JPH09106398A - Machine translation method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a suitable translated sentence by automatically making clear the relation of a word to be modified and a modifier when translating the words in the relation of modifier and word to be modified in machine translation processing. SOLUTION: This method is provided with a translation dictionary 22 having the concept information of each piece of vocaburary information together with the vocaburary information of a counter language corresponding to the vocaburary information of a source language, grammar dictionary 23 having concept rule information showing the relation of modifier and word to be modified in the concept information, morpheme analytic processing means 3 for dividing the input source sentence in the source language for each word, and translated sentence generating means 4 for generating the translated sentence in the counter language corresponding to the input source sentence by automatically judging whether the relation of modifier and word to be modified is established or not between the words of the input source sentence divided by the morpheme analytic processing means 3 while referring to the translation dictionary 22 and the grammar dictionary 23.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a machine translation method.

[0002]

2. Description of the Related Art A machine translation device for translating sentences from Japanese into English or from English into Japanese by using a computer system has been put into practical use. "Translation of Japanese machine translation systems with prospects for computerization" is introduced in detail.

In the machine translation process, when translating a word having a modified-modified word relationship, an appropriate translated sentence may not be obtained unless there are a plurality of modified words and the relationship with the modified word is not clear. It was

For example, when a "high English dictionary" is entered, it is not possible to determine whether it is "high English" or "high dictionary" in the conventional machine translation, and it is not possible to determine the modifier of the adjective "high". “High” is translated as “high” or “expe”
Sometimes I couldn't get an accurate translation immediately.

Then, the operator needs an editing operation to clarify the modification relation of the input original sentence, which imposes a considerable burden on the operator.

[0006]

SUMMARY OF THE INVENTION The present invention has been made in view of the above-described points, and automatically determines the relation between modification and modification, obtains an accurate translation, and edits the operator. It is intended to provide a machine translation method capable of reducing the burden.

[0007]

According to the present invention, a translation dictionary having concept information of each vocabulary information together with vocabulary information of a source language and corresponding vocabulary information of a partner language, and a modification-modification relationship in the concept information are described. A grammar dictionary having the concept rule information shown, a morpheme analysis processing unit that divides an input source sentence of a source language into words, and a translation source and a grammatical dictionary are referred to, and the input source sentence divided by the morpheme analysis processing unit And a translated sentence generation unit for automatically determining whether or not a modified-modified relation between words is established and generating a translated sentence of the partner language with respect to the input original sentence.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION In the embodiments of the present invention, a Japanese to English translation device will be described as an example, and a schematic configuration diagram of a machine translation device for carrying out the machine translation method of the present invention will be described below.
Shown in the figure.

[0009] (1) is an original sentence input section for inputting an original sentence (Japanese) to be translated, (2) is a dictionary, and a morphological analysis dictionary in which lexical information for decomposing the input original sentence into words (morphemes) is stored. (21), a translation dictionary (22) that stores concept information indicating the concept of each vocabulary information together with the vocabulary information of the other language (English) corresponding to the vocabulary information of the source language (Japanese),
Along with the grammatical information of each language, the grammar dictionary (2) that stores the conceptual rule information indicating the modification of the concept information and the modified
3).

(3) is a morpheme analysis processing unit that divides the original sentence input by the original sentence input unit (1) into words (morpheme) by referring to the morpheme analysis dictionary (21), and (4) is the morpheme analysis unit. A translation generation unit that translates the input original sentence divided into words by the analysis processing unit (3) into a partner language (English) while performing syntax analysis with reference to the translation dictionary (22) and the grammar dictionary (23). , (5) are translation output units that output the translations obtained by the translation generation unit (4) to, for example, a CRT or a printer.

FIG. 2 illustrates a part of the contents of the translation dictionary,
FIG. 3 shows an example of the concept rule information.

Now, when the "high English dictionary" is input from the original sentence input section (1), the morphological analysis processing section (2) refers to the morphological analysis dictionary (21) and sets high / English / Dictionary Adjectives Nouns / adjectives Case particles Divide into words (morphemes) like nouns.

Next, in the translated sentence generation unit (4), in order to determine the modification-modification relationship of the original sentence divided for each word, first the conceptual information of "high", "English", and "dictionary" are translated respectively. From the dictionary (22), the concept information “vertical distance”, “degree”, “value” of “high” as a modifier,
Conceptual information of "English" and "Dictionary" as modified phrases: "Academic", "Book" -Modified rules for grammar dictionary (2
Obtained from the concept rule information in 3).

That is, in this case, "vertical distance" and "degree" do not have a modified-modified relationship with "academic" and "book", and "value" has a modified-modified relationship with "academic". It doesn't hold, but "book" does.

Therefore, "high" representing "value" modifies the "dictionary" representing "book". And
“Expensive” that represents “value” as a translation of “high”
e ”is selected, and an expansive English dict is selected as a translation.
ionary is generated. The generated translated text is output by the translated text output unit (5).

FIG. 4 shows another example of the input original text.
After the morphological analysis processing, the translation generation unit (4) refers to the translation dictionary (22) and the grammar dictionary (23) to determine the modified-modified relation as shown in FIG. 4 and generate the translated text. .

In these translations, other grammatical rules are used, for example, when a noun is followed by "no", the sobo noun and the next noun are arranged, or a word indicating a human is followed by "no". In some cases, it is generated based on various rules such as adding "'s" to the word and placing it before the next noun.

[0018]

As is apparent from the above description, the present invention is as follows.
The vocabulary information of each word has concept information, and the modification of each concept information-The concept rule information that is the rule of the relation to be modified modifies the input source sentence divided into words- It is possible to accurately and automatically determine whether or not the modified relationship is established, and obtain a translated sentence that matches the meaning of the original sentence.

Thus, the operator's burden of editing the original sentence or the translated sentence can be greatly reduced, and the efficiency of the translation work can be improved.

[Brief description of the drawings]

FIG. 1 is a schematic configuration diagram of a machine translation device for implementing the present invention.

FIG. 2 is a diagram showing an example of a translation dictionary.

FIG. 3 is a diagram showing an example of concept rule information.

FIG. 4 is a diagram showing an example of translation.

[Explanation of symbols]

 1 ... Original sentence input unit 2 ... Dictionary 21 ... Morphological analysis dictionary 22 ... Translation dictionary 23 ... Grammar dictionary 3 ... Morphological analysis processing unit 4 ... Translation generation unit 5 ... Translation output section

Claims (1)

[Claims] 1. A first step of dividing an input source sentence of a source language into words by a morphological analysis processing means, and a vocabulary information of a partner language corresponding to the vocabulary information of the source language and a corresponding vocabulary information of a partner language by a translation generation means. A second step of referring to a translation dictionary having concept information and a grammar dictionary having concept rule information indicating a modification-modification relationship in the concept information, and divided by the morpheme analysis processing means in the first step. 3. A machine translation method comprising: a third step of automatically determining whether or not a modification-modified relationship between words in an input source sentence is satisfied, and generating a translated sentence in a partner language for the input source sentence.