Semantic Role Labeling

22.5 FrameNet

1. What generalization does FrameNet capture that PropBank misses?
   Commonality across different verbs and between verbs and nouns: in “the price of bananas increased 5%”, “…rose 5%” and “a 5% rise in the price of bananas”, the same thing went up by the same amount.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 504
2. What is a frame in Fillmore’s sense?
   The holistic background knowledge that unites a set of words, e.g. air travel for reservation, flight, travel, buy, price, fare, meal, plane; related to models and scripts in AI and cognitive science.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 505
3. What does a FrameNet frame contain?
   A background knowledge structure defining frame-specific semantic roles (frame elements), the predicates (lexical units) that evoke it, and labeled example sentences. Each word evokes a frame and profiles some aspect of it.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 505
4. What is the difference between core and non-core frame elements?
   Core roles are frame-specific; non-core roles express general properties like time and location, similar to PropBank ArgMs.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 505
5. Name core frame elements of the Change_position_on_a_scale frame.
   ITEM (entity with a position on the scale), ATTRIBUTE (the scalar property), DIFFERENCE (distance changed), INITIAL_VALUE, FINAL_VALUE, INITIAL_STATE, FINAL_STATE, VALUE_RANGE.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 505
6. Label “Oil rose in price by 2%” with FrameNet roles.
   [ITEM Oil] rose [ATTRIBUTE in price] [DIFFERENCE by 2%].
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 505
7. How does FrameNet relate frames to each other?
   Frames can inherit from each other and be linked by relations like causation: Cause_change_of_position_on_a_scale adds an AGENT (“[AGENT They] raised [ITEM the price of their soda] [DIFFERENCE by 2%]”).
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 506
8. What datasets and resources exist for semantic roles and semantic frames?
   FrameNet (frames, frame elements, lexical units; also for Spanish, German, Japanese, Portuguese, Italian, Chinese…), PropBank (English and Chinese), NomBank (noun predicates), VerbNet (verb classes), and the CoNLL-2005 and CoNLL-2012 SRL evaluation sets.
   • Jurafsky & Martin, Speech and Language Processing, 3rd ed. draft, ch. 22, §22.5, p. 506