ModErn Text Analysis
META Enumerates Textual Applications
|
A re-implementation of (some of) the evalb metrics. More...
#include <evalb.h>
Public Member Functions | |
uint64_t | matched () const |
uint64_t | proposed_total () const |
uint64_t | gold_total () const |
double | labeled_precision () const |
double | labeled_recall () const |
double | labeled_f1 () const |
double | perfect () const |
double | average_crossing () const |
double | zero_crossing () const |
void | add_tree (parse_tree proposed, parse_tree gold) |
Private Attributes | |
uint64_t | proposed_correct_ = 0 |
The number of correct constituents in proposed parse trees. | |
uint64_t | proposed_total_ = 0 |
The number of total constituents in proposed parse trees. | |
uint64_t | gold_total_ = 0 |
The number of total constituents in gold parse trees. | |
uint64_t | perfect_ = 0 |
The number of parse trees that were perfect matches with gold trees. | |
uint64_t | crossed_ = 0 |
The total number of crossings in proposed trees. | |
uint64_t | zero_crossing_ = 0 |
The total number of proposed trees that had no crossings. | |
uint64_t | total_trees_ = 0 |
The total number of parse trees. | |
A re-implementation of (some of) the evalb metrics.
You should, of course, always double check with the real evalb in any paper results, but this can be used internally for e.g. convergence testing on a dev set.
|
inline |
|
inline |
|
inline |
|
inline |
|
inline |
|
inline |
|
inline |
|
inline |
|
inline |
void meta::parser::evalb::add_tree | ( | parse_tree | proposed, |
parse_tree | gold | ||
) |
proposed | The proposed parse |
gold | The gold standard parse |