explosion/spaCy
1
1
Fork 0
mirror of https://github.com/explosion/spaCy.git synced 2024-12-25 09:26:27 +03:00
Code Issues Packages Projects Releases Wiki Activity

* Adjust scorer to account for tokenization mistakes

Browse Source
This commit is contained in:
Matthew Honnibal 2015-03-24 04:26:37 +01:00
parent 221f43c370
commit 2e12dec76e
1 changed files with 7 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

9
spacy/scorer.py
View File

@ -9,11 +9,13 @@ class Scorer(object):
self.ents_fp = 0
self.ents_fn = 0
self.total = 1e-100
self.mistokened = 0
self.n_tokens = 0
self.eval_punct = eval_punct
@property
def tags_acc(self):
return (self.tags_corr / self.total) * 100
return ((self.tags_corr - self.mistokened) / (self.n_tokens - self.mistokened)) * 100
@property
def uas(self):
@ -39,12 +41,15 @@ class Scorer(object):
assert len(tokens) == len(gold)
for i, token in enumerate(tokens):
if gold.orths.get(token.idx) != token.orth_:
self.mistokened += 1
if not self.skip_token(i, token, gold):
self.total += 1
if token.head.i == gold.heads[i]:
self.heads_corr += 1
self.labels_corr += token.dep_ == gold.labels[i]
self.tags_corr += token.tag_ == gold.tags[i]
self.tags_corr += token.tag_ == gold.tags[i]
self.n_tokens += 1
gold_ents = set((start, end, label) for (start, end, label) in gold.ents)
guess_ents = set(tokens.ents)
if verbose and gold_ents: