Table 2. Systematic Evaluation of Text Mining Events from 1176 PubMed Abstracts Related to Arabidopsis.
| Event Type | No. of Evaluated Events | Correct Event Type |
Correct Event Type and Arguments |
||
|---|---|---|---|---|---|
| Single-argument binding | 259 | 22 | 8% | 19 | 7% |
| Double-argument binding | 431 | 402 | 93% | 317 | 74% |
| Phosphorylation | 314 | 307 | 98% | 204 | 65% |
| Transcription | 89 | 53 | 60% | 41 | 46% |
| Gene expression | 53 | 49 | 92% | 43 | 81% |
| Regulation (unspecified) | 385 | 360 | 94% | 270 | 70% |
| Positive regulation | 208 | 166 | 80% | 119 | 57% |
| Negative regulation | 48 | 44 | 92% | 35 | 73% |
| All | 1787 | 1403 | 79% | 1048 | 59% |
This PLEV corpus is available as Supplemental Data Set 1 online. For every event type, the evaluated events were randomly picked. All regulatory events assigned to a certain polarity (positive/negative) were also evaluated as unspecified regulation.