JPH07282061A - Machine translation device - Google Patents

Abstract

PURPOSE:To prevent errorneous translated result from being outputted and to output correct translated result by using grammar information extracted from a question sentence (interrogative sentence) at the time of translating a response sentence in question and answertype sentences in an inputted sentence. CONSTITUTION:The machine translation device is provided with the functions of Japanes-English translation and English-Japanese translation. The device is provided with an input part 1 reading the sentence of a language to be translated, an analysis processing part 2 analyzing the sentence of the language inputted by using a dictionary storage part 5 and a grammar rule storage part 6, a conversion processing part 3 converting the sentence of the inputted language by using the dictionary storage part 5 and the grammar rule storage part into a target language, a grammar information holding part 7 and an output part 4 outputting the target language being the translated result. The conversion processing part 3 converts the response sentence inputted by succeeding the interrogative sentence into the second language based on extracted grammar information. When extracted grammar information is of the tense of the interrogative sentence, the tense of the response sentence is made to match with that of the interrogative sentence based on the information.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation device for automatically translating Japanese sentences into English sentences.

[0002]

2. Description of the Related Art Japanese Unexamined Patent Publication (Kokai) No. 4-152466 describes a key for acquiring information indicating a document structure and a context condition from a text by using a character face and a character type, and a document to be translated is sentence by sentence from the beginning. When reading and translating, if a key indicating the role of the next sentence is included in the sentence, that information is stored and used in the translation process of the next sentence. Is disclosed. According to this method, the user can easily set the context condition and specify the translation method, so that the function of translating the style or the phrase depending on the context condition can be enhanced.

[0003]

However, since the conventional machine translation device does not pay attention to the translation of the question-and-answer form sentence composed of the question sentence (question sentence) and the answer sentence to it,
Where is the sentence in the input language sentence 1 tomorrow? Sentence 2 I go to school. Sentence 3 Do you not go to school today? Sentence 4 Yes. When a question-and-answer part consisting of a question sentence and an answer sentence to it is included, sentence 1 Where will you go tomorrow? Sentence 2 I go to school. Sentence 3 Don't you go to school today? Sentence 4 Yes . translated and so on, in order to output an incorrect translation result,
Humans needed to correct the output translation results.

The present invention has been made in consideration of such circumstances, and a question and answer portion composed of a question sentence and an answer sentence to the sentence, for example, the above sentence 1 to sentence 4 is added to the sentence of the input language. Sentence 1 Where will you go tomorrow? Sentence 2 I will go to school. Sentence 3 Don't you go to school today? Sentence 4 It is intended to provide a machine translation device capable of outputting a correct translation result such as No.

[0005]

SUMMARY OF THE INVENTION According to the present invention, an input unit for inputting a sentence in a first language, a conversion processing unit for converting an input sentence in a first language into a sentence in a second language, An analysis processing unit that analyzes whether or not the input sentence is a question sentence and its answer sentence, and a grammar of the question sentence converted into the second language when the input sentence is the question sentence and its answer sentence A grammatical information holding unit that extracts and stores information is provided, and the conversion processing unit converts an answer sentence that is input subsequent to the interrogative sentence into a second language based on the extracted grammatical information. The present invention provides a machine translation device that does.

For the input section, for example, a device such as a keyboard for a user to manually input a sentence or a reading device for directly reading a sentence from a recording medium (paper, magnetic disk, optical disk) is used. The conversion processing unit, the analysis processing unit, and the grammar information processing unit are configured by a microcomputer having a CPU, a ROM, and a RAM, and the dictionaries for the first language and the second language required for the conversion processing unit and the analysis processing unit. It is desirable that the grammar rule and the grammar rule are stored in advance in the ROM dictionary storage unit and the grammar rule storage unit.

According to the present invention, the grammatical information holding unit extracts and stores, as grammatical information, information representing the tense of the question sentence,
The conversion processing unit converts the answer sentence input after the interrogative sentence into the second sentence.
When converting to a language, it is preferable to match the tense of the answer sentence with the tense of the question sentence based on the extracted information.

According to the present invention, the first language is Japanese, the second language is English, and the grammatical information holding unit extracts information representing the type of question sentence as grammatical information, and the conversion processing unit When the interrogative sentence is a negative interrogative sentence, when converting the answer sentence that is input following the interrogative sentence into the second language, the word "yes" indicating affirmation is converted to "No" and the word "no" indicating negation. Is preferably converted to “Yes”.

According to the present invention, the first language is English, the second language is Japanese, and the grammatical information holding unit extracts information indicating the type of question sentence as grammatical information, and the conversion processing unit When an interrogative sentence is a negative interrogative sentence, "N
Preferably, "o" is converted to the word "yes" for affirmation and "Yes" is converted to the word "no" for negation.

In the present invention, the grammar information extracted by the grammatical information holding unit from the interrogative sentence converted into the second language is, for example, the tense (past, present, future, etc.) of the interrogative sentence and the type of interrogative sentence (normal). Question text, negative question text, etc.).

Then, when the grammatical information holding unit extracts such grammatical information, the conversion processing unit converts the answer sentence input after the question sentence into the second language based on the extracted grammatical information. .

That is, when the extracted grammatical information is the tense of the question sentence, the tense of the answer sentence is matched with the tense of the question sentence based on the information. Therefore, if the answer sentence "Where are you going tomorrow?" Is input as the first language and the second language is English, then the question sentence is "Where will you go
tomorrow? "is translated into English. Then, the tense "future" is extracted from the converted English sentence, and the answer sentence is converted to "I will go to school." Based on the future tense.

Further, when the extracted grammatical information is the type of question sentence and the question sentence is a negative question sentence, if the first and second languages are Japanese and English, respectively, the affirmation included in the answer sentence The word "yes" is converted to "No" and the negative word "no" is converted to "Yes". Therefore, if the answer sentence "Yes." Is input after the question sentence "Are you not going to school today?", The question sentence is first converted to "Don't you go to school today?". The answer sentence “Yes.” Is converted into “No.”.

If the question sentence is a negative question sentence and the first and second languages are English and Japanese, respectively, "No" included in the answer sentence is "Yes" and "Yes" is "No". , "Respectively.

[0015]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described in detail below based on the embodiments shown in the drawings. This does not limit the invention. FIG. 1 is a block diagram showing a main part of a machine translation device according to the present invention. Referring to FIG. 1, the machine translation device has functions of Japanese-English translation and English-Japanese translation, and is input using an input unit 1 for reading a sentence in a language to be translated, a dictionary storage unit 5, and a grammar rule storage unit 6. An analysis processing unit 2 for analyzing sentences in different languages, a conversion processing unit 3 for converting sentences in the input language into a target language using the dictionary storage unit 5 and the grammar rule storage unit 5, and a grammar information holding unit 7. , And an output unit 4 for outputting a target language as a translation result.

The input unit 1 may directly input a sentence not only by file information but also by a keyboard. As the output unit 4, a CRT, a liquid crystal display device, various printers, or the like can be used.

FIG. 2 is a flow chart showing the flow of processing in the machine translation apparatus shown in FIG. Referring to FIG. 2, first, the flag of the grammar information holding unit 7 is initialized (step S1). Next, if there is an untranslated sentence, one sentence is read by the input unit 1 (step S3).

Next, analysis processing is performed (step S
4) Then, it is judged whether or not it is a question sentence (step S5), and if it is a question sentence, the "question sentence processed flag" of the grammar information holding unit 7 is set to ON (step S6), and the conversion processing unit 3 Normal conversion processing is performed (step S7).

Next, the grammatical information holding 7 stores the grammatical information such as the tense of the converted sentence (step S8). In step S5, if it is not a question sentence, it is checked whether the "question sentence processed flag" of the grammar information holding unit 7 is ON or OFF (step S9). If it is OFF, the process proceeds to step S7. If it is ON, the conversion processing unit 3 performs conversion processing according to the grammatical information held in the grammatical information holding unit 7 (step S10).

Then, the "questioned sentence processed flag" is OF
Set to F (step S11). After the processing of steps S8 and S11, the output unit 4 outputs the target language (step S12). Next, the process proceeds to step S2, and the above process is repeated while there are untranslated sentences. When there are no untranslated sentences, the process ends.

As a specific example, a process for sentences 1 and 2 when the Japanese sentence (sentence 1, sentence 2) shown in FIG. 4 is input will be described. Since sentence 1 is a question sentence, the process proceeds from step S5 to step S6, and the "question sentence processed flag" of the grammar information holding unit 7 is turned on.

Then, in step S7, "Where will you
"go tomorrow?", and the information "future tense" is stored in the grammar information holding unit 7 in step S8. Next, since sentence 2 is not an interrogative sentence, the process proceeds from step S5 to step S9. When the "question sentence processed flag" of the grammar information holding unit 7 is referred to in step S9, since it is ON, the process proceeds to step S10, and the grammar information is retained. Based on the "future tense" information stored in Part 7, "I will go to
school. "is translated into the future tense.

FIG. 3 is a flow chart showing the flow of processing when the rule relating to the negative question sentence stored in the grammar rule storage unit 6 is applied. Referring to FIG. 3, first, the flag of the grammar information holding unit 7 is initialized (step S10).
1). Next, if there is an untranslated sentence, one sentence is read by the input unit 1 (step S103).

Next, analysis processing is performed (step S10).
4) It is determined whether the negative question flag is ON (step S105). If not, the conversion processing unit 3 performs normal conversion processing (step S106). Next, it is judged whether or not the negative question sentence is converted (step S107),
If it has been converted into a negative question sentence, the negative question flag of the grammar information holding unit 7 is set to ON (step S10).
8).

If the negative question flag is ON in step S105, it is determined whether or not there is a translation word candidate in which a condition "used when answering a negative question sentence" is described (step S109). ), If there is a corresponding translated word, a conversion process is performed using the translated word (step S110).

Next, the negative interrogative sentence flag is set to OFF (step S111). After the processing of step S108 and step S111, the output unit 4 outputs the target language (step S112). Next, the process proceeds to step S102, and the above process is repeated while there are untranslated sentences. When there are no untranslated sentences, the process ends.

As a specific example, a process for sentences 3 and 4 when the Japanese sentence (sentence 3, sentence 4) shown in FIG. 4 is input will be described. When sentence 3 "Do you not go to school today?" Is entered, the negative question flag is not ON, so in step S106, "Don't you go to scho".
ol today? ”. Since this English is a negative question sentence, the negative question flag is set to ON in step S108.

Next, when the sentence 4 is input, the negative question flag is ON. Therefore, "No" is prepared for the translation word candidate of "Yes", and the condition "used when answering the negative question sentence" is set. If it is described, the conversion process is performed using "No" (step S110). Therefore, when answering a negative question sentence, a function of translating "Yes" into "No" can be realized. Similarly, when the English sentences (sentences 3 and 4) in FIG. 4 are input to the input unit 1, the Japanese sentences (sentences 3 and 4) in FIG. 4 are output from the output unit 4.

Here, the aforementioned steps S106 and S
The process of detecting a negative question sentence in 107 will be described in detail by taking as an example the case where the sentence 3 of FIG. 4, “Are you not going to school today?” Is read into the input unit 1.

The conversion processing unit 3, the grammar rule storage unit 6 and the grammar information holding unit 7 shown in FIG. 1 are configured as shown in FIG. That is, the conversion processing unit 3 includes a syntax tree holding unit 31, a translation word assigning unit 22, and a syntax tree conversion unit 33,
The grammar rule storage unit 6 stores “rules for converting to negative sentences”,
The “rule for converting to a question sentence”, the “rule for moving a subject to the beginning of a sentence”, and the “rule for making a preposition” are stored, and the grammar information holding unit 7 stores a “negative rule applied flag”, a “question rule applied flag” and Includes "negative question flag".

First, referring to FIG. 5, the analysis result sent to the conversion processing unit 3 is held in the syntax tree holding unit 31 in the form of a syntax tree.
At this time, the analysis result is expressed like syntax tree 1 with reference to FIG. Next, a translation word is assigned to each element of the syntax tree 1 in the translation word assigning unit 32, and converted into the syntax tree 2. The syntax tree 2 is held in the syntax tree holding unit 31 again.
Next, in the syntax tree conversion unit 33, the rules stored in the grammar rule storage unit 6 are applied to convert the syntax tree,
The process of storing the converted syntax tree in the syntax tree holding unit 31 is repeated, the syntax tree 2 is converted little by little, and the syntax tree 3 is converted.
Get the basic structure of English expressed as.

Next, in the syntax tree conversion unit 33, the “rule for converting into an interrogative sentence” stored in the grammar rule storage unit 6 is applied to the syntax tree 3 and converted into the syntax tree 5, and the grammar information is obtained. The “question rule applied flag” provided in the holding unit 7 is set to ON, and the syntax tree 5 is stored in the syntax tree holding unit 31.

Next, the processing contents in step S107 will be described in detail. Referring to FIG. 5, the syntax tree conversion unit 33 provided in the conversion processing unit 3 stores the contents in the grammar rule storage unit 6. The "negative rule applied flag" of the grammar information holding unit 7 when there are no rules to be applied
If both the "question rule applied flag" and "ON" are set to ON, it can be determined that "a negative question has been generated", so the "negative question flag" is set to ON.

In the above process, "flag O
The process for "N / OFF" is shown in FIG. 7, and the process for "execute the conversion rule" is shown in FIG. In FIG. 6, VERB and VB are verbs, JDO is an auxiliary verb, JO is a particle, NOUN and NN are nouns, AV is an adverb, PR is a preposition, and PG is a word that has no translation. It is shown that.

[0035]

【The invention's effect】

(1) By providing a grammatical information holding unit in the machine translation device, when translating an answer sentence out of the question-and-answer sentence in the input sentence,
Since the grammatical information extracted from the question sentence (question sentence) is used, a correct translation result can be output without outputting an incorrect translation result. (2) As a grammar rule, a grammar rule regarding a negative question sentence can be built in, and a user can obtain a correct translation result without instructing the negative question sentence to be translated.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a main part of a machine translation device showing an embodiment of the present invention.

FIG. 2 is a flowchart showing a flow of processing in the machine translation device shown in FIG.

FIG. 3 is a flowchart showing a flow of processing in the machine translation device shown in FIG.

FIG. 4 is an explanatory diagram showing a relationship between a Japanese sentence and an English sentence translated according to the present invention.

FIG. 5 is a detailed view of a main part of FIG.

FIG. 6 is an explanatory diagram of a syntax tree held in a syntax tree holding unit of the embodiment.

FIG. 7 is a flowchart illustrating a part of the processing flow of the embodiment.

FIG. 8 is a flowchart showing a part of the processing flow of the embodiment.

[Explanation of symbols]

 1 input unit 2 analysis processing unit 3 conversion processing unit 4 output unit 5 dictionary storage unit 6 grammar rule storage unit 7 grammar information storage unit

Claims (4)

[Claims] 1. An input unit for inputting a sentence in a first language, a conversion processing unit for converting an input sentence in a first language into a sentence in a second language, and the input sentence is a question sentence and its answer sentence. And a grammatical information holding unit for extracting and storing grammatical information of the question sentence converted into the second language when the input sentence is the question sentence and its answer sentence. The machine translation device according to claim 1, wherein the conversion processing unit converts the answer sentence input following the question sentence into the second language based on the extracted grammatical information. 2. A grammatical information holding unit extracts and stores, as grammatical information, information indicating the tense of an interrogative sentence, and a conversion processing unit converts an answer sentence input following the interrogative sentence into a second language. The machine translation device according to claim 1, wherein the tense of the answer sentence is matched with the tense of the question sentence based on the extracted information. 3. The first language is Japanese and the second language is English, and the grammatical information holding unit extracts information indicating the type of question sentence as grammatical information, and the conversion processing unit sets the question sentence Is a negative interrogative sentence, when converting the answer sentence input after the interrogative sentence into the second language, the word “yes” indicating affirmation is converted to “No” and the word “no” indicating negative is converted. The machine translation apparatus according to claim 1, wherein the machine translation apparatus converts the machine translation into "Yes". 4. The first language is English, the second language is Japanese, and the grammatical information holding unit extracts information indicating the type of question sentence as grammatical information, and the conversion processing unit sets the question sentence. Is a negative interrogative sentence, when converting the answer sentence that is input following the interrogative sentence into the second language, "No" is converted into the word "Yes" that represents affirmation, and "Yes" is converted into negative. The machine translation apparatus according to claim 1, wherein the machine translation apparatus converts the word into "No".