Update README.md
Browse files
README.md
CHANGED
@@ -10,4 +10,78 @@ language:
|
|
10 |
- ca
|
11 |
---
|
12 |
# Catalan punctuation and capisalization restoration model
|
13 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
10 |
- ca
|
11 |
---
|
12 |
# Catalan punctuation and capisalization restoration model
|
13 |
+
## Details of the model
|
14 |
+
|
15 |
+
Explicarlo
|
16 |
+
|
17 |
+
## Details of the dataset
|
18 |
+
The dataset used for training the model has been XXXXXXX
|
19 |
+
|
20 |
+
## Evaluation Metrics
|
21 |
+
|
22 |
+
## Funding
|
23 |
+
|
24 |
+
## How to use the model
|
25 |
+
|
26 |
+
```py
|
27 |
+
from transformers import pipeline, AutoModelForTokenClassification, AutoTokenizer
|
28 |
+
import torch
|
29 |
+
|
30 |
+
def get_result_text_es_pt (list_entity, text, lang):
|
31 |
+
result_words = []
|
32 |
+
if lang == "es":
|
33 |
+
punc_tags = ['¿', '?', '¡', '!', ',', '.', ':']
|
34 |
+
else:
|
35 |
+
punc_tags = ['?', '!', ',', '.', ':']
|
36 |
+
|
37 |
+
for entity in list_entity:
|
38 |
+
tag = entity["entity"]
|
39 |
+
word = entity["word"]
|
40 |
+
start = entity["start"]
|
41 |
+
end = entity["end"]
|
42 |
+
|
43 |
+
# check punctuation
|
44 |
+
punc_in = next((p for p in punc_tags if p in tag), "")
|
45 |
+
|
46 |
+
subword = False
|
47 |
+
# check subwords
|
48 |
+
if word[0] == "#":
|
49 |
+
subword = True
|
50 |
+
if punc_in != "":
|
51 |
+
word = result_words[-1].replace(punc_in, "") + text[start:end]
|
52 |
+
else:
|
53 |
+
word = result_words[-1] + text[start:end]
|
54 |
+
|
55 |
+
if tag == "l":
|
56 |
+
word = word
|
57 |
+
elif tag == "u":
|
58 |
+
word = word.capitalize()
|
59 |
+
# case with punctuation
|
60 |
+
else:
|
61 |
+
if tag[-1] == "l":
|
62 |
+
word = (punc_in + word) if punc_in in ["¿", "¡"] else (word + punc_in)
|
63 |
+
elif tag[-1] == "u":
|
64 |
+
word = (punc_in + word.capitalize()) if punc_in in ["¿", "¡"] else (word.capitalize() + punc_in)
|
65 |
+
|
66 |
+
if subword == True:
|
67 |
+
result_words[-1] = word
|
68 |
+
else:
|
69 |
+
result_words.append(word)
|
70 |
+
|
71 |
+
return " ".join(result_words)
|
72 |
+
|
73 |
+
lang = "es"
|
74 |
+
model_path = "VOCALINLP/spanish_capitalization_punctuation_restoration_sanivert"
|
75 |
+
|
76 |
+
model = AutoModelForTokenClassification.from_pretrained(model_path)
|
77 |
+
tokenizer = AutoTokenizer.from_pretrained(model_path)
|
78 |
+
|
79 |
+
pipe = pipeline("token-classification", model=model, tokenizer=tokenizer)
|
80 |
+
text = "el paciente presenta los siguientes síntomas náuseas vértigo disnea fiebre y dolor abdominal"
|
81 |
+
result = pipe(text)
|
82 |
+
|
83 |
+
print("Source text: "+ text)
|
84 |
+
result_text = get_result_text_es_pt(result, text, lang)
|
85 |
+
print("Restored text: " +result_text)```
|
86 |
+
|
87 |
+
> Created by [VOCALI SISSTEMAS INTELIGENTES/@VOCALINLP](https://twitter.com/vocalinet)
|