JPH1074201A - Machine translation system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To efficiently execute the editing work of a translation result by storing the attribute of a case in a word or phrase corresponding to a translated result and retrieving a translated result document not by a character string but by the attribute. SOLUTION: After the end of translation processing in a grammar part, a translated result sentence is stored together with a case information in an output sentence storing area by a control program. When a subjective case retrieval start is instructed, an attribute indicating a subjective case is retrieved from the output sentence storing area by an attribute retrieving part (S51). When the retrieval succeeds (S53: YES), a word or phrase having the attribute is inversely displayed on an output device by a display part (S55). When the retrieved subjective case is unnecessary (S57: YES), deletion is instructed (S59) and the inversely displayed part is deleted by a deleting part 47.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation apparatus for translating a first language given as a source language into another target language such as a second language.

[0002]

2. Description of the Related Art Conventionally, in a machine translator for translating a first language, which is a source language, into a second language, which is a target language, in order to obtain a better translation result, the original language or the target language which is the source language is used. There are several ways to edit a translation. The search function is one of them. For example, if there is a word that is not registered in the dictionary in the original text, it can not be translated, so you can use it to search for such a word and replace it with the word registered in the dictionary . Further, when there is an inappropriate word as a translated word in the translation result sentence, it is also possible to search for that word and replace it with an appropriate word.

Such a search function searches a document for a character string identical to a character string given as a search character string. In order to use this function, the character string used in the document must be used. It was assumed that the columns were known.

[0004]

However, in the actual editing work, the content to be searched is not always known as a character string. In such a case, the user must review the entire document by himself / herself. No. For example, you may want to check how the subject is used in a document, but the subject may be anything.

[0005] English requires a subject, but Japanese does not. Therefore, in a machine translation result using English as a source language and Japanese as a target language, the subject may be translated as unnatural as ordinary Japanese. At this time, in editing the translation result sentence, it is conceivable to delete an extra subject in order to make natural Japanese.

However, in many cases, the character string of the subject cannot be determined. Since any word can be the subject, it is extremely difficult to search for all the subjects by character strings. It is possible for the user to imagine to some extent by the contents of the original text, but it is difficult to translate the original text whose content is unknown.

[0007] Conversely, when the source language is Japanese and the target language is English, even if there is no subject, it will always be translated into English which requires the subject. May be translated. If this tentative subject is, for example, "it", the translation result sentence is checked by "it".
When the search is performed by, words that need not be searched, such as "it" as an object, are searched.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem. A case attribute is stored in a corresponding phrase of a translation result, and a search of a translation result document is performed using an attribute instead of a character string. Accordingly, it is an object of the present invention to provide a machine translation apparatus that can efficiently edit a translation result.

[0009]

In order to achieve this object, a machine translation apparatus according to claim 1 inputs a first language sentence as a source language and inputs instructions for performing various operations. Input means, a translation dictionary storing information necessary to translate the first language sentence given as the source language into the second language sentence as the target language, and the given first language sentence,
Translation means for translating into a second language sentence while referring to the information stored in the translation dictionary, wherein the role as a case in the sentence found during the translation processing by the translation means corresponds to the second language sentence. Attribute storing means for storing as an attribute of a phrase; attribute searching means for searching for the attribute stored by the attribute storing means; and a phrase having the attribute searched by the attribute searching means in a second language sentence as a translation result. Display means shown in FIG.

The attribute storing means stores the role as a case in the sentence found during the translation processing as an attribute. Therefore, the attribute searching means searches for the stored attribute, and the display means checks the attribute of the searched attribute. Display the phrase in the translation.

Further, in the machine translation apparatus according to the second aspect,
There is a deletion unit for deleting the phrase searched by the attribute search unit. The phrase searched by the deletion means is easily deleted.

Further, in the machine translation device according to the third aspect,
An attribute selection unit for selecting an attribute to be searched by the attribute search unit is provided. When a plurality of types of attributes are stored in the attribute storage unit, an attribute of a type to be searched (for example, nominative) is selected by the attribute selection unit, and then the search by the attribute search unit is performed. When only the attribute selected by the attribute selection unit is stored in the attribute storage unit, the attribute search unit searches for a phrase corresponding to the attribute stored from the attribute storage unit.

[0013]

Embodiments of the present invention will be described below with reference to the drawings.

Here, an English-Japanese machine translator for translating English as a source language into Japanese as a target language will be described as an example.

First, the configuration of the entire English-Japanese machine translation apparatus of the present embodiment will be described with reference to FIG. An input device 10 such as a keyboard as an input means for inputting an English sentence to be translated and various instructions, an English sentence and instruction contents input by the input device 10, a translation result sentence (second language sentence)
Output device 5 such as a CRT as an output means for outputting information
0 is a central processing unit (CP) for controlling the entire apparatus.
U) 20.

The RAM 30 serving as a storage means includes a CPU 20
And an input sentence storage area 31 for storing an input English sentence, an output sentence storage area 32 for storing a Japanese sentence as a translation result sentence, and a work area 34. Further, the output sentence storage area 32 also has a role as an attribute storage unit.

A ROM 40, which stores programs and dictionaries for translation, is connected to the CPU 20, and includes a control program 41 for performing various processes, and a dictionary prepared in advance by a system for translating given English sentences into Japanese sentences. And a grammar unit 43 as a translation unit that stores grammar rules for proceeding translation while referring to the dictionary information.

The control program 41 incorporates an attribute search section 45 as attribute search means, a display control section 46 constituting display means, and a delete section 47 as delete means.

Further, the external storage device 60 is connected to the CPU 20 and stores a technical term dictionary prepared by the user as desired, a user dictionary created by the user to reflect the translation result, and the like.

Next, the flow of the translation processing in the grammar section 43 of the present embodiment will be described with reference to the flowchart of FIG.

When an English sentence to be translated is input from the input device 10 and an instruction to execute translation is given, the control program 41
Thus, when the given English sentence is one sentence, the sentence is sent to the grammar unit 43 (S21: YES), and a translation process is performed. When there are a plurality of given English sentences (S21: N
O), is divided into sentences one by one by the control program (S
23), sent to the grammar unit 43 for each sentence.

Here, it is assumed that the English sentence given as a translation target is the document shown in FIG. As shown in FIG. 3A, since there are a plurality of English sentences (S21: NO),
Each sentence is separated by the control program 41 (S2
3) As a result, the first sentence, "Do you k
now him? ”Is sent to the grammar unit 43 in this order. In the grammar unit 43, translation from English to Japanese is performed, but since it is a known technique, detailed description is omitted.

The grammar section first analyzes the structure of the English sentence (S25). In this process, the role of each phrase in the English sentence, that is, the elements such as the nominative case, the purpose case, and the complement, are determined.

FIG. 4A schematically shows the result of the English analysis of the first sentence. In FIG. 4A, “know” is a verb, “you” is the nominative of “know”, and “him” is “
know ".

The information that the English sentence was a question sentence is stored in "know".

After analyzing the structure of the English sentence, each word is converted to Japanese (S27). The conversion result at this time is schematically shown in FIG. As shown in FIG. 4B, "you" is "you", "him" is "he", "
know "is converted to" know. "

Thereafter, Japanese language generation processing is performed to generate a grammatically correct sentence as Japanese (S29). Here, processing such as supplementing an appropriate particle to each case element or ending the inflected word is performed. FIG. 4C schematically shows the result of supplementing the particles. Since "you" is a nominative case, the particle "ha" representing the nominative case is supplemented.
The target case is supplemented with "", and the result is as shown in FIG. At this time, it became clear in the above process,
The attribute of nominative is left attached to the phrase.

FIG. 4D schematically shows the result of generating the first sentence in Japanese. In FIG. 4D, the attribute "Nominative" is stored in "you".

With the above, the translation processing in the grammar unit 43 is completed.
According to the control program 41, the Japanese sentence "Do you know him?" Is stored in the output sentence storage area 32 with the case information shown in FIG. Only the code is output to the output device 50 as a translation result.

If there are a plurality of English sentences initially given, untranslated English sentences remain (S31: YES), so that subsequent English sentences are similarly translated one by one.

FIG. 3B schematically shows the translation result output to the output device 50, and FIG. 3C schematically shows the Japanese generation result stored in the output sentence storage area 32. Since an English sentence always needs a subject, as shown in FIG. 3B, all translated Japanese sentences also have a subject.
However, Japanese sentences seem to be unnatural if there are too many subjects, so it is necessary to perform editing work to remove extra subjects and make them natural Japanese sentences.

Next, the flow of nominative search and deletion in this embodiment will be described with reference to the flowchart of FIG.

When the start of the nominative search is instructed, the attribute search section 45 searches the output sentence storage area 32 for an attribute of the nominative case (S51). It is searched up to the particles that represent the nominative. If the search succeeds (S53: YES),
The display control unit 41 in the control program 41 reversely displays the phrase having this attribute on the output device 50 (S
55). Instead of the reverse display, the word indicating the nominative character can be underlined, or the word can be blinked. In short, any form of display of the searched phrase can be used as long as it can be distinguished and recognized from other parts in the translated sentence.

FIG. 6 is an image diagram of the result of the nominative search.

In the example shown in FIG. 3 (c), since the nominative attribute is given to the beginning of the first sentence, "you are", as a result of searching for this attribute, the first sentence, "you are" Is highlighted. The user checks the subject highlighted on the output device 50, and if there is no need to delete the subject (S5).
7: NO), a next candidate search instruction is issued (S61). If the searched subject is unnecessary in the Japanese sentence (S57: YE
S), when an instruction to delete is made, the reverse display portion is deleted by the deletion unit 47 (S59). At this time, as shown in FIG. 6, the particles are also highlighted, so that the user can delete the entire unnecessary portion including the particles by pressing the key once. Thereafter, the next candidate search is instructed (S6).
1) The same processing is performed. FIG. 3D shows an example of the result of the above search and deletion work. As compared with FIG. 3B, there is no extra subject, and a more natural Japanese document is obtained.

The manner of storing the attributes is not limited to the above method. In the above-described embodiment, the output sentence is stored in the output sentence storage area 32 in a form added to the corresponding phrase. Conceivable. At this time, in the nominative search, a position corresponding to the numerical value stored in the nominative storage area is obtained from the translation result sentence output to the output device 50, and that portion is highlighted.

The attribute to be stored is not limited to the nominative.
A case such as a purpose case or an opposition case may be stored so as to be searchable, or the nominative case may be further subdivided. That is, an attribute that can be distinguished from other nouns, which can be considered when translating Japanese into English, may be added to the temporary subject so that only the temporary subject can be searched.

As described above, when various attributes are to be searched, the following two embodiments exist.

There is a method of storing a plurality of, preferably all, attributes of a phrase when the translated text is stored in the output text storage area 32. At this time, the attribute to be searched is arbitrarily specified from the input device 10, the attribute search unit 45 searches only the word of the specified attribute, and the display control unit 46 sequentially inverts the corresponding word in the translated sentence. You can do so.

By doing so, for example, it is possible to retrieve only the objective case in addition to the nominative case. in this case,
Although the storage capacity of the output sentence storage area 32 increases, the storage capacity does not increase in the following embodiment.

That is, only the designated attribute is stored and stored in the output sentence saving area 32 together with the translated sentence. That is, after arbitrarily specifying the attribute to be searched from the input device 10, only the specified attribute is stored in the output sentence storage area 32 together with the translated sentence. Then, only the phrase having the stored attribute is searched from the output sentence storage area 32.

By doing so, it is possible to select an attribute to be searched. At this time, the input device 1
0 constitutes the attribute selecting means of the present invention.

[0043]

As is apparent from the above description, in the machine translation apparatus according to the first aspect, by searching not the character string of the translation result but the attribute, the phrase which cannot be searched by the character string can be easily searched. And efficiently edit the translation result document.

Further, in the machine translation apparatus according to the second aspect,
Move the cursor to the phrase you want to delete as before,
It is possible to save the trouble of deleting the data by designating the range or deleting one word at a time, thereby improving work efficiency.

In the machine translation device according to the third aspect,
Since the attribute to be searched can be arbitrarily specified, the attribute can be easily detected and edited from various viewpoints.

[Brief description of the drawings]

FIG. 1 is a block diagram showing a configuration of the present embodiment.

FIG. 2 is a flowchart illustrating a flow of a translation process according to the embodiment.

FIG. 3 is a diagram schematically showing an input sentence and an output sentence at each stage of the embodiment.

FIG. 4 is a diagram schematically showing an analysis result, a conversion result, and a generation result in each stage of the embodiment.

FIG. 5 is a flowchart showing a flow of a nominative search process in the embodiment.

FIG. 6 is a screen image diagram of a search result of a nominative search in the embodiment.

[Explanation of symbols]

 Reference Signs List 10 input device 20 CPU 30 RAM 31 input sentence storage area 32 output sentence storage area 34 work area 40 ROM 41 control program 42 dictionary part 43 grammar part 45 attribute search part 46 display part 47 deletion part 50 output device 60 external storage device

Claims (3)

[Claims] An input means for inputting a first language sentence as a source language and inputting instructions for performing various operations, and a second language as a target language using a first language sentence given as a source language A translation dictionary storing information necessary for translation into a language sentence; and a translation means for translating a given first language sentence into a second language sentence while referring to the information stored in the translation dictionary. In the machine translation device, an attribute storage unit that stores a role as a case in a sentence, which is found during the translation processing by the translation unit, as an attribute of a corresponding phrase in a second language sentence, and is stored by the attribute storage unit. A machine translation device comprising: an attribute search unit configured to search for a searched attribute; and a display unit indicating a phrase having the attribute searched by the attribute search unit in a second language sentence as a translation result. 2. The machine translation apparatus according to claim 1, further comprising a deletion unit for deleting a phrase searched by said attribute search unit. 3. The machine translation apparatus according to claim 1, further comprising an attribute selection unit that selects an attribute to be searched by the attribute search unit.