Table 4
Causes of Successful and Missed Matches by KM in Two Documents*
| KM Component/Method | Causes of Successful KM Matches | Causes of Failed KM Matches |
| Heuristic disambiguation of competing candidate terms | 18 (18%) | 5 (2%) |
| Abbreviation/acronym/hyphen handling | 13 (13%) | 22 (10%) |
| Heuristic overmatch utilization | 34 (35%) | 7 (3%) |
| Heuristic use of form-rules | 11 (11%) | 2 (1%) |
| Heuristic multiword phrase pairing | 10 (10%) | 6 (3%) |
| KM determination of part of speech | N/A | 4 (2%) |
| MRCON-contained overmatches | N/A | 44 (19%) |
| Failure to correct document misspelling | N/A | 2 (1%) |
| Concept not present in MRCON | N/A | 141 (62%) |
| Other reasons for success or failure | 12 (12%) | 13 (6%) |
| Total | 98 | 246 |
-
↵* “Successful KM matches” were those gold standard meaningful terms matched correctly by KM but not MetaMap. The “Failed KM matches” were all gold standard meaningful terms not correctly identified by KM, irrespective of MetaMap's performance.
