victormiller
commited on
Commit
•
6475c3e
1
Parent(s):
5ab0a5d
Update curated.py
Browse files- curated.py +24 -24
curated.py
CHANGED
@@ -129,10 +129,10 @@ wikipedia_filter = pd.DataFrame(
|
|
129 |
"0.00%",
|
130 |
],
|
131 |
"Percent Removed After Local Dedup": [
|
132 |
-
"",
|
133 |
],
|
134 |
"Total Percentage Remaining": [
|
135 |
-
"",
|
136 |
],
|
137 |
}
|
138 |
)
|
@@ -187,10 +187,10 @@ dmm_filter = pd.DataFrame(
|
|
187 |
"0.00%",
|
188 |
],
|
189 |
"Percent Removed After Local Dedup": [
|
190 |
-
"",
|
191 |
],
|
192 |
"Total Percentage Remaining": [
|
193 |
-
"%",
|
194 |
],
|
195 |
}
|
196 |
)
|
@@ -217,10 +217,10 @@ uspto_filter = pd.DataFrame(
|
|
217 |
"0.01%",
|
218 |
],
|
219 |
"Percent Removed After Local Dedup": [
|
220 |
-
"",
|
221 |
],
|
222 |
"Total Percentage Remaining": [
|
223 |
-
"%",
|
224 |
],
|
225 |
}
|
226 |
)
|
@@ -246,10 +246,10 @@ pg19_filter = pd.DataFrame(
|
|
246 |
"0.17%",
|
247 |
],
|
248 |
"Percent Removed After Local Dedup": [
|
249 |
-
"",
|
250 |
],
|
251 |
"Total Percentage Remaining": [
|
252 |
-
"%",
|
253 |
],
|
254 |
}
|
255 |
)
|
@@ -276,10 +276,10 @@ hn_filter = pd.DataFrame(
|
|
276 |
"0.34%",
|
277 |
],
|
278 |
"Percent Removed After Local Dedup": [
|
279 |
-
"",
|
280 |
],
|
281 |
"Total Percentage Remaining": [
|
282 |
-
"%",
|
283 |
],
|
284 |
}
|
285 |
)
|
@@ -306,10 +306,10 @@ uirc_filter = pd.DataFrame(
|
|
306 |
"1.12%",
|
307 |
],
|
308 |
"Percent Removed After Local Dedup": [
|
309 |
-
"",
|
310 |
],
|
311 |
"Total Percentage Remaining": [
|
312 |
-
"%",
|
313 |
],
|
314 |
}
|
315 |
)
|
@@ -335,10 +335,10 @@ up_filter = pd.DataFrame(
|
|
335 |
"0.00%",
|
336 |
],
|
337 |
"Percent Removed After Local Dedup": [
|
338 |
-
"",
|
339 |
],
|
340 |
"Total Percentage Remaining": [
|
341 |
-
"%",
|
342 |
],
|
343 |
}
|
344 |
)
|
@@ -364,10 +364,10 @@ se_filter = pd.DataFrame(
|
|
364 |
"0.00%",
|
365 |
],
|
366 |
"Percent Removed After Local Dedup": [
|
367 |
-
"",
|
368 |
],
|
369 |
"Total Percentage Remaining": [
|
370 |
-
"%",
|
371 |
],
|
372 |
}
|
373 |
)
|
@@ -393,10 +393,10 @@ arx_filter = pd.DataFrame(
|
|
393 |
"0.07%",
|
394 |
],
|
395 |
"Percent Removed After Local Dedup": [
|
396 |
-
"",
|
397 |
],
|
398 |
"Total Percentage Remaining": [
|
399 |
-
"%",
|
400 |
],
|
401 |
}
|
402 |
)
|
@@ -422,10 +422,10 @@ s2o_filter = pd.DataFrame(
|
|
422 |
"0.00%",
|
423 |
],
|
424 |
"Percent Removed After Local Dedup": [
|
425 |
-
"",
|
426 |
],
|
427 |
"Total Percentage Remaining": [
|
428 |
-
"%",
|
429 |
],
|
430 |
}
|
431 |
)
|
@@ -451,10 +451,10 @@ med_filter = pd.DataFrame(
|
|
451 |
"0.02%",
|
452 |
],
|
453 |
"Percent Removed After Local Dedup": [
|
454 |
-
"",
|
455 |
],
|
456 |
"Total Percentage Remaining": [
|
457 |
-
"%",
|
458 |
],
|
459 |
}
|
460 |
)
|
@@ -480,10 +480,10 @@ phil_filter = pd.DataFrame(
|
|
480 |
"0.12%",
|
481 |
],
|
482 |
"Percent Removed After Local Dedup": [
|
483 |
-
"",
|
484 |
],
|
485 |
"Total Percentage Remaining": [
|
486 |
-
"%",
|
487 |
],
|
488 |
}
|
489 |
)
|
|
|
129 |
"0.00%",
|
130 |
],
|
131 |
"Percent Removed After Local Dedup": [
|
132 |
+
"0.31%",
|
133 |
],
|
134 |
"Total Percentage Remaining": [
|
135 |
+
"97.84%",
|
136 |
],
|
137 |
}
|
138 |
)
|
|
|
187 |
"0.00%",
|
188 |
],
|
189 |
"Percent Removed After Local Dedup": [
|
190 |
+
"0.00%",
|
191 |
],
|
192 |
"Total Percentage Remaining": [
|
193 |
+
"100.00%",
|
194 |
],
|
195 |
}
|
196 |
)
|
|
|
217 |
"0.01%",
|
218 |
],
|
219 |
"Percent Removed After Local Dedup": [
|
220 |
+
"22.94%",
|
221 |
],
|
222 |
"Total Percentage Remaining": [
|
223 |
+
"75.60%",
|
224 |
],
|
225 |
}
|
226 |
)
|
|
|
246 |
"0.17%",
|
247 |
],
|
248 |
"Percent Removed After Local Dedup": [
|
249 |
+
"0.80%",
|
250 |
],
|
251 |
"Total Percentage Remaining": [
|
252 |
+
"98.78%",
|
253 |
],
|
254 |
}
|
255 |
)
|
|
|
276 |
"0.34%",
|
277 |
],
|
278 |
"Percent Removed After Local Dedup": [
|
279 |
+
"61.84%",
|
280 |
],
|
281 |
"Total Percentage Remaining": [
|
282 |
+
"37.03%",
|
283 |
],
|
284 |
}
|
285 |
)
|
|
|
306 |
"1.12%",
|
307 |
],
|
308 |
"Percent Removed After Local Dedup": [
|
309 |
+
"0.66%",
|
310 |
],
|
311 |
"Total Percentage Remaining": [
|
312 |
+
"60.72%",
|
313 |
],
|
314 |
}
|
315 |
)
|
|
|
335 |
"0.00%",
|
336 |
],
|
337 |
"Percent Removed After Local Dedup": [
|
338 |
+
"1.00%",
|
339 |
],
|
340 |
"Total Percentage Remaining": [
|
341 |
+
"99.00%",
|
342 |
],
|
343 |
}
|
344 |
)
|
|
|
364 |
"0.00%",
|
365 |
],
|
366 |
"Percent Removed After Local Dedup": [
|
367 |
+
"0.00%",
|
368 |
],
|
369 |
"Total Percentage Remaining": [
|
370 |
+
"100.00%",
|
371 |
],
|
372 |
}
|
373 |
)
|
|
|
393 |
"0.07%",
|
394 |
],
|
395 |
"Percent Removed After Local Dedup": [
|
396 |
+
"0.00%",
|
397 |
],
|
398 |
"Total Percentage Remaining": [
|
399 |
+
"92.20%",
|
400 |
],
|
401 |
}
|
402 |
)
|
|
|
422 |
"0.00%",
|
423 |
],
|
424 |
"Percent Removed After Local Dedup": [
|
425 |
+
"0.00%",
|
426 |
],
|
427 |
"Total Percentage Remaining": [
|
428 |
+
"100.00%",
|
429 |
],
|
430 |
}
|
431 |
)
|
|
|
451 |
"0.02%",
|
452 |
],
|
453 |
"Percent Removed After Local Dedup": [
|
454 |
+
"0.00%",
|
455 |
],
|
456 |
"Total Percentage Remaining": [
|
457 |
+
"91.14%",
|
458 |
],
|
459 |
}
|
460 |
)
|
|
|
480 |
"0.12%",
|
481 |
],
|
482 |
"Percent Removed After Local Dedup": [
|
483 |
+
"0.00%",
|
484 |
],
|
485 |
"Total Percentage Remaining": [
|
486 |
+
"79.22%",
|
487 |
],
|
488 |
}
|
489 |
)
|