Information Extraction: Relations, Events, and Time

21.3 Extracting events

1. What is an event mention for event extraction?
   Any expression denoting an event or state that can be assigned to a point or interval in time.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 21, §21.3, p. 484
2. Do events always correspond to verbs in English?
   No: most verbs introduce events, but noun phrases can too (the move, the increase), and some verbs don’t, like phrasal took effect or light verbs.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 21, §21.3, p. 485
3. What is a light verb?
   A verb with little meaning of its own, whose event is expressed by its direct object noun: in took a flight, flight defines the event; take, make, have just provide syntax.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 21, §21.3, p. 485
4. What attributes are extracted for events in TempEval-style tasks?
   Event class (action, state, reporting, perception…), plus aspect, tense and modality, e.g. said = (class=REPORTING, tense=PAST, aspect=PERFECTIVE).
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 21, §21.3, p. 485
5. How is event extraction modeled?
   As supervised learning: IOB sequence labeling to detect events, then multi-class classifiers for class and attributes, using encoders or features like affixes, nominalization suffixes (-tion), POS, light-verb government, and stems.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 21, §21.3, p. 485