mirror of
https://github.com/moses-smt/mosesdecoder.git
synced 2024-12-29 06:52:34 +03:00
403 lines
13 KiB
C++
403 lines
13 KiB
C++
// memscore - in-memory phrase scoring for Statistical Machine Translation
|
|
// Christian Hardmeier, FBK-irst, Trento, 2010
|
|
// $Id$
|
|
|
|
#include <cassert>
|
|
#include <cstring>
|
|
#include <fstream>
|
|
|
|
#include "scorer-impl.h"
|
|
#include "timestamp.h"
|
|
#include "lexdecom.h"
|
|
|
|
#ifdef ENABLE_CHANNEL_SCORER
|
|
#include "channel-scorer.h"
|
|
#endif
|
|
|
|
const std::vector<String> &PhraseScorerFactory::scorer_list()
|
|
{
|
|
static std::vector<String> list;
|
|
if(list.size() == 0) {
|
|
list.push_back("ml - maximum likelihood score (relative frequency)");
|
|
list.push_back("wittenbell - Witten-Bell smoothing");
|
|
list.push_back("absdiscount - absolute discounting");
|
|
list.push_back("kndiscount1 - Knesser-Ney discounting");
|
|
list.push_back("kndiscount3 - modified Knesser-Ney discounting");
|
|
list.push_back("lexweights <weightfile> - lexical weights (Koehn et al., NAACL 2003)");
|
|
#ifdef ENABLE_CHANNEL_SCORER
|
|
list.push_back("channel <sigma> <srclm> <tgtlm> - channel adaptation");
|
|
#endif
|
|
list.push_back("const <constant> - constant phrase penalty");
|
|
list.push_back("lexdecomp <weightfile> - lexical decomposition smoothing");
|
|
}
|
|
|
|
return list;
|
|
}
|
|
|
|
PhraseScorer *PhraseScorerFactory::create_scorer(const char *argv[], int &argp, bool reverse)
|
|
{
|
|
const char *arg = argv[argp++];
|
|
if(arg == NULL)
|
|
usage();
|
|
|
|
if(!strcmp(arg, "ml"))
|
|
return MLPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else if(!strcmp(arg, "wittenbell"))
|
|
return WittenBellPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else if(!strcmp(arg, "absdiscount"))
|
|
return AbsoluteDiscountPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else if(!strcmp(arg, "kndiscount1"))
|
|
return KNDiscount1PhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else if(!strcmp(arg, "kndiscount3"))
|
|
return KNDiscount3PhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else if(!strcmp(arg, "lexweights"))
|
|
return LexicalWeightPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
#ifdef ENABLE_CHANNEL_SCORER
|
|
else if(!strcmp(arg, "channel"))
|
|
return ChannelAdaptationPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
#endif
|
|
else if(!strcmp(arg, "const"))
|
|
return ConstantPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else if (!strcmp(arg, "lexdecomp"))
|
|
return LexicalDecompositionPhraseScorer::create_scorer(argv, argp, reverse, *this);
|
|
else {
|
|
std::cerr << "Unknown phrase scorer type: " << arg << std::endl << std::endl;
|
|
usage();
|
|
}
|
|
}
|
|
|
|
PhraseScorer *MLPhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
return new MLPhraseScorer(ptf.get_phrase_table(), reverse);
|
|
}
|
|
|
|
#if 1
|
|
void MLPhraseScorer::do_score_phrases() {}
|
|
#else
|
|
void MLPhraseScorer::do_score_phrases()
|
|
{
|
|
Score bla = 0;
|
|
Timestamp t_it;
|
|
for(Count i = 0; i < 200; i++) {
|
|
for(PhraseTable::iterator it = phrase_table_.begin(); it != phrase_table_.end(); ++it) {
|
|
PhrasePairInfo ppair = *it;
|
|
Phrase tgt = ppair.get_tgt();
|
|
bla += bla * ppair.get_count() / phrase_table_.get_tgt_phrase(tgt).get_count();
|
|
}
|
|
}
|
|
std::cerr << "Time for 200 iterations with ML estimation: " << (t_it.elapsed_time() / 1000) << " ms" << std::endl;
|
|
std::cerr << bla << std::endl;
|
|
}
|
|
#endif
|
|
|
|
Score MLPhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
return static_cast<Score>(it->get_count()) / tgt_phrase.get_count();
|
|
}
|
|
|
|
PhraseScorer *WittenBellPhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
return new WittenBellPhraseScorer(ptf.get_phrase_table(), reverse);
|
|
}
|
|
|
|
Score WittenBellPhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
return static_cast<Score>(it->get_count()) / (tgt_phrase.get_count() + tgt_phrase.get_distinct());
|
|
}
|
|
|
|
|
|
PhraseScorer *AbsoluteDiscountPhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
return new AbsoluteDiscountPhraseScorer(ptf.get_phrase_table(), reverse);
|
|
}
|
|
// p(s|t) = (c(s,t) - beta) / c(t) <-- absolute discounting
|
|
|
|
void AbsoluteDiscountPhraseScorer::do_score_phrases()
|
|
{
|
|
Count n1 = 0, n2 = 0;
|
|
|
|
for(PhraseTable::iterator it = phrase_table_.begin(); it != phrase_table_.end(); ++it) {
|
|
PhrasePairInfo ppinfo = *it;
|
|
Count c = ppinfo.get_count();
|
|
switch(c) {
|
|
case 1:
|
|
n1++;
|
|
break;
|
|
case 2:
|
|
n2++;
|
|
}
|
|
}
|
|
|
|
discount_ = static_cast<Score>(n1) / (n1 + 2*n2);
|
|
}
|
|
|
|
inline Score AbsoluteDiscountPhraseScorer::get_discount()
|
|
{
|
|
return discount_;
|
|
}
|
|
|
|
Score AbsoluteDiscountPhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
return (it->get_count() - discount_) / tgt_phrase.get_count();
|
|
}
|
|
|
|
PhraseScorer *KNDiscount1PhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
return new KNDiscount1PhraseScorer(ptf.get_phrase_table(), reverse);
|
|
}
|
|
|
|
|
|
void KNDiscount1PhraseScorer::do_score_phrases()
|
|
{
|
|
Count n1 = 0, n2 = 0;
|
|
Count total_count = 0;
|
|
|
|
for(PhraseTable::iterator it = phrase_table_.begin(); it != phrase_table_.end(); ++it) {
|
|
PhrasePairInfo ppinfo = *it;
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
total_count += tgt_phrase.get_count();
|
|
|
|
Count c = ppinfo.get_count();
|
|
switch(c) {
|
|
case 1:
|
|
n1++;
|
|
break;
|
|
case 2:
|
|
n2++;
|
|
}
|
|
}
|
|
|
|
discount_ = static_cast<Score>(n1) / (n1 + 2*n2);
|
|
total_count_ = static_cast<Count>(total_count);
|
|
|
|
}
|
|
|
|
Score KNDiscount1PhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
PhraseInfo &src_phrase = phrase_table_.get_src_phrase(it->get_src());
|
|
return ((it->get_count() - discount_) / tgt_phrase.get_count()) + (discount_ * tgt_phrase.get_distinct() / tgt_phrase.get_count())*(src_phrase.get_count() / total_count_);
|
|
}
|
|
|
|
PhraseScorer *KNDiscount3PhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
return new KNDiscount3PhraseScorer(ptf.get_phrase_table(), reverse);
|
|
}
|
|
|
|
|
|
void KNDiscount3PhraseScorer::do_score_phrases()
|
|
{
|
|
Count n1 = 0, n2 = 0, n3 = 0, n4 = 0;
|
|
Count total_count = 0; //total number of source or target phrases (including repetitions)
|
|
Count total_distinct_n1 = 0; //sum_{s} n1plus(s,*)
|
|
Count total_distinct_n2 = 0;
|
|
Count total_distinct_n3plus = 0;
|
|
Score y;
|
|
|
|
for(PhraseTable::iterator it = phrase_table_.begin(); it != phrase_table_.end(); ++it) {
|
|
PhrasePairInfo ppinfo = *it;
|
|
PhraseInfo &src_phrase = phrase_table_.get_src_phrase(it->get_src());
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
|
|
total_count += src_phrase.get_count();
|
|
|
|
Count c = ppinfo.get_count();
|
|
switch(c) {
|
|
case 1:
|
|
n1++;
|
|
tgt_phrase.inc_n1();
|
|
src_phrase.inc_n1();
|
|
total_distinct_n1++;
|
|
break;
|
|
case 2:
|
|
n2++;
|
|
tgt_phrase.inc_n2();
|
|
src_phrase.inc_n2();
|
|
total_distinct_n2++;
|
|
break;
|
|
case 3:
|
|
n3++;
|
|
tgt_phrase.inc_n3plus();
|
|
src_phrase.inc_n3plus();
|
|
total_distinct_n3plus++;
|
|
break;
|
|
case 4:
|
|
n4++;
|
|
tgt_phrase.inc_n3plus();
|
|
src_phrase.inc_n3plus();
|
|
total_distinct_n3plus++;
|
|
}
|
|
|
|
}
|
|
|
|
y = (Score)(n1) / (n1 + 2*n2);
|
|
discount1_ = static_cast<Score> (1) - (2)*(y)*(n2 / n1);
|
|
discount2_ = static_cast<Score> (2) - (3)*(y)*(n3 / n2);
|
|
discount3plus_ = static_cast<Score> (3) - (4)*(y)*(n4 / n3);
|
|
total_distinct_n1_ = static_cast<Count>(total_distinct_n1);
|
|
total_distinct_n2_ = static_cast<Count>(total_distinct_n2);
|
|
total_distinct_n3plus_ = static_cast<Count>(total_distinct_n3plus);
|
|
}
|
|
|
|
Score KNDiscount3PhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
PhrasePairInfo ppinfo = *it;
|
|
PhraseInfo &tgt_phrase = phrase_table_.get_tgt_phrase(it->get_tgt());
|
|
PhraseInfo &src_phrase = phrase_table_.get_src_phrase(it->get_src());
|
|
|
|
Score norm = (discount1_ * tgt_phrase.get_n1() + discount2_ * tgt_phrase.get_n2() + discount3plus_ * tgt_phrase.get_n3plus()) / tgt_phrase.get_count();
|
|
Count c = ppinfo.get_count();
|
|
switch(c) {
|
|
case 1:
|
|
return ((it->get_count() - discount1_) / tgt_phrase.get_count()) + \
|
|
norm*(src_phrase.get_n1() / total_distinct_n1_);
|
|
break;
|
|
case 2:
|
|
return ((it->get_count() - discount2_) / tgt_phrase.get_count()) + \
|
|
norm*(src_phrase.get_n2() / total_distinct_n2_);
|
|
break;
|
|
default:
|
|
return ((it->get_count() - discount3plus_) / tgt_phrase.get_count()) + \
|
|
norm*(src_phrase.get_n3plus() / total_distinct_n3plus_);
|
|
}
|
|
}
|
|
|
|
PhraseScorer *LexicalWeightPhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
bool overall_max = true;
|
|
|
|
if(argv[argp] == NULL)
|
|
usage();
|
|
|
|
if(!strcmp(argv[argp], "-AlignmentCount")) {
|
|
overall_max = false;
|
|
argp++;
|
|
if(argv[argp] == NULL)
|
|
usage();
|
|
}
|
|
|
|
String lwfile(argv[argp++]);
|
|
return new LexicalWeightPhraseScorer(ptf.get_phrase_table(), reverse, lwfile, overall_max);
|
|
}
|
|
|
|
LexicalWeightPhraseScorer::LexicalWeightPhraseScorer(PhraseTable &pd, bool reverse, const String &weightfile, bool overall_max) :
|
|
PhraseScorer(pd, reverse), overall_max_score_(overall_max), null_word_(PhraseText::index_word("NULL"))
|
|
{
|
|
std::ifstream wfile(weightfile.c_str());
|
|
|
|
while(!wfile.eof()) {
|
|
if(wfile.fail()) {
|
|
std::cerr << "Problem reading file: " << weightfile << std::endl;
|
|
exit(1);
|
|
}
|
|
|
|
String src, tgt;
|
|
Score weight;
|
|
|
|
wfile >> src >> tgt >> weight;
|
|
Count src_id = PhraseText::index_word(src);
|
|
Count tgt_id = PhraseText::index_word(tgt);
|
|
weight_map_.insert(std::make_pair(std::make_pair(src_id, tgt_id), weight));
|
|
}
|
|
|
|
wfile.close();
|
|
}
|
|
|
|
Score LexicalWeightPhraseScorer::get_weight(const String &s_src, const String &s_tgt) const
|
|
{
|
|
Count src = PhraseText::index_word(s_src);
|
|
Count tgt = PhraseText::index_word(s_tgt);
|
|
return get_weight(src, tgt);
|
|
}
|
|
|
|
inline Score LexicalWeightPhraseScorer::get_weight(Count src, Count tgt) const
|
|
{
|
|
WeightMapType_::const_iterator it = weight_map_.find(std::make_pair(src, tgt));
|
|
if(it == weight_map_.end())
|
|
return 0.00001; // default value copied from Philipp Koehn's scorer
|
|
return it->second;
|
|
}
|
|
|
|
#if 1
|
|
void LexicalWeightPhraseScorer::do_score_phrases() {}
|
|
#else
|
|
void LexicalWeightPhraseScorer::do_score_phrases()
|
|
{
|
|
Score bla = 0;
|
|
Timestamp t_it;
|
|
for(Count i = 0; i < 200; i++) {
|
|
for(PhraseTable::iterator it = phrase_table_.begin(); it != phrase_table_.end(); ++it) {
|
|
PhrasePairInfo ppair = *it;
|
|
Phrase src = ppair.get_src();
|
|
Phrase tgt = ppair.get_tgt();
|
|
bla += bla * get_score(src, tgt);
|
|
}
|
|
}
|
|
std::cerr << "Time for 200 iterations with lexical weights: " << (t_it.elapsed_time() / 1000) << " ms" << std::endl;
|
|
std::cerr << bla << std::endl;
|
|
}
|
|
#endif
|
|
|
|
Score LexicalWeightPhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
const Phrase src = it->get_src();
|
|
const Phrase tgt = it->get_tgt();
|
|
const PhraseText &src_phrase = phrase_table_.get_src_phrase(src).get_phrase();
|
|
const PhraseText &tgt_phrase = phrase_table_.get_tgt_phrase(tgt).get_phrase();
|
|
const PhrasePairInfo &ppair = *it;
|
|
|
|
Count max_count = 0;
|
|
|
|
Score maxlex = 0;
|
|
PhrasePairInfo::AlignmentVector av = ppair.get_alignments();
|
|
for(PhrasePairInfo::AlignmentVector::const_iterator it = av.begin(); it != av.end(); ++it) {
|
|
const PhraseAlignment &alig = it->first;
|
|
const Count alig_cnt = it->second;
|
|
|
|
assert(alig.get_source_length() == src_phrase.size() && alig.get_target_length() == tgt_phrase.size());
|
|
|
|
if(!overall_max_score_ && alig_cnt < max_count)
|
|
continue;
|
|
max_count = alig_cnt;
|
|
|
|
Score lex = 1;
|
|
for(Count s = 0; s < src_phrase.size(); s++) {
|
|
Score factor = 0;
|
|
Count na = 0;
|
|
for(Count t = 0; t < tgt_phrase.size(); t++)
|
|
if(alig.is_aligned(s, t)) {
|
|
const Score w = get_weight(src_phrase[s], tgt_phrase[t]);
|
|
factor += w;
|
|
na++;
|
|
}
|
|
|
|
if(na > 0)
|
|
lex *= factor / na;
|
|
else
|
|
lex *= get_weight(src_phrase[s], null_word_);
|
|
}
|
|
|
|
if(lex > maxlex)
|
|
maxlex = lex;
|
|
}
|
|
|
|
return maxlex;
|
|
}
|
|
|
|
PhraseScorer *ConstantPhraseScorer::create_scorer(const char *argv[], int &argp, bool reverse, const PhraseScorerFactory &ptf)
|
|
{
|
|
if(argv[argp] == NULL)
|
|
usage();
|
|
Score c = atof(argv[argp++]);
|
|
return new ConstantPhraseScorer(ptf.get_phrase_table(), reverse, c);
|
|
}
|
|
|
|
Score ConstantPhraseScorer::do_get_score(const PhraseTable::const_iterator &it)
|
|
{
|
|
return constant_;
|
|
}
|
|
|