const char * linkage_get_disjunct_str(const Linkage linkage, WordIdx w)
{
Disjunct *dj;
if (NULL == linkage) return "";
if (NULL == linkage->disjunct_list_str)
{
lg_compute_disjunct_strings(linkage);
}
if (linkage->num_words <= w) return NULL; /* bounds-check */
/* dj will be null if the word wasn't used in the parse. */
dj = linkage->chosen_disjuncts[w];
if (NULL == dj) return "";
return linkage->disjunct_list_str[w];
}
void lg_corpus_linkage_senses(Linkage linkage)
{
const char * infword;
Sentence sent = linkage->sent;
Dictionary dict = sent->dict;
Corpus *corp = dict->corpus;
int nwords = sent->length;
Linkage_info *lifo = linkage->info;
int w;
if (lifo->sense_list) return;
/* Set up the disjunct strings first */
lg_compute_disjunct_strings(sent, lifo);
lifo->nwords = nwords;
lifo->sense_list = (Sense **) malloc(nwords * sizeof (Sense *));
memset(lifo->sense_list, 0, nwords * sizeof (Sense *));
/* Decrement nwords, so as to ignore the RIGHT-WALL */
nwords --;
/* Loop over each word in the sentence (skipping LEFT-WALL, which is
* word 0. */
for (w=1; w<nwords; w++)
{
Disjunct *disj = sent->parse_info->chosen_disjuncts[w];
/* disj is NULL if word did not participate in parse */
if (NULL == disj)
{
continue;
}
infword = disj->string;
lifo->sense_list[w] = lg_corpus_senses(corp, infword,
lifo->disjunct_list_str[w], w);
}
}
/**
* lg_corpus_score -- compute parse-ranking score for sentence.
*
* Given a parsed sentence, this routine will compute a parse ranking
* score, based on the probabilites of observing the indicated set of
* disjuncts in the statistics database.
*
* The score is stored in the Linkage_info->corpus_cost struct member.
*
* The score is currently computed as the average -log_2 conditional
* probability p(d|w) of observing disjunct 'd', given word 'w'.
* Lower scores are better -- they indicate more likely parses.
*/
void lg_corpus_score(Sentence sent, Linkage_info *lifo)
{
const char *infword, *djstr;
double tot_score = 0.0f;
Corpus *corp = sent->dict->corpus;
int nwords = sent->length;
int w;
/* No-op if the database is not open */
if (NULL == corp->dbconn) return;
lg_compute_disjunct_strings(sent, lifo);
/* Decrement nwords, so as to ignore the RIGHT-WALL */
nwords --;
/* Loop over each word in the sentence (skipping LEFT-WALL, which is
* word 0. */
for (w=1; w<nwords; w++)
{
Disjunct *disj = sent->parse_info->chosen_disjuncts[w];
/* disj is NULL if word did not participate in parse */
if (NULL == disj)
{
tot_score += LOW_SCORE;
continue;
}
infword = disj->string;
djstr = lifo->disjunct_list_str[w];
tot_score += get_disjunct_score(corp, infword, djstr);
}
/* Decrement nwords, so as to ignore the LEFT-WALL */
--nwords;
tot_score /= nwords;
lifo->corpus_cost = tot_score;
}
double lg_corpus_disjunct_score(Linkage linkage, int w)
{
double score;
const char *infword, *djstr;
Sentence sent = linkage->sent;
Linkage_info *lifo = linkage->info;
Corpus *corp = sent->dict->corpus;
Disjunct *disj;
/* No-op if the database is not open */
if (NULL == corp->dbconn) return LOW_SCORE;
/* disj is NULL if word did not participate in parse */
disj = sent->parse_info->chosen_disjuncts[w];
if (NULL == disj) return LOW_SCORE;
lg_compute_disjunct_strings(sent, lifo);
infword = disj->string;
djstr = lifo->disjunct_list_str[w];
score = get_disjunct_score(corp, infword, djstr);
return score;
}
