Table 4 Trigger/event classification performance, measured on the EVEX test set: The first column (Count) shows prediction results based on the counts of trigger words (test set examples). The second column (Sum of frequency) shows the number of respective events of those triggers in the EVEX database. For instance, the first row (True-Positive) shows that the classifier has correctly predicted 352 test set trigger words to be correct triggers, while these words account for 4,602 extracted events in the EVEX resource

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