Hugging Face's logo

Spaces:
darylfunggg
/
text-analysis
Running

App Files Community
Daryl Fung commited on
Commit
f696d1d
·
1 Parent(s): aa83e86

change font size

Browse files
Files changed (11) hide show
  1. keyphrase_extraction.py +1 -1
  2. keyword_extraction.py +1 -1
  3. results/DAD/1_gram.png +0 -0
  4. results/DAD/1_keyword.png +0 -0
  5. results/DAD/2_gram.png +0 -0
  6. results/DAD/2_keyword.png +0 -0
  7. results/DAD/3_gram.png +0 -0
  8. results/DAD/3_keyword.png +0 -0
  9. results/DAD/key_phrase.html +53 -59
  10. results/DAD/top_keyphrase.png +0 -0
  11. word.py +1 -1
keyphrase_extraction.py CHANGED
@@ -35,8 +35,8 @@ def get_top_key_phrases(text, top_n, save_output):
35
  keyword_df = pd.DataFrame({'words': phrases[:top_n], 'scores': scores[:top_n]})
36
  plt.title("Word Count")
37
  plt.figure(figsize=(24, 8))
 
38
  sns.barplot(data=keyword_df, y='words', x='scores', palette='blend:#7AB,#EDA')
39
- plt.xticks(rotation=-10, fontsize=8)
40
  plt.savefig(save_output, dpi=300)
41
  plt.close()
42
 
 
35
  keyword_df = pd.DataFrame({'words': phrases[:top_n], 'scores': scores[:top_n]})
36
  plt.title("Word Count")
37
  plt.figure(figsize=(24, 8))
38
+ plt.yticks(fontsize=14)
39
  sns.barplot(data=keyword_df, y='words', x='scores', palette='blend:#7AB,#EDA')
 
40
  plt.savefig(save_output, dpi=300)
41
  plt.close()
42
 
keyword_extraction.py CHANGED
@@ -38,8 +38,8 @@ def keyword_extract(doc, kw_model, n_grams, save_output='results/'):
38
  keyword_df = pd.DataFrame({'words': words, 'scores': scores})
39
  plt.title("Word Count")
40
  plt.figure(figsize=(24, 8))
 
41
  sns.barplot(data=keyword_df, y='words', x='scores', palette='blend:#7AB,#EDA')
42
- plt.xticks(rotation=-10, fontsize=12-n_grams)
43
  plt.savefig(save_output, dpi=300)
44
  plt.close()
45
 
 
38
  keyword_df = pd.DataFrame({'words': words, 'scores': scores})
39
  plt.title("Word Count")
40
  plt.figure(figsize=(24, 8))
41
+ plt.yticks(fontsize=14)
42
  sns.barplot(data=keyword_df, y='words', x='scores', palette='blend:#7AB,#EDA')
 
43
  plt.savefig(save_output, dpi=300)
44
  plt.close()
45
 
results/DAD/1_gram.png CHANGED
results/DAD/1_keyword.png CHANGED
results/DAD/2_gram.png CHANGED
results/DAD/2_keyword.png CHANGED
results/DAD/3_gram.png CHANGED
results/DAD/3_keyword.png CHANGED
results/DAD/key_phrase.html CHANGED
@@ -6,11 +6,10 @@
6
 
7
  <body style="font-size: 16px; font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol'; padding: 4rem 2rem; direction: ltr">
8
  <figure style="margin-bottom: 6rem">
9
- <div class="entities" style="line-height: 2.5; direction: ltr">
10
- <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
11
- CIHI
12
- Database
13
- <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
14
  </mark>
15
 
16
  <mark class="entity" style="background: hsl(121, 100%, 100.0%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
@@ -54,10 +53,10 @@ Database
54
  </mark>
55
  to capture
56
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
57
- day surgery
58
-
59
-
60
-
61
  Alberta
62
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
63
  </mark>
@@ -66,7 +65,7 @@ Alberta
66
  Alberta
67
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
68
  </mark>
69
- </br>
70
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
71
  The discharge abstract database
72
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
@@ -103,11 +102,11 @@ Alberta
103
  </mark>
104
  and
105
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
106
- transfers
107
-
108
-
109
-
110
- British Columbia
111
  Data
112
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
113
  </mark>
@@ -116,7 +115,7 @@ Data
116
  British Columbia
117
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
118
  </mark>
119
- </br>Data on
120
  <mark class="entity" style="background: hsl(121, 100%, 59.09090909090909%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
121
  discharges transfers
122
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.09</span>
@@ -267,9 +266,9 @@ Data
267
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
268
  </mark>
269
  basis to the respective
270
- <mark class="entity" style="background: hsl(121, 100%, 86.36363636363636%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
271
  Ministries of Health The DAD
272
- <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
273
  </mark>
274
 
275
  <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
@@ -311,12 +310,7 @@ Data
311
  year
312
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
313
  </mark>
314
- </br></br></br></br>
315
- <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
316
- Manitoba
317
- <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
318
- </mark>
319
- </br>Health data maintained by
320
  <mark class="entity" style="background: hsl(121, 100%, 63.63636363636363%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
321
  Manitoba Health
322
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.08</span>
@@ -463,11 +457,11 @@ Data
463
  </mark>
464
  about
465
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
466
- inpatient and day surgery services
467
-
468
-
469
-
470
- New Brunswick
471
  Patient
472
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
473
  </mark>
@@ -476,7 +470,7 @@ Patient
476
  New Brunswick
477
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
478
  </mark>
479
- </br>Patient
480
  <mark class="entity" style="background: hsl(121, 100%, 50.0%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
481
  discharge information
482
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.11</span>
@@ -513,10 +507,10 @@ Patient
513
  </mark>
514
  and
515
  <mark class="entity" style="background: hsl(121, 100%, 86.36363636363636%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
516
- transfers
517
-
518
-
519
-
520
  Newfoundland
521
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
522
  </mark>
@@ -535,7 +529,7 @@ Newfoundland
535
  Labrador
536
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
537
  </mark>
538
- </br>
539
  <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
540
  The Provincial Discharge Abstract Database
541
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
@@ -650,17 +644,12 @@ Newfoundland
650
  NL
651
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
652
  </mark>
653
- </br></br></br></br>
654
- <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
655
- Nova Scotia
656
- <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
657
- </mark>
658
-
659
  <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
660
  Nova Scotia
661
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
662
  </mark>
663
- </br>Contains
664
  <mark class="entity" style="background: hsl(121, 100%, 59.09090909090909%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
665
  information
666
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.09</span>
@@ -672,10 +661,10 @@ Newfoundland
672
  </mark>
673
  recorded in
674
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
675
- a Nova Scotia hospital
676
-
677
-
678
-
679
  Ontario
680
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
681
  </mark>
@@ -684,12 +673,12 @@ Ontario
684
  Nova Scotia
685
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
686
  </mark>
687
- hospital</br></br></br></br>
688
  <mark class="entity" style="background: hsl(121, 100%, 86.36363636363636%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
689
  Ontario
690
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
691
  </mark>
692
- </br>
693
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
694
  The Discharge Abstract Database
695
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
@@ -755,26 +744,31 @@ Ontario
755
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
756
  </mark>
757
  of
758
- <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
759
- health
760
-
761
-
762
-
763
- Saskatchewan
764
- Captures administrative clinical and demographic information
765
- <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
766
  </mark>
767
 
768
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
769
  Saskatchewan
770
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
771
  </mark>
772
- </br>
773
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
774
  Captures
775
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
776
  </mark>
777
- administrative clinical and demographic information on
 
 
 
 
 
778
  <mark class="entity" style="background: hsl(121, 100%, 63.63636363636363%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
779
  discharges
780
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.08</span>
@@ -794,7 +788,7 @@ Captures administrative clinical and demographic information
794
  transfers
795
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
796
  </mark>
797
- </br></div>
798
  </figure>
799
  </body>
800
  </html>
 
6
 
7
  <body style="font-size: 16px; font-family: -apple-system, BlinkMacSystemFont, 'Segoe UI', Helvetica, Arial, sans-serif, 'Apple Color Emoji', 'Segoe UI Emoji', 'Segoe UI Symbol'; padding: 4rem 2rem; direction: ltr">
8
  <figure style="margin-bottom: 6rem">
9
+ <div class="entities" style="line-height: 2.5; direction: ltr"> </br>CIHI </br>
10
+ <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
11
+ Database
12
+ <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
 
13
  </mark>
14
 
15
  <mark class="entity" style="background: hsl(121, 100%, 100.0%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
 
53
  </mark>
54
  to capture
55
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
56
+ day surgery
57
+
58
+
59
+
60
  Alberta
61
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
62
  </mark>
 
65
  Alberta
66
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
67
  </mark>
68
+ </br>
69
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
70
  The discharge abstract database
71
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
 
102
  </mark>
103
  and
104
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
105
+ transfers
106
+
107
+
108
+
109
+ British Columbia
110
  Data
111
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
112
  </mark>
 
115
  British Columbia
116
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
117
  </mark>
118
+ </br>Data on
119
  <mark class="entity" style="background: hsl(121, 100%, 59.09090909090909%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
120
  discharges transfers
121
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.09</span>
 
266
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
267
  </mark>
268
  basis to the respective
269
+ <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
270
  Ministries of Health The DAD
271
+ <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
272
  </mark>
273
 
274
  <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
 
310
  year
311
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
312
  </mark>
313
+ </br> </br> </br> </br>Manitoba </br>Health data maintained by
 
 
 
 
 
314
  <mark class="entity" style="background: hsl(121, 100%, 63.63636363636363%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
315
  Manitoba Health
316
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.08</span>
 
457
  </mark>
458
  about
459
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
460
+ inpatient and day surgery services
461
+
462
+
463
+
464
+ New Brunswick
465
  Patient
466
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
467
  </mark>
 
470
  New Brunswick
471
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
472
  </mark>
473
+ </br>Patient
474
  <mark class="entity" style="background: hsl(121, 100%, 50.0%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
475
  discharge information
476
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.11</span>
 
507
  </mark>
508
  and
509
  <mark class="entity" style="background: hsl(121, 100%, 86.36363636363636%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
510
+ transfers
511
+
512
+
513
+
514
  Newfoundland
515
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
516
  </mark>
 
529
  Labrador
530
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
531
  </mark>
532
+ </br>
533
  <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
534
  The Provincial Discharge Abstract Database
535
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
 
644
  NL
645
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
646
  </mark>
647
+ </br> </br> </br> </br>
 
 
 
 
 
648
  <mark class="entity" style="background: hsl(121, 100%, 72.72727272727272%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
649
  Nova Scotia
650
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
651
  </mark>
652
+ </br>Contains
653
  <mark class="entity" style="background: hsl(121, 100%, 59.09090909090909%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
654
  information
655
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.09</span>
 
661
  </mark>
662
  recorded in
663
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
664
+ a Nova Scotia hospital
665
+
666
+
667
+
668
  Ontario
669
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
670
  </mark>
 
673
  Nova Scotia
674
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.06</span>
675
  </mark>
676
+ hospital </br> </br> </br> </br>
677
  <mark class="entity" style="background: hsl(121, 100%, 86.36363636363636%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
678
  Ontario
679
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
680
  </mark>
681
+ </br>
682
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
683
  The Discharge Abstract Database
684
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
 
744
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
745
  </mark>
746
  of
747
+ <mark class="entity" style="background: hsl(121, 100%, 86.36363636363636%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
748
+ health
749
+
750
+
751
+
752
+ Saskatchewan
753
+ Captures
754
+ <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.03</span>
755
  </mark>
756
 
757
  <mark class="entity" style="background: hsl(121, 100%, 81.81818181818181%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
758
  Saskatchewan
759
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.04</span>
760
  </mark>
761
+ </br>
762
  <mark class="entity" style="background: hsl(121, 100%, 77.27272727272727%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
763
  Captures
764
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
765
  </mark>
766
+
767
+ <mark class="entity" style="background: hsl(121, 100%, 68.18181818181817%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
768
+ administrative clinical and demographic information
769
+ <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.07</span>
770
+ </mark>
771
+ on
772
  <mark class="entity" style="background: hsl(121, 100%, 63.63636363636363%); padding: 0.45em 0.6em; margin: 0 0.25em; line-height: 1; border-radius: 0.35em;">
773
  discharges
774
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.08</span>
 
788
  transfers
789
  <span style="font-size: 0.8em; font-weight: bold; line-height: 1; border-radius: 0.35em; vertical-align: middle; margin-left: 0.5rem">0.05</span>
790
  </mark>
791
+ </br> </div>
792
  </figure>
793
  </body>
794
  </html>
results/DAD/top_keyphrase.png CHANGED
word.py CHANGED
@@ -34,8 +34,8 @@ def show_gram_plot(doc, n_grams, top_words=10, save_output='results/'):
34
  tokens_df = pd.DataFrame({'word': words, 'count': count})
35
  plt.title("Word Count")
36
  plt.figure(figsize=(24, 8))
 
37
  sns.barplot(data=tokens_df, y='word', x='count', palette='blend:#7AB,#EDA')
38
- plt.xticks(rotation=-10, fontsize=12-n_grams)
39
  plt.savefig(save_output, dpi=300)
40
  plt.close()
41
 
 
34
  tokens_df = pd.DataFrame({'word': words, 'count': count})
35
  plt.title("Word Count")
36
  plt.figure(figsize=(24, 8))
37
+ plt.yticks(fontsize=14)
38
  sns.barplot(data=tokens_df, y='word', x='count', palette='blend:#7AB,#EDA')
 
39
  plt.savefig(save_output, dpi=300)
40
  plt.close()
41