/**
* Do the following pruning steps until nothing happens:
* power pp power pp power pp....
* Make sure you do them both at least once.
*/
void pp_and_power_prune(Sentence sent, Parse_Options opts)
{
power_prune(sent, opts);
pp_prune(sent, opts);
return;
// Not reached. We can actually gain a few percent of
// performance be skipping the loop below. Mostly, it just
// does a lot of work, and pretty much finds nothing.
// And so we skip it.
#ifdef ONLY_IF_YOU_THINK_THIS_IS_WORTH_IT
for (;;) {
if (pp_prune(sent, opts) == 0) break;
if (power_prune(sent, opts) == 0) break;
}
#endif
}
/**
* Prune useless disjuncts.
*/
void pp_and_power_prune(Sentence sent, Parse_Options opts)
{
power_table pt;
power_table_alloc(sent, &pt);
power_table_init(sent, &pt);
power_prune(sent, opts, &pt);
if (pp_prune(sent, opts) > 0)
power_prune(sent, opts, &pt);
/* No benefit for now to make additional pp_prune() & power_prune() -
* additional deletions are very rare and even then most of the
* times only one disjunct is deleted. */
power_table_delete(&pt);
return;
}
void my_prepare_to_parse(Sentence sent, Parse_Options opts) {
/* assumes that the sentence expression lists have been generated */
/* this does all the necessary pruning and building of and */
/* structures. */
int i, has_conjunction;
// build_sentence_disjuncts(sent, opts->disjunct_cost);
// if (verbosity > 2) {
//printf("After expanding expressions into disjuncts:") ;
//print_disjunct_counts(sent);
// }
print_time(opts, "Built disjuncts");
for (i=0; i<sent->length; i++) {
sent->word[i].d = eliminate_duplicate_disjuncts(sent->word[i].d);
}
print_time(opts, "Eliminated duplicate disjuncts");
if (verbosity > 2) {
printf("\nAfter expression pruning and duplicate elimination:\n");
print_disjunct_counts(sent);
}
null_links = (opts->min_null_count > 0);
has_conjunction = sentence_contains_conjunction(sent);
set_connector_length_limits(sent, opts);
build_deletable(sent, has_conjunction);
build_effective_dist(sent, has_conjunction);
/* why do we do these here instead of in
first_prepare_to_parse() only? The
reason is that the deletable region
depends on if null links are in use.
with null_links everything is deletable */
if (!has_conjunction) {
pp_and_power_prune(sent, RUTHLESS, opts);
} else {
pp_and_power_prune(sent, GENTLE, opts);
/*if (verbosity > 2) {
printf("\nAfter Gentle power pruning:\n");
print_disjunct_counts(sent);
}
*/
/*print_time(opts, "Finished gentle power pruning"); */
conjunction_prune(sent, opts);
if (verbosity > 2) {
printf("\nAfter conjunction pruning:\n");
print_disjunct_counts(sent);
print_statistics();
}
print_time(opts, "Done conjunction pruning");
build_conjunction_tables(sent);
install_fat_connectors(sent);
install_special_conjunctive_connectors(sent);
if (verbosity > 2) {
printf("After conjunctions, disjuncts counts:\n");
print_disjunct_counts(sent);
}
set_connector_length_limits(sent, opts);
/* have to do this again cause of the
new fat connectors and disjuncts */
print_time(opts, "Constructed fat disjuncts");
prune(sent);
print_time(opts, "Pruned fat disjuncts");
for (i=0; i<sent->length; i++) {
sent->word[i].d = eliminate_duplicate_disjuncts(sent->word[i].d);
}
if (verbosity > 2) {
printf("After pruning and duplicate elimination:\n");
print_disjunct_counts(sent);
}
print_time(opts, "Eliminated duplicate disjuncts (again)");
if (verbosity > 2) print_AND_statistics(sent);
power_prune(sent, RUTHLESS, opts);
}
/*
if (verbosity > 2) {
printf("\nAfter RUTHLESS power-pruning:\n");
print_disjunct_counts(sent);
}
*/
/* print time for power pruning used to be here */
/* now done in power_prune itself */
print_time(opts, "Initialized fast matcher and hash table");
}