Natural Language Understanding Wiki


Given a set of scores between pairs of mentions, form clusters to optimize a coreference metrics.

Luo et al. (2004)[1] apply beam-search over Bell trees which they define as search trees obtained by going from an one-mention root node to full-clusters leave nodes by either starting a new cluster or adding a mention to an existing one. They didn't compare their algorithm to anything else and reported scores on true mentions only. Nicolae and Nicolae (2006)[2] later found no improvement of their method compared to a greedy baseline.

Nicolae and Nicolae (2006)[2] propose BestCut, a slightly modified version of the well-known MinCut algorithm. They show that BestCut works better than beam search and greedy decoding.


  1. Luo, X., Ittycheriah, A., Jing, H., Kambhatla, N., & Roukos, S. (2004). A mention-synchronous coreference resolution algorithm based on the Bell tree. In Proceedings of the 42nd Annual Meeting of the Association for Computational Linguistics.
  2. 2.0 2.1 Nicolae, C., & Nicolae, G. (2006). BestCut: A Graph Algorithm for Coreference Resolution. EMNLP 2006, 275–283.