Table 4 |
||||
|
Number of correct and incorrect terms for each of the rewrite and suppression rules. |
||||
|
Rule |
Most frequent |
Random |
||
|
|
||||
|
Rewrite rules |
Correct |
Incorrect |
Correct |
Incorrect |
|
|
||||
|
Syntactic inversion |
50 |
0 |
100 |
0 |
|
Possessives |
50 |
0 |
100 |
0 |
|
Short/long form |
49 |
1 |
98 |
2 |
|
Angular brackets |
50 |
0 |
97 |
3 |
|
Semantic type |
50 |
0 |
100 |
0 |
|
Begin parentheses |
1 |
25 |
- |
- |
|
End parentheses |
49 |
1 |
96 |
4 |
|
Begin brackets |
38 |
12 |
91 |
9 |
|
End brackets |
46 |
4 |
95 |
5 |
|
|
||||
|
Suppression rules |
||||
|
|
||||
|
Dosages |
50 |
0 |
100 |
0 |
|
Short token |
50 |
0 |
100 |
0 |
|
At-sign |
- |
- |
- |
- |
|
EC numbers |
50 |
0 |
99 |
0 |
|
Any classification |
50 |
0 |
100 |
0 |
|
Any underspecification |
50 |
0 |
100 |
0 |
|
Miscellaneous |
50 |
0 |
100 |
0 |
|
Words > 5 |
0 |
50 |
5 |
95 |
|
|
||||
|
The calculations are based on the, for every rule, 50 most frequently found terms in the corpus and 100 randomly selected terms in the corpus (if available). The At-sign rule has no values because terms suppressed by this rule were not found in the corpus. |
||||
|
Hettne et al. Journal of Biomedical Semantics 2010 1:5 doi:10.1186/2041-1480-1-5 |
||||
