论文信息 - Modeling Discriminative Global Inference

Modeling Discriminative Global Inference

Many recent advances in complex domains such as natural language processing (NLP) have taken a discriminative approach in conjunction with the global application of structural and domain specific constraints. We introduce LBJ, a new modeling language for specifying exact inference systems of this type, combining ideas from machine learning, optimization, first order logic (FOL), and object oriented programming (OOP). Expressive constraints are specified declaratively as arbitrary FOL formulas over functions and objects. The language's run-time library translates them to a mathematical programming representation from which an exact solution is computed. In addition, the compiler leverages an existing OOP language: objects and functions are grounded as the OOP objects and methods that encapsulate the user's data.

Dan Roth | Nicholas Rizzolo | D. Roth | Nicholas Rizzolo

[1] Stuart J. Russell,et al. BLOG: Probabilistic Models with Unknown Objects , 2005, IJCAI.

[2] Avi Pfeffer,et al. IBAL: A Probabilistic Rational Programming Language , 2001, IJCAI.

[3] Dan Roth,et al. The Necessity of Syntactic Parsing for Semantic Role Labeling , 2005, IJCAI.

[4] Michael Collins,et al. Discriminative Training Methods for Hidden Markov Models: Theory and Experiments with Perceptron Algorithms , 2002, EMNLP.

[5] Daniel Gildea,et al. The Proposition Bank: An Annotated Corpus of Semantic Roles , 2005, CL.

[6] Dan Roth,et al. Semantic Role Labeling Via Integer Linear Programming Inference , 2004, COLING.

[7] Dan Roth,et al. Integer linear programming inference for conditional random fields , 2005, ICML.

[8] Mirella Lapata,et al. Aggregation via Set Partitioning for Natural Language Generation , 2006, NAACL.

[9] Michael Strube,et al. Beyond the Pipeline: Discrete Optimization in NLP , 2005, CoNLL.

[10] Matthew Richardson,et al. Markov logic networks , 2006, Machine Learning.