AUTOMATIC TRAINING OF A SYNTACTIC AND SEMANTIC PARSER USING A GENETIC ALGORITHM

摘要

Disclosed are methods, systems, and computer-readable mediums for automatic training of a syntactic and semantic parser using a genetic algorithm. An initial population is created, where the initial population comprises a vector of parameters for elements of syntactic and semantic descriptions of a source sentence. A natural language compiler (NLC) system is used to translate the sentence from the source language into a target language based on the syntactic and semantic descriptions of the source sentence. A vector of quality ratings is generated where each quality rating in the vector of quality ratings is of a corresponding parameter in the vector of parameters. Quality ratings are evaluated according to specific criterion, which comprise parameters such as a BLEU score and a number of emergency sentences. A number of parameters in the vector of parameters are replaced with adjusted parameters.

主权项

1. A method comprising:
creating, by a processing device, an initial population comprising a vector of parameters for elements of syntactic and semantic descriptions of a source sentence; using a natural language compiler (NLC) system to translate the source sentence into a resulting translation based on universal syntactic and semantic descriptions of the source sentence; generating a vector of quality ratings, wherein each quality rating in the vector of quality ratings is of a corresponding parameter in the vector of parameters; and replacing a number of parameters in the vector of parameters with adjusted parameters, wherein replacing the number of parameters comprises:
randomly selecting a first parameter from the vector of parameters and adjusting the first parameter to create a first adjusted parameter;computing a quality rating of a translation, corresponding to the first adjusted parameter;comparing the quality rating of the translation, corresponding to the first adjusted parameter to a quality rating of a translation, corresponding to the first parameter; andreplacing the first parameter with the first adjusted parameter if the quality rating of the translation, corresponding to the first adjusted parameter is better than the quality rating of the translation, corresponding to the first parameter.