JPH07191997A - Interpretation device - Google Patents

Abstract

PURPOSE:To support dialogues of plural persons who use different languages by translating a sentence according to the style directed by a style management means which directs the style of sentence according to the input sentence. CONSTITUTION:A dialogue storage part 42 stores a speaker number, speaker, and I/O language and I/O style of sentence from among a style of sentence management part 4 which manages the sytle of sentence by two persons. When a user directs the style of sentence, a dialogue control part 41 of the part 4 directs a language generation part 3 to output the directed style of sentence and when the sentence style is not directed, the part 41 acquires the style of inputted sentence from the part 1 and directs the style of sentence as the style of sentence of the output one. The language conversion part converts the structure of analyzing the input sentence by a language analysis part 1 into the internal structure of another language according to a conversion dictionary. The part 3 prepares a sentence from the internal structure according to the grammar of the output language to be specified by the grammar rule file and the sentence style directed by the part 41.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an interpreter, and more particularly, to an interpreter with a style adjustment function for supporting a dialogue between two people who speak different languages.

[0002]

2. Description of the Related Art A conventional interpreter incorporates a bidirectional machine translation device as it is, and translates a sentence entered by a user regardless of the style of the sentence entered by the other party.

[0003]

In such a conventional interpreter, although the other party uses a polite body, a situation occurs in which this input is translated by an ordinary body or a conversational body.
They had the drawback of becoming an obstacle to mutual communication.

It is an object of the present invention to provide an interpreter device which can assist the interaction of two people who speak different languages.

[0005]

An interpreter according to the present invention comprises a style managing means for designating a style according to an input sentence, and a translating means for translating a sentence to be translated according to a style designated by a style managing section. It is characterized by including.

Further, when the style is designated, the style managing means may indicate the designated style.

[0007]

[Function] The style management means indicates the style corresponding to the input sentence,
Since the translation means translates the sentence to be translated according to the style designated by the style management means, mechanical interpretation is performed in accordance with the designated style, so that mutual communication between users is smoothly performed. obtain.

[0008]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The embodiment of the interpreting device of the present invention is an interpreting device for translating between two languages using machine translation, in order to assist the dialogue between two humans who speak different languages, the sentences input by both of them are used. By providing a style management unit for managing the style, the interpreter is characterized in that the style of the sentence output by the translation unit is matched with the style of the sentence input by the other party.

In another embodiment, a mechanism for detecting the style of the input sentence of the other party is provided and a function of designating the style for the machine translation module is provided when the sentence input by the user is translated. It is characterized by the fact that the user can directly specify the style when he / she does not want to adjust the style automatically. The feature is that it can be adapted to the style and if the user wants to say it violently, the user can directly specify the style.

Embodiments of the present invention will be described in detail below.

FIG. 1 is a block diagram of an interpreting device between Japanese and English according to an embodiment of the interpreting device of the present invention, FIG. 2 is a block diagram showing an internal structure of a style managing unit, and FIG. 3 is a style managing unit. 3 is a flowchart for explaining the flow of processing of the dialogue control unit. In this embodiment, two users are used as an intermediary, but in the following description, one of the two users is called a “user” and the other is called an “other”. Call.

FIG. 1 shows a general Japanese-English bidirectional translator,
It can be roughly divided into a newly added style management unit 4.

In FIG. 1, 1 is a language analysis unit that analyzes an input sentence according to the grammar of the sentence language specified in the Japanese grammar rule file 6 or the English grammar rule file 7, and 2 is the structure of the analysis result. Is converted into an internal structure of another language according to a conversion dictionary 5 between two languages, which will be described later, according to the grammar of the output language specified by the Japanese grammar rule file 6 or the English grammar rule file 7. A language generation unit that creates sentences, 4 is a style management unit that manages the style of a dialogue between two parties (a core part of this embodiment, which will be described in detail later), and 5 is a structure and vocabulary between Japanese and English. Is a Japanese-English conversion dictionary that can be used to convert both Japanese to English or English to Japanese, and 6 is a Japanese grammar rule file that is commonly used for parsing and generating Japanese. Le, 7 is the English grammar rules files that are shared by the generation and English of the analysis. The language analysis unit 1, the language conversion unit 2, the language generation unit 3, the Japanese-English conversion dictionary 5, the Japanese grammar rule file 6 and the English grammar rule file 7 constitute a translation means.

The present embodiment is a vocabulary-led bidirectional translation method, and is described in the literature Coling 92, Peter Whi.
tick, Shake and Bakertra
Shake and B as in nslation
It is based on ake.

In FIG. 2, the style management unit 4 tells the language generation unit 3 to use the designated style as the style of the output sentence when there is a style instruction, and the language when there is no style instruction. A dialogue control unit 41 that acquires the style of a sentence input from the analysis unit 1 and informs the language generation unit 3 that the style is the style of an output sentence, a speech number, a speaker, an input language, an output language, and an input. Dialogue storage unit 42 for storing styles and output styles
It consists of The arrow in FIG. 2 indicates the dialogue control unit 4.
1 and the dialogue storage unit 42, the dialogue control unit 41 and the user, the dialogue control unit 41 and the language analysis unit 1, and the dialogue control unit 41 and the language generation unit 3 exchange data.

The interpreter having the above-described structure operates as follows.

When the user designates a style, the dialogue control section 41 instructs the language generation section 3 to set the designated style as an output style. When the user does not instruct the style, the dialogue control unit 41 acquires the style of the sentence input from the language analysis unit 1 and instructs the language generation unit 3 to set the style as the style of the output sentence. The language analysis unit 1 analyzes the input sentence according to the grammar of the language of the sentence specified by the Japanese grammar rule file 6 or the English grammar rule file 7. The language conversion unit 2 converts the structure of the analysis result into the internal structure of another language according to the conversion dictionary 5 between the two languages. The language generation unit 3 creates a sentence from the converted internal structure in accordance with the grammar of the output language specified by the Japanese grammar rule file 6 or the English grammar rule file 7 and the style specified by the dialogue control unit 41.

Table 1 is a table for explaining an example of dialogue using this embodiment.

[0019]

[Table 1]

Table 1 is an example of a dialogue between the English person A and the Japanese person B, and shows the input and output of the interpreter. In Table 1, at the starting point of the dialogue, the style of the translated sentence is "is", but the translation is detected by detecting that the input of the Japanese sentence is "is, masu". The style of the sentence is changed to the "is, masu" style.

Table 2 is a dialogue storage unit 42 relating to the dialogue of Table 1.
3 is a table for explaining the stored contents of the.

[0022]

[Table 2]

Table 2 shows that the utterance number, speaker, input language, output language, input style, and output style are stored for each utterance.

The process flow of the dialogue control unit 41 of the style management unit 4 will be described with reference to FIG.

In step S1, it is determined whether the user has instructed to directly specify a style.
If there is an instruction from the user, the process proceeds to step S5, and if there is no instruction from the user, step S2.
Proceed to.

In step S2, the instructed style is transmitted to the language generator 3, and the process proceeds to step S9.

In step S3, the language and style of the sentence input by the user are acquired from the language analysis unit 1.

In step S4, the language and style of the sentence input by the user this time are stored in the dialogue storage unit 42.

Next, in step S5, the style of the sentence previously input by the other party is acquired from the dialogue storage unit 42.

Next, in step S6, the language generation unit is instructed to output the style of the sentence previously input by the opponent as the output of the sentence input by the user this time.

In step S7, the translation language and style of the user's input sentence are stored in the dialogue storage unit 42.

Next, the flow chart described above is shown in Table 2.
This will be described using the example shown in. In the current state, the utterance number is 1, the utterer is A, the user is A, and the other party is B. Since no style is specified, the process proceeds from step S1 to step S3. In step S3, the language "English" and the style "normal" of the sentence "Are you writing a paper?" Input by the user A are acquired from the language analysis unit 1. Next, in step S4, the language “English” and the style “normal” of the sentence input by the user A this time are stored in the dialogue storage unit 42. Next, in step S5, the style of the sentence previously input by the opponent B is acquired from the dialogue storage unit 42. Next, step S
6, the language generation unit 3 sets the style of the sentence previously input by the other party as the output style of the sentence input by the user this time.
Instruct. In this case, there is no previous time, so the default value '
The style of'is' is the style of the output. Then, the language generation unit 3 generates the sentence “Is a paper written?”. Next, in step S7, the dialogue storage unit 42 stores the translation language “Japanese” and the style “that is” for the user's input sentence.
Save to.

Next, the utterance number is 2, the utterer is B, the user is B, and the other party is A. Since the style is not specified,
The process proceeds from step S1 to step S3. In step S3, the language "Japanese" and the style "Masu" of the sentence "Yes, that is." Entered by the user B is acquired from the language analysis unit 1. Next, in step S4, the language "Japanese" and the style "Masuma" of the sentence input by the user B this time is stored in the dialogue storage unit 42. Next, in step S <b> 5, the style “normal” of the sentence previously input by the opponent A is acquired from the dialogue storage unit 42. Next, in step S6, the language generation unit 3 is instructed to set the style "normal" of the sentence previously input by the other party as the output style of the sentence input by the user this time. The language generation unit 3 generates the sentence “Yes, I am.”.
Next, in step S7, the translation language “English” and the style “normal” for the user's input sentence are stored in the dialogue storage unit 42.

Next, the utterance number is 3, the utterer is A, the user is A, and the other party is B. Since the style is not specified,
The process proceeds from step S1 to step S3. In step S3, the sentence'What i entered by the user A by the language analysis unit 1
Get the language'English 'and the style'normal' of s the subject? Next, in step S4, the language "English" and the style "normal" of the sentence input by the user A this time are stored in the dialogue storage unit 4
Save to 2. Next, in step S5, the style of the sentence previously input by the opponent B, "masu", is added to the dialogue storage unit 42.
Get more. Next, in step S6, the language generation unit 3 is instructed to set the style "Masu", which is the style of the sentence previously input by the other party, as the output style of the sentence input by the user this time. The language generation unit 3 generates the sentence "What about?". Next, in step S7, the translation languages “Japanese” and style “Masuma” for the user's input sentence are stored in the dialogue storage unit 42.

Next, the utterance number is 4, the utterer is B, the user is B, and the other party is A. Since the style is not specified,
The process proceeds from step S1 to step S3. In step S3, the language "Japanese" and the style of the sentence "About the history of Japan."
You'll get. Next, in step S4, the language of the sentence entered by user B this time is "Japanese" and style ",
"Masu" is stored in the dialogue storage unit 42. Next, step S
In 5, the style of the sentence input by the opponent A last time, “normal”, is acquired from the dialogue storage unit 42. Next, in step S6, the language generation unit 3 is instructed to set the style "normal" of the sentence previously input by the other party as the output style of the sentence input by the user this time. The language generator 3 uses the sentence'It is about Japa.
'nesehistory' is generated. Next, in step S7, the translation language “English” and the style “normal” for the user's input sentence are stored in the dialogue storage unit 42.

Next, the utterance number is 5, the utterer is A, the user is A, and the other party is B. Since the style is not specified,
The process proceeds from step S1 to step S3. In step S3, the sentence'Oh, it entered by the user A by the language analysis unit 1 is entered.
seems interesting to me.'language 'English'and style'
Get'normal '. Next, in step S4, the language “English” and the style “normal” of the sentence input by the user A this time are stored in the dialogue storage unit 42. Next, in step S5, the style “Masu”, which is the style of the sentence previously input by the opponent B, is acquired from the dialogue storage unit 42. Next, in step S6, the language generation unit 3 is instructed to set the style of the sentence previously input by the other party, "masu," as the output style of the sentence input by the user this time. The language generation unit 3 generates the sentence "It looks interesting." Next, in step S7,
Translation languages'Japanese 'and Stylistic'for user's input sentence
It is saved in 42 in the dialogue memory.

As described above, in the embodiment, since the sentence input by the user is translated according to the style of the sentence input by the other party, there is an advantage that mutual communication becomes smooth.

[0038]

[Effects of the Invention] The style management unit indicates the style corresponding to the input sentence, and the translation means translates the sentence to be translated according to the style specified by the style management unit. Therefore, the machine conforms to the designated style. Interpretation is done so that users can communicate smoothly with each other.

[Brief description of drawings]

FIG. 1 is a block diagram of an interpreter between Japanese and English according to an embodiment of the interpreter of the present invention.

FIG. 2 is a block diagram showing an internal structure of a style management unit.

FIG. 3 is a flowchart illustrating a flow of processing of a style management unit.

[Explanation of symbols]

 1 language analysis unit 2 language conversion unit 3 language generation unit 4 style management unit 5 conversion dictionary 6 Japanese grammar rule file 7 English grammar rule file 41 dialogue control unit 42 dialogue storage unit

Claims (2)

[Claims] 1. An interpreter comprising: a style management unit for designating a style according to an input sentence, and a translation unit for translating a sentence to be translated according to the style designated by the style management unit. apparatus. 2. The interpreter according to claim 1, wherein the style management unit, when a style is designated, designates the designated style.