schlevik
commited on
Commit
·
35b2ba6
1
Parent(s):
7b45cb5
fix text script
Browse files
cas.py
CHANGED
@@ -168,95 +168,122 @@ class CAS(datasets.GeneratorBasedBuilder):
|
|
168 |
|
169 |
def _generate_examples(self, datadir):
|
170 |
key = 0
|
171 |
-
for file in ["CAS_neg.txt", "CAS_spec.txt"]:
|
172 |
-
|
173 |
-
|
174 |
-
|
175 |
-
|
176 |
-
|
177 |
-
|
178 |
-
|
|
|
|
|
|
|
|
|
|
|
179 |
|
180 |
-
|
181 |
-
|
182 |
-
|
183 |
-
|
184 |
-
|
185 |
-
|
186 |
-
|
187 |
-
|
188 |
-
|
|
|
189 |
|
190 |
-
|
191 |
-
|
192 |
-
"
|
193 |
-
|
194 |
-
|
195 |
-
|
196 |
-
|
197 |
-
|
198 |
-
|
199 |
-
|
200 |
-
|
201 |
-
|
202 |
-
|
203 |
-
|
204 |
-
|
205 |
-
|
206 |
-
|
207 |
-
|
208 |
-
|
209 |
-
|
210 |
-
|
211 |
-
|
212 |
-
|
213 |
-
|
214 |
-
|
215 |
-
|
216 |
-
|
217 |
-
|
218 |
-
|
219 |
-
|
220 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
221 |
}
|
222 |
-
|
223 |
-
|
224 |
-
|
225 |
-
idces = np.argwhere(dic["id_docs"] == doc_id)[:, 0]
|
226 |
-
text = [dic["words"][id] for id in idces]
|
227 |
-
POS_tags_ = [dic["POS_tags"][id] for id in idces]
|
228 |
|
229 |
-
|
230 |
-
|
231 |
-
"
|
232 |
-
"
|
233 |
-
"
|
234 |
-
"
|
235 |
-
"
|
236 |
-
"coreferences": [],
|
237 |
}
|
|
|
238 |
key += 1
|
239 |
|
240 |
-
data["passages"] = [
|
241 |
-
{
|
242 |
-
"id": str(key + i),
|
243 |
-
"type": "sentence",
|
244 |
-
"text": [text[i]],
|
245 |
-
"offsets": [[i, i + 1]],
|
246 |
-
}
|
247 |
-
for i in range(len(text))
|
248 |
-
]
|
249 |
-
key += len(text)
|
250 |
-
|
251 |
-
for i in range(len(text)):
|
252 |
-
entity = {
|
253 |
-
"id": key,
|
254 |
-
"type": "POS_tag",
|
255 |
-
"text": [POS_tags_[i]],
|
256 |
-
"offsets": [[i, i + 1]],
|
257 |
-
"normalized": [],
|
258 |
-
}
|
259 |
-
data["entities"].append(entity)
|
260 |
-
key += 1
|
261 |
-
|
262 |
yield key, data
|
|
|
168 |
|
169 |
def _generate_examples(self, datadir):
|
170 |
key = 0
|
171 |
+
# for file in ["CAS_neg.txt", "CAS_spec.txt"]:
|
172 |
+
file = 'CAS_neg.txt'
|
173 |
+
filepath = os.path.join(datadir, file)
|
174 |
+
filepath2 = os.path.join(datadir, 'CAS_spec.txt')
|
175 |
+
label = "negation" if "neg" in file else "speculation"
|
176 |
+
id_docs = []
|
177 |
+
id_docs_2 = []
|
178 |
+
id_words = []
|
179 |
+
words = []
|
180 |
+
lemmas = []
|
181 |
+
POS_tags = []
|
182 |
+
NER_tags = []
|
183 |
+
NER_tags_2 = []
|
184 |
|
185 |
+
with open(filepath) as f:
|
186 |
+
for line in f.readlines():
|
187 |
+
line_content = line.split("\t")
|
188 |
+
if len(line_content) > 1:
|
189 |
+
id_docs.append(line_content[0])
|
190 |
+
id_words.append(line_content[1])
|
191 |
+
words.append(line_content[2])
|
192 |
+
lemmas.append(line_content[3])
|
193 |
+
POS_tags.append(line_content[4])
|
194 |
+
NER_tags.append(line_content[5].strip())
|
195 |
|
196 |
+
with open(filepath2) as f:
|
197 |
+
for line in f.readlines():
|
198 |
+
line_content = line.split("\t")
|
199 |
+
if len(line_content) > 1:
|
200 |
+
id_docs_2.append(line_content[0])
|
201 |
+
NER_tags_2.append(line_content[5].strip())
|
202 |
+
|
203 |
+
dic = {
|
204 |
+
"id_docs": np.array(list(map(int, id_docs))),
|
205 |
+
"id_words": id_words,
|
206 |
+
"words": words,
|
207 |
+
"lemmas": lemmas,
|
208 |
+
"POS_tags": POS_tags,
|
209 |
+
"NER_tags": NER_tags
|
210 |
+
}
|
211 |
+
dic2 = {
|
212 |
+
"id_docs": np.array(list(map(int, id_docs_2))),
|
213 |
+
"NER_tags": NER_tags_2
|
214 |
+
}
|
215 |
+
if self.config.schema == "source":
|
216 |
+
for doc_id in set(dic["id_docs"]):
|
217 |
+
idces = np.argwhere(dic["id_docs"] == doc_id)[:, 0]
|
218 |
+
text = [dic["words"][id] for id in idces]
|
219 |
+
text_lemmas = [dic["lemmas"][id] for id in idces]
|
220 |
+
POS_tags_ = [dic["POS_tags"][id] for id in idces]
|
221 |
+
yield key, {
|
222 |
+
"id": key,
|
223 |
+
"document_id": doc_id,
|
224 |
+
"text": text,
|
225 |
+
"lemmas": text_lemmas,
|
226 |
+
"POS_tags": POS_tags_,
|
227 |
+
"labels": [label],
|
228 |
+
}
|
229 |
+
key += 1
|
230 |
+
elif self.config.schema == "bigbio_text":
|
231 |
+
for doc_id in set(dic["id_docs"]):
|
232 |
+
idces = np.argwhere(dic["id_docs"] == doc_id)[:, 0]
|
233 |
+
idces_2 = np.argwhere(dic2["id_docs"] == doc_id)[:, 0]
|
234 |
+
|
235 |
+
text = " ".join([dic["words"][id] for id in idces])
|
236 |
+
label_tokens = [dic["NER_tags"][id] for id in idces]
|
237 |
+
label2_tokens = [dic2["NER_tags"][id] for id in idces_2]
|
238 |
+
label_ = []
|
239 |
+
if not all(l == '***' for l in label_tokens):
|
240 |
+
label_.append("negation")
|
241 |
+
if not all(l == '***' for l in label2_tokens):
|
242 |
+
label_.append("speculation")
|
243 |
+
yield key, {
|
244 |
+
"id": key,
|
245 |
+
"document_id": doc_id,
|
246 |
+
"text": text,
|
247 |
+
"labels": label_,
|
248 |
+
}
|
249 |
+
key += 1
|
250 |
+
elif self.config.schema == "bigbio_kb":
|
251 |
+
for doc_id in set(dic["id_docs"]):
|
252 |
+
idces = np.argwhere(dic["id_docs"] == doc_id)[:, 0]
|
253 |
+
text = [dic["words"][id] for id in idces]
|
254 |
+
POS_tags_ = [dic["POS_tags"][id] for id in idces]
|
255 |
+
|
256 |
+
data = {
|
257 |
+
"id": str(key),
|
258 |
+
"document_id": doc_id,
|
259 |
+
"passages": [],
|
260 |
+
"entities": [],
|
261 |
+
"relations": [],
|
262 |
+
"events": [],
|
263 |
+
"coreferences": [],
|
264 |
+
}
|
265 |
+
key += 1
|
266 |
+
|
267 |
+
data["passages"] = [
|
268 |
+
{
|
269 |
+
"id": str(key + i),
|
270 |
+
"type": "sentence",
|
271 |
+
"text": [text[i]],
|
272 |
+
"offsets": [[i, i + 1]],
|
273 |
}
|
274 |
+
for i in range(len(text))
|
275 |
+
]
|
276 |
+
key += len(text)
|
|
|
|
|
|
|
277 |
|
278 |
+
for i in range(len(text)):
|
279 |
+
entity = {
|
280 |
+
"id": key,
|
281 |
+
"type": "POS_tag",
|
282 |
+
"text": [POS_tags_[i]],
|
283 |
+
"offsets": [[i, i + 1]],
|
284 |
+
"normalized": [],
|
|
|
285 |
}
|
286 |
+
data["entities"].append(entity)
|
287 |
key += 1
|
288 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
289 |
yield key, data
|