File size: 9,480 Bytes
158b61b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 |
/***********************************************************************
Moses - factored phrase-based language decoder
Copyright (C) 2010 University of Edinburgh
This library is free software; you can redistribute it and/or
modify it under the terms of the GNU Lesser General Public
License as published by the Free Software Foundation; either
version 2.1 of the License, or (at your option) any later version.
This library is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
Lesser General Public License for more details.
You should have received a copy of the GNU Lesser General Public
License along with this library; if not, write to the Free Software
Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
***********************************************************************/
#define BOOST_TEST_MODULE BackwardTest
#include <boost/test/unit_test.hpp>
#include "lm/config.hh"
#include "lm/left.hh"
#include "lm/model.hh"
#include "lm/state.hh"
#include "moses/Sentence.h"
#include "moses/TypeDef.h"
#include "moses/StaticData.h"
#include "moses/parameters/AllOptions.h"
//#include "BackwardLMState.h"
#include "moses/LM/Backward.h"
#include "moses/LM/BackwardLMState.h"
#include "moses/Util.h"
#include "lm/state.hh"
#include "lm/left.hh"
#include <vector>
using namespace Moses;
//using namespace std;
/*
template <class M> void Foo() {
Moses::BackwardLanguageModel<M> *backwardLM;
// = new Moses::BackwardLanguageModel<M>( filename, factorType, lazy );
}
template <class M> void Everything() {
// Foo<M>();
}
*/
namespace Moses
{
// Apparently some Boost versions use templates and are pretty strict about types matching.
#define SLOPPY_CHECK_CLOSE(ref, value, tol) BOOST_CHECK_CLOSE(static_cast<double>(ref), static_cast<double>(value), static_cast<double>(tol));
AllOptions::ptr DefaultOptions(new AllOptions);
class BackwardLanguageModelTest
{
public:
BackwardLanguageModelTest() :
dummyInput(new Sentence(DefaultOptions)),
backwardLM(
static_cast< BackwardLanguageModel<lm::ngram::ProbingModel> * >(
ConstructBackwardLM(
"LM1=1.0",
boost::unit_test::framework::master_test_suite().argv[1],
0,
false)
)
) {
// This space intentionally left blank
}
~BackwardLanguageModelTest() {
delete dummyInput;
delete backwardLM;
}
void testEmptyHypothesis() {
FFState *ffState = const_cast< FFState * >(backwardLM->EmptyHypothesisState( *dummyInput ));
BOOST_CHECK( ffState != NULL );
delete ffState;
}
void testCalcScore() {
double p_the = -1.383059;
double p_licenses = -2.360783;
double p_for = -1.661813;
double p_most = -2.360783;
// double p_software = -1.62042;
double p_the_licenses = -0.9625873;
double p_licenses_for = -1.661557;
double p_for_most = -0.4526253;
// double p_most_software = -1.70295;
double p_the_licenses_for = p_the_licenses + p_licenses_for;
// double p_licenses_for_most = p_licenses_for + p_for_most;
// the
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"the",
NULL);
BOOST_CHECK( phrase.GetSize() == 1 );
float fullScore;
float ngramScore;
size_t oovCount;
backwardLM->CalcScore(phrase, fullScore, ngramScore, oovCount);
BOOST_CHECK( oovCount == 0 );
SLOPPY_CHECK_CLOSE( TransformLMScore(p_the), fullScore, 0.01);
SLOPPY_CHECK_CLOSE( TransformLMScore( 0.0 ), ngramScore, 0.01);
}
// the licenses
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"the licenses",
NULL);
BOOST_CHECK( phrase.GetSize() == 2 );
float fullScore;
float ngramScore;
size_t oovCount;
backwardLM->CalcScore(phrase, fullScore, ngramScore, oovCount);
BOOST_CHECK( oovCount == 0 );
SLOPPY_CHECK_CLOSE( TransformLMScore(p_licenses + p_the_licenses), fullScore, 0.01);
// Check ngramScore is 0.0
BOOST_CHECK_GT(0.0001, ngramScore);
BOOST_CHECK_LT(-0.0001, ngramScore);
}
// the licenses for
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"the licenses for",
NULL);
BOOST_CHECK( phrase.GetSize() == 3 );
float fullScore;
float ngramScore;
size_t oovCount;
backwardLM->CalcScore(phrase, fullScore, ngramScore, oovCount);
BOOST_CHECK( oovCount == 0 );
SLOPPY_CHECK_CLOSE( TransformLMScore( p_the_licenses_for ), ngramScore, 0.01);
SLOPPY_CHECK_CLOSE( TransformLMScore(p_for + p_licenses_for + p_the_licenses), fullScore, 0.01);
}
// the licenses for most
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"the licenses for most",
NULL);
BOOST_CHECK( phrase.GetSize() == 4 );
float fullScore;
float ngramScore;
size_t oovCount;
backwardLM->CalcScore(phrase, fullScore, ngramScore, oovCount);
BOOST_CHECK( oovCount == 0 );
SLOPPY_CHECK_CLOSE( TransformLMScore( p_the_licenses + p_licenses_for ), ngramScore, 0.01);
SLOPPY_CHECK_CLOSE( TransformLMScore(p_most + p_for_most + p_licenses_for + p_the_licenses), fullScore, 0.01);
}
}
void testEvaluate() {
FFState *nextState;
FFState *prevState = const_cast< FFState * >(backwardLM->EmptyHypothesisState( *dummyInput ));
double p_most = -2.360783;
double p_for = -1.661813;
double p_licenses = -2.360783;
double p_the = -1.383059;
double p_eos = -1.457693;
double p_most_for = -0.4526253;
double p_for_licenses = -1.661557;
double p_licenses_the = -0.9625873;
double p_the_eos = -1.940311;
// the
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"the",
NULL);
BOOST_CHECK( phrase.GetSize() == 1 );
float score;
nextState = backwardLM->Evaluate(phrase, prevState, score);
// p(the) * p(</s> | the) / p(</s>)
SLOPPY_CHECK_CLOSE( (p_the + p_the_eos - p_eos), score, 0.01);
delete prevState;
prevState = nextState;
}
// the licenses
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"licenses",
NULL);
BOOST_CHECK( phrase.GetSize() == 1 );
float score;
nextState = backwardLM->Evaluate(phrase, prevState, score);
// p(licenses) * p(licenses | the) / p(the)
SLOPPY_CHECK_CLOSE( (p_licenses + p_licenses_the - p_the), score, 0.01);
delete prevState;
prevState = nextState;
}
// the licenses for
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"for",
NULL);
BOOST_CHECK( phrase.GetSize() == 1 );
float score;
nextState = backwardLM->Evaluate(phrase, prevState, score);
// p(for) * p(for | licenses) / p(licenses)
SLOPPY_CHECK_CLOSE( (p_for + p_for_licenses - p_licenses), score, 0.01);
delete prevState;
prevState = nextState;
}
// the licenses for most
{
Phrase phrase;
BOOST_CHECK( phrase.GetSize() == 0 );
std::vector<FactorType> outputFactorOrder;
outputFactorOrder.push_back(0);
phrase.CreateFromString(
Input,
outputFactorOrder,
"most",
NULL);
BOOST_CHECK( phrase.GetSize() == 1 );
float score;
nextState = backwardLM->Evaluate(phrase, prevState, score);
// p(most) * p(most | for) / p(for)
SLOPPY_CHECK_CLOSE( (p_most + p_most_for - p_for), score, 0.01);
delete prevState;
prevState = nextState;
}
delete prevState;
}
private:
const Sentence *dummyInput;
BackwardLanguageModel<lm::ngram::ProbingModel> *backwardLM;
};
}
const char *FileLocation()
{
if (boost::unit_test::framework::master_test_suite().argc < 2) {
BOOST_FAIL("Jamfile must specify arpa file for this test, but did not");
}
return boost::unit_test::framework::master_test_suite().argv[1];
}
BOOST_AUTO_TEST_CASE(ProbingAll)
{
BackwardLanguageModelTest test;
test.testEmptyHypothesis();
test.testCalcScore();
test.testEvaluate();
}
|