Example-based translation method and system employing...

Details Example-based translation method and system employing... Example-based translation method and system employing...

2000-06-05

2002-05-21

Thomas, Joseph (Department: 2644)

Data processing: speech signal processing, linguistics, language

Linguistics

Translation machine

C704S007000, C704S009000

Reexamination Certificate

active

06393388

ABSTRACT:

BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to a translating apparatus and a translating method, and particularly to a translating apparatus and a translating method for translating a first language sentence expressed in a first language into a second language sentence expressed in a second language using examples.
2. Description of the Related Art
Translating apparatuses for translating a first language sentence expressed in a first language into a second language sentence expressed in a second language can be generally divided into the three classes: rule-driven translation apparatuses, example-using translation apparatuses, and example-driven translation apparatuses.
FIG. 1
shows the construction of an example of a rule-driven translation apparatus. In
FIG. 1
, an inputting part
1
consists of for example a keyboard or a voice recognition device or a character recognition device; a first language sentence is inputted into it and is converted into a form such as text data and outputted. That is, when the inputting part
1
consists of a keyboard, a first language sentence is inputted by the keyboard being operated and test data corresponding to that operation is outputted. When the inputting part
1
is a voice recognition device, a first language sentence is inputted as voice and voice-recognized. Text data corresponding to the results of the voice recognition is then outputted. When the inputting part
1
is a character recognition device, a first language sentence written for example on paper is inputted (read) and this first language sentence is character-recognized. Text data corresponding to the results of this character recognition is then outputted.
The text data corresponding to the first language sentence outputted from the inputting part
1
is supplied to an analyzing part
31
. In the analyzing part
31
, the first language sentence from the inputting part
1
is language-processed (analyzed) on the basis of knowledge and rules relating to the first language, and the results of this analysis are supplied to a converting part
32
. The converting part
32
converts the analysis results from the analyzing part
31
into an intermediate language sentence of a prescribed intermediate language on the basis of knowledge and rules relating to the first language and a second language, and outputs this to a generating part
33
. The generating part
33
generates from the intermediate language sentence supplied from the converting part
32
a corresponding second language sentence, that is, a translation (second language sentence) consisting of the first language sentence translated into the second language, on the basis of knowledge and rules relating to the second language.
An outputting part
16
is made up of for example a display or a voice synthesizing device and a speaker or the like, and displays or outputs in a synthesized voice the second language sentence supplied from the generating part
33
.
FIG. 2
shows the construction of an example of an example-using translating apparatus. Parts in the figure corresponding to parts in
FIG. 1
have been given the same reference numerals and will not be described in the following. Apart from being provided with a collating part
41
and a replacing part
42
, this example-using translating apparatus is of the same construction as the rule-driven translation apparatus of FIG.
1
.
In this example-using translating apparatus, examples in sentence units expressed in a first language and corresponding translations consisting of the examples translated into a second language (hereinafter for convenience these examples and their corresponding translations will be called translation example data) are stored for example in the form of parameters, and in the collating part
41
the first language sentence outputted by the inputting part
1
is collated with the examples and any example matching the first language sentence is thereby detected. When there is an example which matches the first language sentence, the collating part
41
controls the replacing part
42
to replace the first language sentence with the translation corresponding to the example matching it. Accordingly, in the replacing part
42
, the first language sentence is replaced with the example matching it and supplied to the outputting part
16
.
When on the other hand there is no example which matches the first language sentence, the collating part
41
outputs the first language sentence to the analyzing part
31
. Thereafter, in the analyzing part
31
, the converting part
32
and the generating part
33
, the same processing as in the case shown in
FIG. 1
is carried out, and the second language sentence obtained as a result is supplied to the outputting part
16
.
Details of this kind of example-using translating apparatus are disclosed for example in Japanese Unexamined Patent Publication No. H.6-68134.
Next,
FIG. 3
shows the construction of an example of an example-driven translating apparatus. Parts in
FIG. 3
the same as parts in
FIG. 1
have been given the same reference numerals and will not be described in the following.
In this example-driven translating apparatus, a first language sentence outputted by the inputting part
1
is supplied to a converting part
51
, and when the converting part
51
receives the first language sentence from the inputting part
1
it controls a searching part
52
to search for the example most similar to that first language sentence.
That is, translation example data is stored in sentence units in a translation example memory
53
, and the searching part
52
first refers to the translation example memory
53
and searches for an example which matches the first language sentence. When it finds an example which matches the first language sentence, the searching part
52
outputs this example and its corresponding translation to the converting part
51
. In this case, the converting part
51
supplies the translation from the searching part
52
unchanged to the outputting part
16
as a second language sentence.
When on the other hand it cannot find an example which matches the first language sentence, the searching part
52
successively reads the examples stored in the translation example memory
53
and supplies them to a similarity degree calculating part
54
. The searching part
52
makes the similarity degree calculating part
54
calculate a similarity degree expressing the conceptual similarity (the similarity in meaning) between each of the examples and the first language sentence using external knowledge such as for example a thesaurus.
That is, a thesaurus wherein words are classified on the basis of their concepts in a tree structure is stored in a thesaurus memory part
55
. In the thesaurus, nodes of the tree structure are equivalent to meaning concepts and so-called leaf parts are equivalent to words. Referring to this thesaurus, the similarity degree calculating part
54
calculates a degree of similarity between the first language sentence and the examples on the basis of the classes to which concepts common to words constituting the first language sentence and words constituting the examples belong. The searching part
52
then finds in the translation example memory
53
the example of which the similarity degree calculated by the similarity degree calculating part
54
is the highest and supplies the translation corresponding to that example to the converting part
51
.
When the converting part
51
receives the translation from the searching part
52
, it replaces those words of the translation which do not match (correspond with) words of the first language sentence with translations of those words and outputs this to the outputting part
16
as a second language sentence.
Details of this kind of example-driven translating apparatus are disclosed for example in Japanese Unexamined Patent Publication No. H.3-276367. Also, details of methods of calculating the degree of similarity between a first language sentence and an example are also disclosed in for

Affiliated with

Also associated with

Rating

Example-based translation method and system employing... does not yet have a rating. At this time, there are no reviews or comments for this patent.

If you have personal experience with Example-based translation method and system employing..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Example-based translation method and system employing... will most certainly appreciate the feedback.

Rate now

Comments { 0 }

Profile ID: LFUS-PAI-O-2860662