Skip to main content

Table 12 Sense distributions in IMRAD segments

From: The biomedical discourse relation bank

Type-level Sense Introduction Methods Results Abstract Discussion Total
Alternative 4 (13.8%) 3 (10.3%) 7 (24.1%) 0 (0.0%) 15 (51.7%) 29
Background 24 (19.8%) 7 (5.8%) 36 (29.8%) 15 (12.4%) 39 (32.2%) 121
Cause 80 (17.0%) 16 (3.4%) 134 (28.5%) 33 (7.0%) 208 (44.2%) 471
Circumstance 11 (7.1%) 7 (4.5%) 112 (71.8%) 13 (8.3%) 13 (8.3%) 156
Concession 59 (21.7%) 3 (1.1%) 73 (26.8%) 21 (7.7%) 116 (42.6%) 272
Condition 1 (5.3%) 6 (31.6%) 0 0 (0.0%) 1 (5.3%) 11 (57.9%) 19
Conjunction 105 (13.9%) 100 (13.3%) 271 (35.9%) 78 (10.3%) 195 (25.9%) 754
Continuation 80 (19.3%) 121 (29.2%) 112 (27.0%) 17 (4.1%) 85 (20.5%) 415
Contrast 26 (10.6%) 9 (3.7%) 118 (48.0%) 12 (4.9%) 81 (32.9%) 246
Exception 1 (16.7%) 2 (33.3%) 2 (33.3%) 0 (0.0%) 1 (16.7%) 6
Instantiation 17 (23.9%) 0 (0.0%) 9 (12.7%) 3 (4.2%) 42 (59.2%) 71
Purpose 93 (20.2%) 84 (18.3%) 144 (31.3%) 35 (7.6%) 104 (22.6%) 460
Reinforcement 14 (16.5%) 3 (3.5%) 14 (16.5%) 4 (4.7%) 50 (58.8%) 85
Restatement 63 (19.2%) 47 (14.3%) 124 (37.8%) 29 (8.8%) 65 (19.8%) 328
Similarity 0 (0.0%) 0 (0.0%) 2 (40%) 0 (0.0%) 3 (60%) 5
Temporal 41 (8.0%) 259 (50.3%) 0 (0.0%) 22 (4.3%) 52 (10.1%) 515