TE10 and TE11 dataset

This two datasets can be downloaded from TREC website. TE11 can be found here, and TE10 here.

TE-E dataset

We provide different format of the dataset: YAML, JSON, XML.

We include 50 queries in the TE-E dataset. Each query contains the following fields:

  • num: corresponding to the original query in TREC Entity dataset.
  • entity_name, entity_url: the entity name and its domain.
  • narrative: the query narrative.
  • target_entity_type: the answer entity type.
  • answers: a list of answers for this query, consists of the following fields.
    • names: the entity names.
    • homepages: the entity homepages, if the entity has homepages.
    • wikipages: the Wikipedia page urls, if the entity can be found in Wikipedia.