Automatic training of a syntactic and semantic parser using a genetic algorithm

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; and

replacing 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.

2. The method of claim 1 , further comprising comparing the resulting translation to a reference translation to determine a quality of the resulting translation, wherein the vector of quality ratings is generated as a product of determining the quality of the resulting translation.

3. The method of claim 2 , wherein the quality of the resulting translation is based on a BLEU score of the resulting translation and a number of emergency sentences.

4. The method of claim 3 , wherein the emergency sentences comprise at least one sentence that is generated during an emergency mode, wherein the emergency mode is automatically engaged if the NLC system encounters an error during translation.

5. The method of claim 1 , wherein replacing the number of parameters in the vector of parameters with adjusted parameters iterates until a stopping condition is satisfied.

6. The method of claim 5 , wherein the stopping condition comprises at least one parameter in the vector of parameters whose corresponding quality rating attains a preliminary determined level, and when the initial population has degenerated.

7. The method of claim 5 , wherein the number of parameters in the vector of parameters that are replaced during an iteration is based on a specified proportion of parameters to be adjusted per iteration.

8. A system comprising:

a processing device configured to:

create an initial population comprising a vector of parameters for elements of universal syntactic and semantic descriptions of a source sentence;

use a natural language compiler (NLC) system to translate the sentence from a source language into a target language based on the universal syntactic and semantic descriptions of the source sentence;

generate 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

replace a number of parameters in the vector of parameters with adjusted parameters, wherein to replace the number of parameters the processing circuit is configured to:

randomly select a first parameter from the vector of parameters and adjusting the first parameter to create a first adjusted parameter;

compute a quality rating of a translation, corresponding to the first adjusted parameter;

compare the quality rating of the translation, corresponding to the first adjusted parameter to a quality rating of a translation, corresponding to the first parameter; and

replace 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.

9. The system of claim 8 , wherein the processing device is further configured to compare the resulting translation to a reference translation to determine a quality of the resulting translation, wherein the vector of quality ratings is generated as a product of determining the quality of the resulting translation.

10. The system of claim 9 , wherein the quality of the resulting translation is based on a BLEU score of the resulting translation and a number of emergency sentences.

11. The system of claim 10 , wherein the emergency sentences comprise at least one sentence that is generated during an emergency mode, wherein the emergency mode is automatically engaged by the processing device if the NLC system encounters an error during translation.

12. The system of claim 8 , wherein replacing the number of parameters in the vector of parameters with adjusted parameters iterates until a stopping condition is satisfied.

13. The system of claim 12 , wherein the stopping condition comprises at least one parameter in the vector of parameters whose corresponding quality rating attains a preliminary determined level, and when the initial population has degenerated.

14. The system of claim 12 , wherein the number of parameters in the vector of parameters that are replaced during an iteration is based on a specified proportion of parameters to be adjusted per iteration.

15. A non-transitory computer-readable medium having instructions stored thereon, the instructions comprising:

instructions to create an initial population comprising a vector of parameters for elements of syntactic and semantic descriptions of a source sentence;

instructions to use a natural language compiler (NLC) system to translate the sentence from a source language into a target language based on the universal syntactic and semantic descriptions of the source sentence;

instructions to generate 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

instructions to replace a number of parameters in the vector of parameters with adjusted parameters, wherein the instructions to replace the number of parameters comprise:

instructions to randomly select a first parameter from the vector of parameters and adjusting the first parameter to create a first adjusted parameter;

instructions to compute a quality rating of a translation, corresponding to the first adjusted parameter;

instructions to compare the quality rating of the translation, corresponding to the first adjusted parameter to a quality rating of a translation, corresponding to the first parameter; and

instructions to replace 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.

16. The non-transitory computer-readable medium of claim 15 , further comprising instructions to compare the resulting translation to a reference translation to determine a quality of the resulting translation, wherein the vector of quality ratings is generated as a product of determining the quality of the resulting translation.

17. The non-transitory computer-readable medium of claim 16 , wherein the quality of the resulting translation is based on a BLEU score of the resulting translation and a number of emergency sentences.

18. The non-transitory computer-readable medium of claim 17 , wherein the emergency sentences comprise at least one sentence that is generated during an emergency mode, wherein the emergency mode is automatically engaged if the NLC system encounters an error during translation.

19. The non-transitory computer-readable medium of claim 15 , wherein replacing the number of parameters in the vector of parameters with adjusted parameters iterates until a stopping condition is satisfied.

20. The non-transitory computer-readable medium of claim 19 , wherein the stopping condition comprises at least one parameter in the vector of parameters whose corresponding quality rating attains a preliminary determined level, and when the initial population has degenerated.