System and process for grammatical inference

• US 8,296,129 B2
• Filed: 04/29/2004
• Issued: 10/23/2012
• Est. Priority Date: 04/29/2003
• Status: Expired due to Fees

• Current Assignee: Telstra Corporation Limited (Telstra Group Ltd.)
• Original Assignee: Telstra Corporation Limited (Telstra Group Ltd.)
• Inventors: Starkie, Bradford Craig
• Primary Examiner(s): Hudspeth, David R
• Assistant Examiner(s): SPOONER, LAMONT M
• Application Number:
  US10/554,861
• Publication Number:
  US 20080126078A1
• Time in Patent Office: 3,099 Days
• Field of Search: 704/1, 704/9, 704/10, 704/251, 704/257
• US Class Current: 704/9
• CPC Class Codes: G06F 40/211 Syntactic parsing, e.g. based on context-free grammar [CFG] or unification grammars
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique

Abstract

A grammatical inference system for inferring a grammar from a plurality of example sentences. The system selects sentences having a common suffix or prefix component; identifies the other of said suffix or prefix component of each selected sentence; generating rules for generating the example sentences and the other components; reduces the right hand side of each rule on the basis of the right hand sides of the other rules; and generates a grammar on the basis of the reduced rules.

Claims

1. A non-transitory computer-readable non-volatile storage medium having stored thereon computer-executable instructions that, when executed by one or more processors of a computer system, cause the computer system to perform a process for inferring a grammar from a plurality of example sentences, the process including:

selecting ones of said example sentences having a common suffix or prefix component;

identifying the other of said suffix or prefix component of each selected sentence;

generating rules for generating the example sentences and the other components of the selected sentences, each of the rules having a left-hand side and a right-hand side;

reducing the right hand side of each rule on the basis of the right hand sides of the other rules using typed leftmost reduction or typed rightmost reduction;

identifying one or more shortest prefix components common only to negative example sentences of said plurality of example sentences;

generating rules for generating said one or more shortest prefix components;

removing one or more of said one or more shortest prefix components by removing one or more of said rules; and

generating a grammar on the basis of the reduced rules.

2. A computer-readable non-volatile storage medium as claimed in claim 1, wherein the process includes deleting rules that violate the constraints of a target class for said grammar.

3. A computer-readable non-volatile storage medium as claimed in claim 2, wherein said deleting includes removing one or more of left recursion, unreachable rules and badly formed signatures.

4. A computer-readable non-volatile storage medium as claimed in claim 1, wherein said reducing includes resolving conflicts from left to right.

5. A computer-readable non-volatile storage medium as claimed in claim 1, wherein said reducing includes resolving conflicts from right to left.

6. A computer-readable non-volatile storage medium as claimed in claim 1, wherein said reducing is repeated until no more reductions can be made, subject to one or more constraints of the grammar.

7. A computer-readable non-volatile storage medium as claimed in claim 1 , wherein the process includes selecting from said rules one or more rules to be unfolded on the basis of said one or more shortest prefix components;

and wherein said step of removing includes unfolding one of more of the selected rules.

8. A computer-readable non-volatile storage medium as claimed in claim 1, wherein said reducing corresponds to one of typed leftmost reduction and typed rightmost reduction.

9. A computer-readable non-volatile storage medium as claimed in claim 1, wherein the example sentences include a plurality of positive and negative example sentences, including:

generating in the limit a class of grammar from only the positive example sentences; and

removing recursion from the grammar.

10. A computer-readable non-volatile storage medium as claimed in claim 9, wherein said step of removing recursion includes removing from the grammar rules that generate negative example sentences.

11. A computer-readable non-volatile storage medium as claimed in claim 1, wherein said step of reducing includes resolving conflicts from left to right or from right to left.

12. A computer-readable non-volatile storage medium as claimed in claim 11, wherein said step of reducing is repeated until no more reductions can be made, subject to one or more constraints of the grammar.

13. A computer-readable non-volatile storage medium as claimed in claim 1, wherein the example sentences include positive and negative example sentences, and the process includes:

identifying in the limit a grammar from only the positive example sentences using machine learning; and

wherein said generating includes generating, on the basis of said grammar and said plurality of positive and negative example sentences, an output grammar that can generate all of the positive example sentences but cannot generate any of the negative example sentences.

14. A non-transitory computer-readable non-volatile storage medium having stored thereon computer-executable instructions that, when executed by one or more processors of a computer system, cause the computer system to perform a process for inferring a grammar from a plurality of positive and negative example sentences and a starting grammar, the process including:

identifying one or more shortest prefix components common only to said plurality of negative example sentences;

identifying rules of the starting grammar for generating said one or more shortest prefix components;

removing one or more of said one or more shortest prefix components by removing one or more of said rules; and

generating a grammar on the basis of the remaining rules.

15. A computer-readable non-volatile storage medium as claimed in claim 14, wherein the process includes:

selecting from said rules one or more rules to be unfolded on the basis of said one or more shortest prefix components; and

wherein said removing includes unfolding one of more of the selected rules.

16. A computer-readable non-volatile storage medium as claimed in claim 14, wherein the example sentences include positive and negative example sentences, and the process includes:

identifying in the limit a grammar from only the positive example sentences using machine learning; and

generating, on the basis of said grammar and said plurality of positive and negative example sentences, an output grammar that can generate all of the positive example sentences but cannot generate any of the negative example sentences.

17. A computer-readable non-volatile storage medium as claimed in claim 14, wherein the process includes:

identifying one or more shortest prefix components common only to negative example sentences of said plurality of example sentences;

generating rules for generating said one or more shortest prefix components; and

removing one or more of said one or more shortest prefix components by removing one or more of said rules.

18. A grammatical inference system, including a memory storing sentences having suffix and prefix components and at least one processor programmed to perform a process for inferring a grammar from a plurality of positive and negative example sentences and a starting grammar, the process including:

selecting ones of said stored sentences having a common suffix or prefix component;

identifying the other of said suffix or prefix component of each selected sentence;

generating rules for generating the example sentences and the other components of the selected sentences, each of the rules having a left-hand side and right-hand side;

reducing the right hand side of each rule on the basis of the right hand sides of the other rules using typed leftmost reduction or typed rightmost reduction;

identifying one or more shortest prefix components common only to negative example sentences of said plurality of example sentences;

generating rules for generating said one or more shortest prefix components;

removing one or more of said one or more shortest prefix components by removing one or more of said rules; and

generating a grammar on the basis of the reduced rules.

19. A grammatical inference system as claimed in claim 18, wherein the processor is further programmed to generate, on the basis of said grammar and said plurality of positive and negative example sentences, an output grammar that can generate all of the positive example sentences but cannot generate any of the negative example sentences.

20. A grammatical inference system as claimed in claim 18, wherein the processor is further programmed to remove recursion from said grammar.

21. A grammatical inference system as claimed in claim 18, wherein the processor is further programmed to convert said grammar to a format for use in a dialog system.

22. A grammatical inference system as claimed in claim 21, wherein said format includes an SRGS grammar.

23. A grammatical inference system as claimed in claim 21, wherein said format includes program code.