INFORMATION EXTRACTION


Goals and Issues of IE

  • Analyze text to identify entities, relations or events
  • Message Understanding Conference (MUC) parallels Hub Evaluations for speech recognition
  • Named-Entity - names of people, organizations and locations
  • Template Element - person - organization relationship in the document
  • Scenario Template - all instances of a particular type of event
  • Coreference - coreferring noun phrases


Components of an IE System

  • Name recognition
  • Noun and verb group recognition
  • Noun phrase recognition
  • Event recognition
  • Reference resolution
  • Discourse level inference