Experiments with learning parsing heuristics

Abstract

Any large language processing software relies in its operation on heuristic decisions concerning the strategy of processing. These decisions are usually "hard-wired" into the software in the form of handcrafted heuristic rules, independent of the nature of the processed texts. We propose an alternative, adaptive approach in which machine learning techniques learn the rules from examples of sentences in each class. We have experimented with a variety of learning techniques on a representative instance of this problem within the realm of parsing. Our approach lead to the discovery of new heuristics that perform significantly better than the current hand-crafted heuristic. We discuss the entire cycle of application of machine learning and suggest a methodology for the use of machine learning as a technique for the adaptive optimisation of language-processing software.