeWriter::createUpdatedEntryFrom(
const ProbabilityEntry *const originalProbabilityEntry, const int newProbability,
const int timestamp) const {
if (mNeedsToDecayWhenUpdating) {
// TODO: Update historical information.
const int updatedProbability = ForgettingCurveUtils::getUpdatedEncodedProbability(
originalProbabilityEntry->getProbability(), newProbability);
return originalProbabilityEntry->createEntryWithUpdatedProbability(updatedProbability);
const HistoricalInfo updatedHistoricalInfo =
ForgettingCurveUtils::createUpdatedHistoricalInfoFrom(
originalProbabilityEntry->getHistoricalInfo(), newProbability, timestamp);
return originalProbabilityEntry->createEntryWithUpdatedProbability(updatedProbability)
.createEntryWithUpdatedHistoricalInfo(&updatedHistoricalInfo);
} else {
return originalProbabilityEntry->createEntryWithUpdatedProbability(newProbability);
}

View File

@ -39,8 +39,39 @@ const int ForgettingCurveUtils::ENCODED_PROBABILITY_STEP = 1;
const float ForgettingCurveUtils::MIN_PROBABILITY_TO_DECAY = 0.03f;
const int ForgettingCurveUtils::DECAY_INTERVAL_SECONDS = 2 * 60 * 60;
const int ForgettingCurveUtils::MAX_LEVEL = 3;
const int ForgettingCurveUtils::MAX_COUNT = 3;
const int ForgettingCurveUtils::MIN_VALID_LEVEL = 1;
const ForgettingCurveUtils::ProbabilityTable ForgettingCurveUtils::sProbabilityTable;
/* static */ const HistoricalInfo ForgettingCurveUtils::createUpdatedHistoricalInfoFrom(
const HistoricalInfo *const originalHistoricalInfo,
const int newProbability, const int timestamp) {
if (newProbability != NOT_A_PROBABILITY && originalHistoricalInfo->getLevel() == 0) {
return HistoricalInfo(timestamp, MIN_VALID_LEVEL /* level */, 0 /* count */);
} else if (originalHistoricalInfo->getTimeStamp() == NOT_A_TIMESTAMP) {
// Initial information.
return HistoricalInfo(timestamp, 0 /* level */, 0 /* count */);
} else {
const int updatedCount = originalHistoricalInfo->getCount() + 1;
if (updatedCount > MAX_COUNT) {
// The count exceeds the max value the level can be incremented.
if (originalHistoricalInfo->getLevel() >= MAX_LEVEL) {
// The level is already max.
return HistoricalInfo(timestamp, originalHistoricalInfo->getLevel(),
originalHistoricalInfo->getCount());
} else {
// Level up.
return HistoricalInfo(timestamp, originalHistoricalInfo->getLevel() + 1,
0 /* count */);
}
} else {
return HistoricalInfo(timestamp, originalHistoricalInfo->getLevel(), updatedCount);
}
}
}
/* static */ int ForgettingCurveUtils::getProbability(const int encodedUnigramProbability,
const int encodedBigramProbability) {
if (encodedUnigramProbability == NOT_A_PROBABILITY) {

View File

@ -20,6 +20,7 @@
#include <vector>
#include "defines.h"
#include "suggest/policyimpl/dictionary/utils/historical_info.h"
namespace latinime {
@ -35,6 +36,10 @@ class ForgettingCurveUtils {
static const int MAX_BIGRAM_COUNT;
static const int MAX_BIGRAM_COUNT_AFTER_GC;
static const HistoricalInfo createUpdatedHistoricalInfoFrom(
const HistoricalInfo *const originalHistoricalInfo, const int newProbability,
const int timestamp);
static int getProbability(const int encodedUnigramProbability,
const int encodedBigramProbability);
@ -76,6 +81,10 @@ class ForgettingCurveUtils {
static const float MIN_PROBABILITY_TO_DECAY;
static const int DECAY_INTERVAL_SECONDS;
static const int MAX_LEVEL;
static const int MAX_COUNT;
static const int MIN_VALID_LEVEL;
static const ProbabilityTable sProbabilityTable;
static int decodeProbability(const int encodedProbability);