Exploiting Diverse Knowledge Sources via Maximum Entropy in NER

From Cohen Courses
Jump to navigationJump to search

This is a review of the paper Borthwick_1998_exploiting_diverse_knowledge_sources_via_maximum_entropy_in_named_entity_recognition by user:sgopal1.

  • The paper just basically adapts the MaxEnt model into NER. It throws in a bunch of binary features in the model and estimation is through iterative scaling [ Refer Berger ].
  • The paper seems to have crawled some websites like www.babynames.com. If some of those baby names were not listed on the website, would the system perform equally well ? What exactly constitutes a new algorithm ( Can I add some more websites to this list and show that it improves the performance and get a paper! ) ?
Retrieved from "http://curtis.ml.cmu.edu/w/courses/index.php?title=Exploiting_Diverse_Knowledge_Sources_via_Maximum_Entropy_in_NER&oldid=987"