Mirror Data

This page contains links to the data used in [LiMoRo04a] and [LiMoRo04b]. There are three files:

• entities.sort: a list of entities from the corpus, sorted by label. Each line in the file is a record for a single entity. The format is: Document_ID:sentence_ID:word_ID:name:NEType:EntityLabel:Nomeaning. The NEType field takes the value '0' for people, '1' for locations, and '2' for organizations. The Nomeaning field always has the value '1'.
• NYT_txt.tar.gz: a tarball of the raw document texts.
• NYT_column.tar.gz: a tarball of the column-formatted texts. NOTE: this tarball will untar into the current working directory The columns of this format are split by tabs. The important indexes of columns are:
  • 0: word number
  • 1: sentence number
  • 2: text
  • 6: named-entity tag