Data processing: speech signal processing – linguistics – language – Linguistics – Translation machine
Reexamination Certificate
1999-12-27
2003-02-18
Edouard, Patrick N. (Department: 2654)
Data processing: speech signal processing, linguistics, language
Linguistics
Translation machine
C704S007000
Reexamination Certificate
active
06523000
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
The present invention relates to a translation supporting apparatus which retrieves, from within a translation example database in which character strings of a first language and translation results of another language corresponding to the character strings are stored, a translation example similar to an input of a user and presents it to the user to support a translation task.
2. Description of the Related Art
From a point of view that a translation task is facilitated if a translation result in the past corresponding to a character string similar to a character string which a user intends to translate is obtained, a translation supporting apparatus has conventionally been proposed which retrieves an example similar to an input of a user from within a translation example database and presents the retrieved example to the user.
For example, an apparatus is proposed in Japanese Patent Laid-Open No. Hei 6-318222 wherein a keyword extracted from within an inputted sentence of a first language and a keyword of a second language obtained by translating the keyword of the first language using a word dictionary are both used to perform similarity retrieval (hereinafter referred to as first prior art). The apparatus is based on a technique of calculating a degree of similarity (the degree of similarity may be hereinafter referred to simply as similarity) with a coincidence degree in a unit of a word called keyword. However, since only a translation example translated literally can be retrieved only with a keyword of a first language, in order to minimize a miss of retrieval, also a translated word of a second language corresponding to the keyword of the first language is used for retrieval.
Another apparatus is disclosed in Japanese Patent Laid-Open No. 10-31676 wherein a synonym dictionary is used and a similarity between words is calculated taking not only a similarity of a character string but also a similarity in meaning into consideration (hereinafter referred to as second prior art). The apparatus performs retrieval including a synonym which is not retrieved only with a similarity of characters and also decreases a retrieval miss.
However, the apparatus of the first prior art and the second prior art both take only a similarity calculated between sentences into consideration and fail to retrieve a sufficiently similar sentence from within a document in which omission of words or anaphoric expressions such as pronouns are used frequently. It is often the case that, for example, although the same expression is used in a certain language, in another language, various expressions must be used properly depending upon the situation. However, where it is considered that omission of words is included frequently or anaphoric expressions such as pronouns are used in an actual writing, all of such situations are not necessarily included in one writing and a key representing a situation frequently appears in preceding and/or succeeding sentences. Nevertheless, the conventional apparatus have a drawback in that, since an example of a similar translation is selected only based on the similarity between one sentence and one sentence, a necessary translation example cannot be presented frequently.
It is to be noted that, although selection of a similar translation example is employed also in a translation method wherein a writing of an original language is divided into a plurality of portions and translation is performed separately for the portions of the writing as disclosed, for example, in Japanese Patent Laid-Open No. Hei 10-63665, also in this technical field, a translation example wherein a particular character string included in a sentence of an original language is included in one sentence is selected. Thus, a similar translation example is selected based only on the similarity between one sentence and one sentence similarly.
As described above, in the first and second prior art apparatus, a translation example is selected using only the similarity between one sentence and one sentence. Therefore, if a sentence with regard to which retrieval is required or individual translation examples include some omission of words or some anaphoric expressions, then they cannot retrieve or select a useful translation example.
SUMMARY OF THE INVENTION
It is an object of the present invention to provide a translation supporting apparatus and method by which a translation example similar to a character string of a translation target can be determined appropriately.
It is another object of the present invention to provide a translation supporting apparatus and method by which not only one sentence but also a context which is composed of sentences around each translation example are taken into consideration so that an appropriate translation example effective for a translation task can be retrieved and presented.
It is a further object of the present invention to provide a translation supporting apparatus and method by which not only one sentence but also a context which is composed of sentences around a sentence of a retrieval request are taken into consideration so that an appropriate translation example effective for a translation task can be retrieved and presented.
In order to attain the objects described above, according to an aspect of the present invention, there is provided a translation supporting apparatus for retrieving and outputting a translation example of a second language similar to a character string of a first language which is a translation target, comprising a translation example storage in which a plurality of character strings of the first language and a plurality of translation examples composed of results of translation of the second language from the character string are stored, a similarity determination for determining, for each of the translation examples, a similarity between the translation example and a retrieval request which is a translation target character string and a similarity between the retrieval request and a translation example context which is another translation example having a predetermined relationship to the translation example and integrating the similarities to determine an integrated similarity, a similarity storage for storing the integrated similarities determined by the similarity determination for the individual translation examples, and an output for referring to the integrated similarities stored in the similarity storage and outputting a translation example similar to the retrieval request.
The similarity determination may set, for each of the translation examples, a translation example which is another translation example which belongs to the same document as the translation example and whose distance from the translation example in the document is within a predetermined range as the translation example context. Further, the similarity determination determines, for each of the translation examples, the integrated similarity as a weighted sum of the similarity between the translation example and the retrieval request and the similarity between the translation example context and the retrieval request. Furthermore, the weight on the similarity between the translation example context and the retrieval request may be smaller than the weight on the similarity between the translation example and the retrieval request. In addition, the weight on the similarity between the translation example context and the retrieval request decreases as the distance between the translation example context and the translation example in the document increases.
According to another aspect of the present invention, there is provided a translation supporting apparatus, comprising a translation target storage apparatus in which a character string of a first language which is a translation target is stored, a translation example database in which character strings of the first language and translation examples of a second language corresponding to the character strings are stored for each document, retrieval request inp
Ando Shinichi
Satoh Kenji
Yamabana Kiyoshi
Edouard Patrick N.
NEC Corporation
Young & Thompson
LandOfFree
TRANSLATION SUPPORTING APPARATUS AND METHOD AND... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with TRANSLATION SUPPORTING APPARATUS AND METHOD AND..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and TRANSLATION SUPPORTING APPARATUS AND METHOD AND... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3173021