Automatic content extraction

Tasks, problem areas and solution methods

Entity detection and tracking

1. entity detection
   • names
     • gazetteers
     • patterns on word sequences
     • patterns on dependency relations (?)
   • noun phrases
     • partial trees
     • entities vs. non-entities
   • pronouns
2. entity disambiguation (e.g. GPE vs. organization)
   • names
     • context information: which conditions are useful?
     • patterns/stochastic rules on dependency relations
   • noun phrases
     • are there ambiguities?
   • (pronouns)
3. determination of mention extent
   • names
   • noun phrases
     • partial dependency trees
4. Reference resolution
   • dependency relations ???

Relation detection and characterization

1. detection and classification of explicit relations
   • patterns on dependency relations

     at	lives in, moved to, hosted by, ...
     near	close to
     part	belongs to, ...
     role	heads, lives in, owns, ...
     social	cooperates with, is father of, meets with, loves, ...

     What benefit if compared to simple patterns?
2. detection and classification of implicit relations
   • natural understanding but no use of world knowledge ???