Spaces:
Runtime error
Runtime error
BerserkerMother
commited on
Commit
•
24c3ed3
1
Parent(s):
27c530a
Adds parser package
Browse filesThis package is responsible for parsing the user prompts into jobs that
are carried by mather to find the best matches.
- elise/src/parser/__init__.py +1 -0
- elise/src/parser/parser.py +60 -0
elise/src/parser/__init__.py
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
from .parser import SentenceParser
|
elise/src/parser/parser.py
ADDED
@@ -0,0 +1,60 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import List, Union
|
2 |
+
|
3 |
+
import transformers
|
4 |
+
|
5 |
+
|
6 |
+
class SentenceParser:
|
7 |
+
mappings = {
|
8 |
+
"Amenity": "Semantic",
|
9 |
+
"Price": "Price",
|
10 |
+
"Hours": "Hours",
|
11 |
+
"Dish": "Menu",
|
12 |
+
"Restaurant_Name": "Name",
|
13 |
+
"Location": "Location",
|
14 |
+
"Cuisine": "Semantic",
|
15 |
+
}
|
16 |
+
|
17 |
+
def __init__(self, pipe):
|
18 |
+
self.pipe = pipe
|
19 |
+
|
20 |
+
@classmethod
|
21 |
+
def from_huggingface(cls, model_path: str):
|
22 |
+
token_classifier = transformers.pipeline(
|
23 |
+
"token-classification",
|
24 |
+
model=model_path,
|
25 |
+
aggregation_strategy="simple",
|
26 |
+
)
|
27 |
+
return cls(token_classifier)
|
28 |
+
|
29 |
+
def get_ner(self, sentences: Union[str, List[str]]):
|
30 |
+
"""Identifies the user intents from prompts
|
31 |
+
returns: list of dictionaries{category: words}
|
32 |
+
"""
|
33 |
+
if isinstance(sentences, str):
|
34 |
+
sentences = [sentences]
|
35 |
+
# format output
|
36 |
+
list_out = []
|
37 |
+
for items in self.pipe(sentences):
|
38 |
+
sentence_ner = {}
|
39 |
+
for recognized_token in items:
|
40 |
+
if recognized_token["entity_group"] in list(sentence_ner.keys()):
|
41 |
+
sentence_ner[recognized_token["entity_group"]].append(
|
42 |
+
recognized_token["word"]
|
43 |
+
)
|
44 |
+
else:
|
45 |
+
sentence_ner[recognized_token["entity_group"]] = [
|
46 |
+
recognized_token["word"]
|
47 |
+
]
|
48 |
+
list_out.append(sentence_ner)
|
49 |
+
return list_out
|
50 |
+
|
51 |
+
def get_jobs(self, ners):
|
52 |
+
"""Maps user's intents to Matcher jobs"""
|
53 |
+
list_out = []
|
54 |
+
for item in ners:
|
55 |
+
jobs = {}
|
56 |
+
for ner, value in item.items():
|
57 |
+
if ner in self.mappings.keys():
|
58 |
+
jobs[self.mappings[ner]] = value
|
59 |
+
list_out.append(jobs)
|
60 |
+
return list_out
|