Method and system of machine translation
First Claim
1. A method for machine translation from a source language to a destination language, said method comprising:
- (a) receiving, at a translation module, at least one sentence in the source language;
(b) searching among and statistically ranking, with a translingual parsing model, a plurality of candidate parses each having;
elements labeled with destination language words corresponding to words of the source language sentence;
syntactic labels indicating a syntactic category of the elements; and
role labels indicating relationships between the elements;
(c) selecting a statistically high ranked parse for the at least one sentence; and
(d) rearranging the parse using the syntactic and role labels, in accordance with word order conventions of the destination language to generate a translingual parse of the source language sentence.
2 Assignments
0 Petitions
Accused Products
Abstract
A system and method is provided for generating a translingual parsing model and for using the model to machine translate a source language to a destination language. The system and method includes receiving a sentence in the source language, and with the translingual parsing model, searching among and statistically ranking candidate parses each having elements labeled with destination language words, syntactic labels, and role labels indicating relationships between the elements. A statistically high ranked parse is selected and rearranged using the syntactic and role labels, in accordance with word order conventions of the destination language to generate a translingual parse of the source language sentence.
45 Citations
40 Claims
-
1. A method for machine translation from a source language to a destination language, said method comprising:
-
(a) receiving, at a translation module, at least one sentence in the source language; (b) searching among and statistically ranking, with a translingual parsing model, a plurality of candidate parses each having; elements labeled with destination language words corresponding to words of the source language sentence; syntactic labels indicating a syntactic category of the elements; and role labels indicating relationships between the elements; (c) selecting a statistically high ranked parse for the at least one sentence; and (d) rearranging the parse using the syntactic and role labels, in accordance with word order conventions of the destination language to generate a translingual parse of the source language sentence. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19)
-
-
20. A method for generating a translingual parsing model for machine translating a source language to a destination language, said method comprising:
-
(a) receiving, at a parser training module, a destination language treebank having parse trees of a plurality of destination language sentences, the parse trees of said destination language sentences having nodes labeled with their syntactic categories in the destination language; (b) generating, using the destination language treebank, a destination language parsing model, including parameters for ranking candidate parse trees for a destination language sentence; (c) receiving, at the parsing module, a second plurality of destination language sentences from a parallel corpus, the parallel corpus including the second plurality of destination language sentences and their respective source language equivalents; (d) applying the destination language parsing model to the second plurality of destination language sentences to generate a ranked list of candidate parse trees for each sentence of the second plurality of destination language sentences; (e) transforming the candidate parse trees by applying, with a tree transformer, a rule set associated with linguistic characteristics of the source and destination languages; (f) assigning, with a role labeler, a linguistic role label to nodes of the candidate parse trees, the role label corresponding to the linguistic role of a node within its respective parse tree; (g) extracting grammar constraints from portions of each candidate parse tree; (h) estimating the translingual parsing model using the extracted grammar constraints and source language sentences of the parallel corpus, the translingual parsing model including parameters sufficient to rank candidate parses, wherein the parameters relate elements of the candidate parses including source language words, destination language words, syntactic labels, and role labels. - View Dependent Claims (21, 22, 23, 24, 25, 26)
-
-
27. A system for machine translation from a source language to a destination language, the system comprising:
-
a translation module configured for receiving at least one sentence in the source language; a translingual parsing model configured for searching among and statistically ranking a plurality of candidate parses each having; elements labeled with destination language words corresponding to words of the source language sentence; syntactic labels indicating a syntactic category of the elements; and role labels indicating relationships between the elements; a parser configured for selecting the highest statistically ranked parse for the at least one sentence; and a reordering module configured for rearranging at least one of the statistically ranked parses, using the syntactic and role labels, in accordance with word order conventions of the destination language to generate a translingual parse of the source language sentence.
-
-
28. A system for generating a translingual parsing model for machine translating a source language to a destination language, said system comprising:
-
a parser training module configured for receiving a destination language treebank having parse trees of a plurality of destination language sentences, the parse trees having nodes labeled with their syntactic categories; the parser training module configured for generating, using the destination language treebank, a destination language parsing model, including parameters for ranking candidate parse trees for a destination language sentence; the parser training module configured for receiving a second plurality of destination language sentences from a parallel corpus, the parallel corpus including the second plurality of destination language sentences and their respective source language equivalents; a destination language parser configured for applying the destination language parsing model to the second plurality of destination language sentences to generate a ranked list of candidate parse trees for each sentence of the second plurality of destination language sentences; a tree transformer configured for transforming the candidate parse trees by applying a rule set associated with linguistic characteristics of the source and destination languages; a role labeler configured to assign a linguistic role label to nodes of the candidate parse trees, the role label corresponding to the linguistic role of a node within its respective parse tree; an extraction module configured for extracting grammar constraints from portions of each candidate parse tree; a translingual grammar estimator configured for estimating the translingual parsing model using the extracted grammar constraints and source language sentences of the parallel corpus; the translingual parsing model including parameters sufficient to rank candidate parses, wherein the parameters relate elements of the candidate parses including source language words, destination language words, syntactic labels, and role labels. - View Dependent Claims (29, 30)
-
-
31. An article of manufacture for machine translating a source language into a destination language, said article of manufacture comprising:
-
a computer usable medium having a computer readable program code embodied therein, said computer usable medium having; computer readable program code for; receiving, at a translation module, at least one sentence in the source language; searching among and statistically ranking, with a translingual parsing model, a plurality of candidate parses each having; elements labeled with destination language words corresponding to words of the source language sentence; syntactic labels indicating a syntactic category of the elements; and role labels indicating relationships between the elements; selecting the highest statistically ranked parse for the at least one sentence; and rearranging the parse using the syntactic and role labels, in accordance with word order conventions of the destination language to generate a translingual parse of the source language sentence. - View Dependent Claims (32, 33, 34, 35)
-
-
36. An automated translation machine for translating a source language into a destination language, said translation machine comprising:
-
means for receiving, at a translation module, at least one sentence in the source language; means for searching among and statistically ranking, with a translingual parsing model, a plurality of candidate parses each having; elements labeled with destination language words corresponding to words of the source language sentence; syntactic labels indicating a syntactic category of the elements; and role labels indicating relationships between the elements; means for selecting the highest statistically ranked parse for the at least one sentence; and means for rearranging the parse using the syntactic and role labels, in accordance with word order conventions of the destination language to generate a translingual parse of the source language sentence. - View Dependent Claims (37, 38, 39, 40)
-
Specification