Natural Language => Knowledge Representation (NL-KR) Project Page


The project aims to adress a major research question that remains unanswered in NLP: whether there are methods for getting from a robust "parse-anything" statistical parser to a semantic representation precise enough for knowledge representation and automated reasoning, without falling afoul of the same problems that stymied the broad application of traditional approaches.

The task chosen for testing the developed methods is solving logic puzzles of the sort found in the Law School Admission Test (LSAT) and the old analytic section of the Graduate Record Exam (GRE). Here are some examples of logic puzzles.



  • Iddo Lev, Bill MacCartney, Christopher D. Manning, and Roger Levy, "Solving Logic Puzzles: From Robust Processing to Precise Semantics", 2nd Workshop on Text Meaning and Interpretation, ACL'2004 [pdf]
  • Galen Andrew and Bill MacCartney, "Statistical resolution of scope ambiguity in natural language" [pdf]

Related Work
  • Patrick Blackburn and Johan Bos, Courses in Computational Semantics: "Representation and Inference for Natural Language" and "Working with Discourse Representation Theory", available at

  • Johan Bos, DORIS 2001. This is a major extension to the above books.

  • Rui P. Chaves, "Non-Redundant Scope Disambiguation in Underspecified Semantics", in Balder ten Cate (Ed.) Proc. of the 8th ESSLLI Student Session, pp. 47-58, 2003.
    This is an extension to the plugging algorithm of B&B, which filters some logically equivalent readings during the plugging process. You can try out the system here.

Contact Information

Comments about the project page? Feel free to email Iddo.