Statistical translation system with features based on phrases or groups of words
First Claim
1. A system for translating a first word set in a source language into a second word set in a target language, the system comprising:
- input means for inputting the first word set into the system;
tagging means for tagging the first word set input to the system so as to at least substantially reduce non-essential variability in the first word set;
translation means including a single a posteriori conditional probability model and a target candidate store for storing target language candidate word sets, wherein the translation means employs the single model to evaluate the target language candidate word sets in order to select the target language candidate word set having a best score with respect to the first word set;
wherein the single model includes a prior model, a plurality of feature functions and a plurality of weighting factors respectively corresponding to the plurality of feature functions, wherein the translation means receives the first word set and, in accordance with the single model, iteratively proceeds through the target language candidate store and each of the candidate word sets and finds the feature functions which are true with respect to the first word set and the second word set, and then multiplies the prior model by the weighting factors corresponding to the feature functions thereby providing resulting scores, the translation means then choosing the best matching target language candidate word set in accordance with the resulting scores; and
output means for out putting the best scoring target language candidate word set as the second word set in the target language.
1 Assignment
0 Petitions
Accused Products
Abstract
A system for translating a first word set in a source language into a second word set in a target language, the system comprising: input means for inputting the first word set into the system; tagging means for tagging the first word set input to the system so as to at least substantially reduce non-essential variability in the first word set; translation means including a single a posteriori conditional probability model and a target candidate store for storing target language candidate word sets, wherein the translation means employs the single model to evaluate the target language candidate word sets in order to select the target language candidate word set having a best score with respect to the first word set; and output means for outputting the best scoring target language candidate word set as the second word set in the target language.
-
Citations
27 Claims
-
1. A system for translating a first word set in a source language into a second word set in a target language, the system comprising:
-
input means for inputting the first word set into the system; tagging means for tagging the first word set input to the system so as to at least substantially reduce non-essential variability in the first word set; translation means including a single a posteriori conditional probability model and a target candidate store for storing target language candidate word sets, wherein the translation means employs the single model to evaluate the target language candidate word sets in order to select the target language candidate word set having a best score with respect to the first word set; wherein the single model includes a prior model, a plurality of feature functions and a plurality of weighting factors respectively corresponding to the plurality of feature functions, wherein the translation means receives the first word set and, in accordance with the single model, iteratively proceeds through the target language candidate store and each of the candidate word sets and finds the feature functions which are true with respect to the first word set and the second word set, and then multiplies the prior model by the weighting factors corresponding to the feature functions thereby providing resulting scores, the translation means then choosing the best matching target language candidate word set in accordance with the resulting scores; and output means for out putting the best scoring target language candidate word set as the second word set in the target language. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25)
-
-
26. A method for translating a first word set in a source language into a second word set in a target language, the method comprising the steps of:
-
(a) storing target language candidate word sets; (b) forming a single a posteriori conditional probability model which includes a prior model, a plurality of feature functions and a plurality of weighting factors respectively corresponding to the plurality of feature functions; (c) inputting the first word set; (d) tagging the input first word set so as to at least substantially reduce non-essential variability in the first word set; (e) determining which features fire regarding the target language candidate word sets and if at least one of the feature functions fires, multiplying the prior model by the weighting factor corresponding to the feature function to provide a resulting score, and if no feature function fires than multiplying the prior model by one; (f) evaluating a next target language candidate word set, if available, as in step (e); and (g) determining a best score from among the resulting scores such that the target language candidate word set having the best score is chosen as the second word set in the target language.
-
-
27. A translation model utilized in a translation system for translating a first word set in a source language into a second word set in a target language, the translation model comprising a prior model, a plurality of feature functions and a plurality of weighting factors respectively corresponding to the plurality of feature functions, wherein the translation model is responsive to the first word set and iteratively proceeds through a target language candidate store of candidate word sets and each of the candidate word sets and finds the feature functions which are true with respect to the first word set and the second word set, and then multiplies the prior model by the weighting factors corresponding to the feature functions thereby providing resulting scores, the translation model then providing the best scoring target language candidate word set in accordance with the resulting scores.
Specification