===== Paper =====

Marco Pennacchiotti, Patrick Pantel: [[http://www.aclweb.org/anthology/D09-1025|Entity Extraction via Ensemble Semantics]], ACL 2009.

===== Questions =====

  - What are "seed instances" good for?
  - What is the difference between B4 and ES-all (apart from different results)?
  - 
    - See the formula 1 (the average precision AP(L)). Describe it in words and write a formula for P(i) (corr(i) should be pretty clear from the text).
    - Draw (or describe in words) a curve of best possible system into Figure 2.
  - Explain the "termness" feature type, give an example of a feature showing termness of a string and explain what "kind of low-termness instances" it helps to eliminate.
  - What is the difference between entity extraction and named entity extraction task?
  - What is sum_{i=1}^{|L|}corr(i) for Athletes and B3 in the evaluation?