Monte-Carlo Sampling for NP-Hard Maximization Problems in the Framework of Weighted Parsing

Abstract. The purpose of this paper is (1) to provide a theoretical justification for the use of Monte-Carlo sampling for approximate resolution of NP-hard maximization problems in the framework of weighted parsing, and (2) to show how such sampling techniques can be efficiently implemented with an explicit control of the error probability. We provide an algorithm to compute the local sampling probability distribution that guarantee that the global sampling probability indeed corresponds to the aimed theoretical score. The proposed sampling strategy significantly differs from existing methods, showing by the same way the bias induced by these methods. 1 Motivations In the framework of Speech Recognition and Natural Language Processing, it is a very common task to search for elements (e.g. sentences, parse trees) a score that depends on the process that was used to produce them. Examples of such tasks include searching a word graph for the most-probable sentence (MPS) according to a Sto...

• Type: Conference
• Year: 2000
• Where: NLP
• Authors: Jean-Cédric Chappelier, Martin Rajman
• Keywords: Monte-Carlo Sampling, Natural Language Processing, NLP 2000, NP-hard Maximization Problems, Sampling Probability