0
$\begingroup$

Suppose I have a news report such as

THE ARCE BATTALION COMMAND HAS REPORTED THAT ABOUT 50 PEASANTS OF VARIOUS AGES HAVE BEEN KIDNAPPED BY TERRORISTS OF THE FARABUNDO MARTI NATIONAL LIBERATION FRONT [FMLN] IN SAN MIGUEL DEPARTMENT.

How might I establish, using something akin to extending Named Entity Recognition in NLP to recognise categories of phrase or sentence such as observations, comments, opinions etc rather very short token sequences such as names or locations, that

  • The report documented an ObservationEvent
  • This ObservationEvent was performed by the ARCE Battalion Command
  • The ObservationEvent produced an Observation (indeed produces a number of Observations detailed below)
  • This Observation described an Event of type ‘kidnapping’
  • ‘Kidnapping’ is a kind of ‘criminal activity’ (drawn from the thesaurus of activities)
  • The kidnapping Event was performed by the Farabundo Marti National Liberation Front (FMNLF)
  • The kidnapping Event occurred in San Miguel Department The FMNLF is an Agent (instantiates the FMNLF as an instance of the glass Agent, ‘things which do stuff’ which includes people, organisations, software, etc)
  • The FMNLF is an Organisation (refines the kind of Agent)
  • An Observation describes a State of the FMNLF of type ‘terrorist’ (noting being a terrorist organisation is not an intrinsic property of an organisation but a temporally bounded ascription by some other, see for example PLO, ANC, etc).
  • A Group participated in the kidnapping Event in the role of ‘victim’ (drawn from the thesaurus of roles)
  • An Observation describes the membership State of the group
  • Around 50 members form part of the Group (note uncertainty in this membership tally)
  • An Organisation (the FMNLF) participated in the kidnapping Event in the role of ‘perpetrator’ (drawn from the thesaurus of roles)

It goes without saying that I will have a large corpus of such news reports, which cover a wide variety of news topics, and of that corpus a number will have been marked up to indicate their structure.

$\endgroup$

0

You must log in to answer this question.

Browse other questions tagged .