diff options
Diffstat (limited to 'src/training/estimate_interpolation.cpp')
| -rw-r--r-- | src/training/estimate_interpolation.cpp | 151 |
1 files changed, 151 insertions, 0 deletions
diff --git a/src/training/estimate_interpolation.cpp b/src/training/estimate_interpolation.cpp new file mode 100644 index 0000000..1a547bc --- /dev/null +++ b/src/training/estimate_interpolation.cpp @@ -0,0 +1,151 @@ +/* + * novel-pinyin, + * A Simplified Chinese Sentence-Based Pinyin Input Method Engine + * Based On Markov Model. + * + * Copyright (C) 2006-2008 Peng Wu + * + * This program is free software; you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation; either version 2 of the License, or + * (at your option) any later version. + * + * This program 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 General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program; if not, write to the Free Software + * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA + */ + +#include <stdio.h> +#include <stdlib.h> +#include <string.h> +#include <locale.h> +#include <math.h> +#include <glib.h> +#include "memory_chunk.h" +#include "novel_types.h" +#include "phrase_index.h" +#include "ngram.h" + +parameter_t compute_interpolation(SingleGram * deleted_bigram, + FacadePhraseIndex * unigram, + SingleGram * bigram){ + bool success; + parameter_t lambda = 0, next_lambda = 0.6; + parameter_t epsilon = 0.001; + + while ( fabs(lambda - next_lambda) > epsilon){ + lambda = next_lambda; + next_lambda = 0; + guint32 table_num = 0; + parameter_t numerator = 0; + parameter_t part_of_denominator = 0; + + PhraseIndexRange range; + range.m_range_begin = token_min; + range.m_range_end = token_max; + + BigramPhraseArray array = g_array_new(FALSE, FALSE, sizeof(BigramPhraseItem)); + deleted_bigram->search(&range, array); + + for ( int i = 0; i < array->len; ++i){ + BigramPhraseItem * item = &g_array_index(array, BigramPhraseItem, i); + //get the phrase token + phrase_token_t token = item->m_token; + guint32 deleted_freq = 0; + assert(deleted_bigram->get_freq(token, deleted_freq)); + { + guint32 freq = 0; + parameter_t elem_poss = 0; + if ( bigram && bigram->get_freq(token, freq)){ + guint32 total_freq; + assert(bigram->get_total_freq(total_freq)); + assert(0 != total_freq); + elem_poss = freq / (parameter_t) total_freq; + } + numerator = lambda * elem_poss; + } + + { + guint32 freq = 0; + parameter_t elem_poss = 0; + PhraseItem item; + if (unigram->get_phrase_item(token, item)){ + guint32 freq = item.get_unigram_frequency(); + guint32 total_freq = unigram->get_phrase_index_total_freq(); + elem_poss = freq / (parameter_t)total_freq; + } + part_of_denominator = ( 1 - lambda) * elem_poss; + } + + if ( 0 == (numerator + part_of_denominator)) + continue; + + next_lambda += deleted_freq * (numerator / (numerator + part_of_denominator)); + } + assert(deleted_bigram->get_total_freq(table_num)); + next_lambda /= table_num; + + g_array_free(array, TRUE); + } + lambda = next_lambda; + return lambda; +} + +int main(int argc, char * argv[]){ + FacadePhraseIndex phrase_index; + + //gb_char binary file + MemoryChunk * chunk = new MemoryChunk; + chunk->load("../../data/gb_char.bin"); + phrase_index.load(1, chunk); + + //gbk_char binary file + chunk = new MemoryChunk; + chunk->load("../../data/gbk_char.bin"); + phrase_index.load(2, chunk); + + Bigram bigram; + bigram.attach("../../data/bigram.db", NULL); + + Bigram deleted_bigram; + deleted_bigram.attach("../../data/deleted_bigram.db", NULL); + + GArray * system_items = g_array_new(FALSE, FALSE, sizeof(phrase_token_t)); + GArray * user_items = g_array_new(FALSE, FALSE, sizeof(phrase_token_t)); + + deleted_bigram.get_all_items(system_items, user_items); + assert(0 == user_items->len); + g_array_free(user_items, TRUE); + + parameter_t lambda_sum = 0; + int lambda_count = 0; + + for ( int i = 0; i < system_items->len; ++i ){ + phrase_token_t * token = &g_array_index(system_items, phrase_token_t, i); + SingleGram * system = NULL, * user = NULL; + bigram.load(*token, system, user); + assert(NULL == user); + SingleGram * deleted_system = NULL, * deleted_user = NULL; + deleted_bigram.load(*token, deleted_system, deleted_user); + assert(NULL == deleted_user); + + parameter_t lambda = compute_interpolation(deleted_system, &phrase_index, system); + + printf("lambda:%f\n", lambda); + + lambda_sum += lambda; + lambda_count ++; + + if (system) delete system; + delete deleted_system; + } + + printf("average lambda:%f\n", (lambda_sum/lambda_count)); + g_array_free(system_items, TRUE); +} + |