Method and apparatus for parsing in a spoken language translation system

• US 6,223,150 B1
• Filed: 01/29/1999
• Issued: 04/24/2001
• Est. Priority Date: 01/29/1999
• Status: Expired due to Fees

• Current Assignee: Sony Corporation (Sony Group Corp.), Sony Electronics Inc. (Sony Group Corp.)
• Original Assignee: Sony Corporation (Sony Group Corp.), Sony Electronics Inc. (Sony Group Corp.)
• Inventors: Duan, Lei, Franz, Alexander M.
• Primary Examiner(s): Thomas, Joseph
• Application Number:
  US09/240,896
• Time in Patent Office: 816 Days
• Field of Search: 704/3, 704/9, 704/2, 704/4, 704/5, 704/6, 704/7, 704/8, 704/1, 704/257, 704/270, 704/277, 707/531, 707/536, 707/100, 707/101, 707/102, 707/104
• US Class Current: 704/9
• CPC Class Codes: G10L 15/26 Speech to text systems G10L15/08 takes precedence

Abstract

A method and apparatus for parsing in a spoken language translation system are provided, wherein an input is received comprising at least one input sentence or expression. A parsing table is accessed and consulted for a next action, wherein the parser looks up in the next action in the parsing table. During parsing operations, the parser may perform shift actions and reduce actions. In performing a shift action, a next item of the input string is shifted onto a stack or intermediate data structure of the parser. A new parse node is generated, and a feature structure or lexical feature structure of the shifted input item is obtained from a morphological analyzer and associated with the new parse node. The new node is placed on the stack or intermediate data structure. In performing a reduce action, a grammar rule and an associated compiled feature structure manipulation are applied. When the manipulations succeed, a new parse node is generated comprising the new feature structures resulting from the successful feature structure manipulations. When the entire input is successfully parsed then an accept action is performed, a rebuilding procedure is performed, and a structural analysis of the input is provided comprising a number of parse trees and sentential feature structures.

Claims

1. A method for performing spoken language translation, comprising:

receiving at least one input expression;

determining at least one next action using at least one parsing table;

shifting at least one component of the at least one input expression into a data structure when the at least one next action is determined to be a shift action;

generating at least one new parse node;

associating at least one feature structure of the at least one component with the at least one new parse node;

placing the at least one new parse node on the data structure;

generating at least one packed shared parse forest comprising a plurality of parse trees and a plurality of nodes;

rebuilding at least one node of the at least one packed shared parse forest; and

outputting at least one structural analysis of the at least one input expression.

2. The method of claim 1, further comprising:

applying at least one grammar rule to the at least one component when the at least one next action is determined to be a reduce action;

executing at least one feature structure manipulation corresponding to the at least one grammar rule; and

determining whether the at least one feature structure manipulation succeeds.

3. The method of claim 1, wherein rebuilding comprises:

performing local ambiguity packing; and

recursively rebuilding the at least one node.

4. The method of claim 3, wherein recursively rebuilding comprises:

marking each of a plurality of nodes for which at least one feature structure is to be rebuilt;

maintaining at least one log comprising each of the plurality of nodes for which the at least one feature structure is to be rebuilt;

locating at least one farthermost marked node from a root node when traversing at least one branch path of the at least one packed shared parse forest;

rebuilding the at least one feature structure of the at least one farthermost marked node;

rebuilding the at least one feature structure of each marked node in succession along the at least one branch path between the at least one farthermost marked and the root node; and

rebuilding the root node.

5. The method of claim 1, further comprising:

generating the at least one parsing table from at least one set of grammar rules;

compiling at least one set of feature structure manipulations corresponding to the at least one set of grammar rules;

maintaining at least one graph-structured stack comprising at least one parsing state.

6. The method of claim 5, wherein the at least one set of grammar rules comprise context-free grammar rules, wherein the context-free grammar rules use a plurality of non-terminal symbols.

7. The method of claim 5, wherein the at least one set of grammar rules comprise English parsing grammar rules and Japanese parsing grammar rules.

8. The method of claim 1, wherein shifting comprises:

creating at least one parse node for a shifted lexicon; and

generating a feature structure for the at least one parse node by coping a feature structure of the lexicon.

9. The method of claim 1, wherein the at least one input expression comprises a grammatical sentence, wherein the grammatical sentence comprises at least one spoken word, wherein at least one feature structure is available for each lexicon of the at least one spoken word.

10. The method of claim 1, wherein the at least one input expression comprises at least one lexical feature structure.

11. The method of claim 1, wherein the at least one structural analysis comprises a plurality of parse trees and sentential feature structures.

12. The method of claim 1, wherein the at least one feature structure comprises information from at least one lexical level of at least one morphological analysis.

13. An apparatus for spoken language translation comprising:

at least one processor;

an input coupled to the at least one processor, the input capable of receiving at least one input expression, the at least one processor configured to translate the received at least one input expression by,

determining at least one next action using at least one parsing table,

shifting at least one component of the at least one input expression into a data structure when the at least one next action is determined to be a shift action,

generating at least one new parse node,

associating at least one feature structure of the at least one component with the at least one new parse node,

placing the at least one new parse node on the data structure,

generating at least one packed shared parse forest comprising a plurality of parse trees and a plurality of nodes,

rebuilding at least one node of the at least one packed shared parse forest; and

an output coupled to the at least one processor, the output capable of providing at least one structural analysis of the at least one input.

14. The apparatus of claim 13, wherein the processor is further configured to translate by:

applying at least one grammar rule to the at least one component when the at least one next action is determined to be a reduce action;

executing at least one feature structure manipulation corresponding to the at least one grammar rule; and

determining whether the at least one feature structure manipulation succeeds.

15. The apparatus of claim 13, wherein rebuilding comprises:

performing local ambiguity packing; and

recursively rebuilding the at least one node.

16. The apparatus of claim 15, wherein recursively rebuilding comprises:

marking each of a plurality of nodes for which at least one feature structure is to be rebuilt;

maintaining at least one log comprising each of the plurality of nodes for which the at least one feature structure is to be rebuilt;

locating at least one farthermost marked node from a root node when traversing at least one branch path of the at least one packed shared parse forest;

rebuilding the at least one feature structure of the at least one farthermost marked node;

rebuilding the at least one feature structure of each marked node in succession along the at least one branch path between the at least one farthermost marked and the root node; and

rebuilding the root node.

17. The apparatus of claim 13, wherein the processor is further configured to translate by:

generating the at least one parsing table from at least one set of grammar rules;

compiling at least one set of feature structure manipulations corresponding to the at least one set of grammar rules;

maintaining at least one graph-structured stack comprising at least one parsing state.

18. The apparatus of claim 17, wherein the at least one set of grammar rules comprise context-free grammar rules, wherein the context-free grammar rules use approximately 78 non-terminal symbols.

19. The apparatus of claim 17, wherein the at least one set of grammar rules comprise English parsing grammar rules and Japanese parsing grammar rules.

20. The apparatus of claim 13, wherein shifting comprises:

creating at least one parse node for a shifted lexicon; and

generating a feature structure for the at least one parse node by coping a feature structure of the lexicon.

21. The apparatus of claim 13, wherein the at least one input expression comprises a grammatical sentence, wherein the grammatical sentence comprises at least one spoken word, wherein at least one feature structure is available for each lexicon of the at least one spoken word.

22. The apparatus of claim 13, wherein the at least one input expression comprises at least one lexical feature structure.

23. The apparatus of claim 13, wherein the at least one structural analysis comprises a plurality of parse trees and sentential feature structures.

24. The apparatus of claim 13, wherein the at least one feature structure comprises information from at least one lexical level of at least one morphological analysis.

25. A computer readable medium containing executable instructions which, when executed in a processing system, causes the system to perform a method for spoken language translation, the method comprising:

receiving at least one input expression;

determining at least one next action using at least one parsing table;

shifting at least one component of the at least one input expression into a data structure when the at least one next action is determined to be a shift action;

generating at least one new parse node;

associating at least one feature structure of the at least one component with the at least one new parse node;

placing the at least one new parse node on the data structure;

generating at least one packed shared parse forest comprising a plurality of parse trees and a plurality of nodes;

rebuilding at least one node of the at least one packed shared parse forest; and

outputting at least one structural analysis of the at least one input expression.

26. The computer readable medium of claim 25, wherein the method further comprises:

applying at least one grammar rule to the at least one component when the at least one next action is determined to be a reduce action;

executing at least one feature structure manipulation corresponding to the at least one grammar rule; and

determining whether the at least one feature structure manipulation succeeds.

27. The computer readable medium of claim 25, wherein rebuilding comprises:

performing local ambiguity packing; and

recursively rebuilding the at least one node.

28. The computer readable medium of claim 27, wherein recursively rebuilding comprises:

marking each of a plurality of nodes for which at least one feature structure is to be rebuilt;

maintaining at least one log comprising each of the plurality of nodes for which the at least one feature structure is to be rebuilt;

locating at least one farthermost marked node from a root node when traversing at least one branch path of the at least one packed shared parse forest;

rebuilding the at least one feature structure of the at least one farthermost marked node;

rebuilding the at least one feature structure of each marked node in succession along the at least one branch path between the at least one farthermost marked and the root node; and

rebuilding the root node.

29. The computer readable medium of claim 25, wherein the method further comprises:

generating the at least one parsing table from at least one set of grammar rules;

compiling at least one set of feature structure manipulations corresponding to the at least one set of grammar rules;

maintaining at least one graph-structured stack comprising at least one parsing state.

30. The computer readable medium of claim 29, wherein the at least one set of grammar rules comprise context-free grammar rules, wherein the context-free grammar rules use approximately 78 non-terminal symbols.

31. The computer readable medium of claim 29, wherein the at least one set of grammar rules comprise English parsing grammar rules and Japanese parsing grammar rules.

32. The computer readable medium of claim 25, wherein shifting comprises:

creating at least one parse node for a shifted lexicon; and

generating a feature structure for the at least one parse node by coping a feature structure of the lexicon.

33. The computer readable medium of claim 25, wherein the at least one input expression comprises a grammatical sentence, wherein the grammatical sentence comprises at least one spoken word, wherein at least one feature structure is available for each lexicon of the at least one spoken word.

34. The computer readable medium of claim 25, wherein the at least one input expression comprises at least one lexical feature structure.

35. The computer readable medium of claim 25, wherein the at least one structural analysis comprises a plurality of parse trees and sentential feature structures.

36. The computer readable medium of claim 25, wherein the at least one feature structure comprises information from at least one lexical level of at least one morphological analysis.