Table 5 Trigger classification performance on the EVEX resource based on trigger counts (test set examples). The prediction measures in this table are calculated based on the values in the first column of Table 4. This table shows how well the classifier is able to classify and distinguish between correct and incorrect trigger words. The last column (Support) shows that there are 363 correct and 233 incorrect trigger words in the test set, i.e, 596 in total

From: Filtering large-scale event collections using a combination of supervised and unsupervised learning for event trigger classification