JPH0785242B2 - Machine translation method - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation system for generating a translated sentence by omitting unnecessary words in a machine translation system.

[0002]

2. Description of the Related Art In the case of translating with a conventional machine translation system, omission of expressions appearing on the input language side has been performed without considering the situation of translation and the characteristics of the target language side.

[0003]

In this conventional translation system, since unnecessary expressions are generated without omission,
Redundant and unnatural sentence is generated. There was a problem that meaningless sentences were generated due to excessive omission. For example, in a situation where there is a dialogue between two parties, "I would like to register for the conferenc
When translating the sentence "e.", the translation is performed without considering the circumstances of the translation and the characteristics of the target Japanese language. Therefore, "I would like to attend the conference", It was sometimes translated into redundant and unnatural sentences to produce the unnecessary expression "I".

The object of the present invention is to eliminate the above drawbacks of the prior art by examining the possibility of omission by referring to the translation situation and the relation with other elements, and by making sure that the omission is properly performed. , Eliminating redundant expressions and allowing the user to easily understand the translated text.

[0005]

The machine translation method of the present invention omits a sentence analysis unit for analyzing an input sentence in a source language given from a sentence input unit, a translation status setting unit for setting a translation status, and an omission. An abbreviated dictionary storage unit that stores possible candidate words, grammatical elements in which this word can be omitted, and dialogue situations that can be omitted, contents of the translation status setting unit and the omitted dictionary storage unit, and contents of translated sentences By comparing with the omission control unit that determines the word to be omitted, and a sentence to be generated in the target language by referring to the word to be omitted sent from the omission control unit and sent to the sentence output unit. And a generation unit.

[0006]

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

FIG. 1 is a functional block diagram showing an embodiment of the present invention. The present invention includes a sentence input unit 1 for inputting an original sentence,
A sentence analysis unit 2 that analyzes an original sentence and outputs an analysis result, a sentence generation unit 3 that generates a sentence of a target language, a sentence output unit 4 that outputs a translated sentence of the generation result, and a situation where translation is performed. , The abbreviated dictionary 7 holding abbreviated candidate words and abbreviated conditions, the sentence generator 3, the translation status setting unit 5 and the abbreviated dictionary 7 are used to determine the abbreviated expressions. The omission control unit 6 is included.

FIG. 2 is a diagram showing an example of the contents of the abbreviation dictionary 7. The abbreviation dictionary is a candidate word for abbreviation and its grammatical elements, corresponding modality expressions that are conditions under which abbreviation can occur,
And holds the translation status that is the condition under which omission can be performed. For example, in Figure 2, the word "I"
It appears as a grammatical element “subject”, and it can be omitted when the modality of the sentence is “hope” and the translation status is “dialogue”.

Next, the operation of this embodiment will be described with reference to FIGS.

The translation status setting unit 5 holds information about the situation under which the sentence is translated. The sentence analysis unit 2 analyzes the input sentence given from the sentence input unit 1 and sends the analysis result to the sentence generation unit 3. The sentence generator 3
Generation is performed based on the analysis result, and the generation result is sent to the omission control unit 6. The omission control unit 6 looks up the omission dictionary 7 with the word of the generation result sent from the sentence generation unit 3 as a key. When there is a matching omission candidate word in the omission dictionary, the omission control unit 6 determines whether the element of the value of the omission dictionary 7 matches the grammatical element of the word in the generation result, or whether the corresponding expression exists in the generation result. , Check whether the translation status matches the value of the translation status setting unit 5. When all of them are true, the omission control unit 6 determines that the omission candidate word can be omitted, and sends the result to the sentence generation unit 3. The sentence generation unit 3 omits the word sent from the omission control unit 6 to perform generation, and outputs the generation result to the sentence output unit 4
Send to.

Next, a specific example will be described. For example, for the translation from English to Japanese in a situation where there is a dialogue between the two parties, "I would like to register
Suppose that the sentence "for the conference." is given as an input sentence. The translation status setting unit 5 determines that the translation status is “2
Holds the value "Inter-person dialogue". As a result of generation by the sentence generation unit 3, ““ I ”is the“ subject ”. "Modality" is "hope." , Etc. are sent to the omission control unit 6. The omission control unit 6 draws the omission dictionary 7 with "I" as a key to obtain the dictionary contents (element: subject, corresponding modality expression: hope, translation status: dialogue). The omission control unit 6
Since the dictionary contents match the generation result and the contents of the translation status setting unit 5, "I" can be omitted and is sent to the sentence generation unit 3. The sentence generation unit 3 omits "I" to perform generation, and outputs the generation result "I want to participate in the conference" to the sentence output unit 4.

[0012]

As described above, the machine translation system according to the present invention can surely carry out an appropriate omission.
The redundant expressions are eliminated, and the user can easily understand the translated sentence.

[Brief description of drawings]

FIG. 1 is a functional block diagram showing an embodiment of the present invention.

FIG. 2 is a diagram showing an example of contents of an abbreviated dictionary.

[Explanation of symbols] 1 sentence input unit 2 sentence analysis unit 3 sentence generation unit 4 sentence output unit 5 translation status setting unit 6 omission control unit 7 omission dictionary

Claims (3)

[Claims] 1. A sentence analysis unit that analyzes an input sentence in a source language given from a sentence input unit, a translation status setting unit that sets a translation status, a candidate word that can be omitted, and this word can be omitted. By comparing the contents of the translation status setting unit and the abbreviated dictionary storage unit with the abbreviated dictionary storage unit that stores grammatical elements and conversation situations that can be omitted, and the contents of the translated sentence, the words to be omitted can be identified. A machine comprising: an omission control unit for determining; and a sentence generation unit for generating a translated sentence in a target language with reference to the words to be omitted sent from the omission control unit and sending the sentence to a sentence output unit. Translation method. 2. A sentence input unit for inputting an original sentence, a sentence analysis unit for analyzing the original sentence input from the sentence input unit and outputting an analysis result, and a sentence for generating a sentence in a target language based on the analysis result. A generation unit, a sentence output unit that outputs the generated sentence, a translation status setting unit that holds the status of translation, an abbreviation dictionary that holds abbreviation candidate words and omission conditions, and the sentence generation unit. A machine translation method comprising: a dialogue status setting unit; and an omission control unit that determines an expression to be omitted based on information in the omission dictionary. 3. The abbreviation control unit draws the abbreviation dictionary using the word of the generation result sent from the sentence generation unit as a key, and when there is a matching abbreviation candidate word in the abbreviation dictionary, the abbreviation dictionary is used. Check whether the element of the value of matches the grammatical element of the word in the generation result, the corresponding expression exists in the generation result, or the translation status matches the value of the translation status setting unit,
If all of them are true, the abbreviation candidate word can be omitted, and the result is sent to the sentence generation unit, and the sentence generation unit omits the word sent from the omission control unit to perform generation, The machine translation system according to claim 2, wherein the generation result is sent to the sentence output unit.