NIST TAC Knowledge Base Population (KBP2014) Entity Linking Track


The Entity Linking (EL) task at NIST TAC-KBP2014 aims to extract named entities from a source collection of textual documents, and link them to an existing Knowledge Base (KB).  An EL system is also required to cluster those NIL entities which don't have corresponding KB entries. This year the task includes the following tracks: Mono-lingual English EL (English text to English KB), Cross-lingual Chinese-to-English EL (Chinese  text to English KB) and Cross-lingual Spanish-to-English (Spanish text to English KB).

What's New

  • Add a new English Entity Discovery and Linking task (full Entity Extraction + Entity Linking + NIL Clustering)
  • Add discussion forums to Cross-lingual tracks (some entity morphs naturally exist in Chinese discussion forums)
  • Provide basic NLP, full IE and rich semantic annotations for some source documents
  • Increase ambiguity and variety in query selection to ensure outstanding advanced algorithms to outperform naive baseline approaches
  • Share some source collections and queries with regular and cold-start slot filling tracks
  • Reduce the entry cost by providing a recommended list of softwares and a reading list of papers


  • Aug 1: Scorers available
  • July 21: Training Corpora and validator available
  • May 17: Annotation guidelines posted