JPH07230464A - Machine translation device - Google Patents

Abstract

(57) [Summary] [Purpose] Appropriate translated word selection for Japanese phrases and clause expressions that can be replaced with a single word in a foreign language without increasing the information added to each word in the bilingual dictionary. A machine translation device capable of performing. [Structure] A word meaning expression example database containing expression examples in which the meaning of another part of speech of a foreign language word is expressed in the form of a phrase or clause containing a Japanese word that is associated with the foreign word in the bilingual dictionary. Set up. A Japanese sentence is converted into a foreign language sentence through a Japanese intermediate expression and a foreign language intermediate expression. When converting the Japanese intermediate expression into the foreign language intermediate expression, it is judged whether or not the phrase or clause in the Japanese intermediate expression matches any of the expression examples included in the meaning expression example database (S
3, S4), while determining another part-of-speech form corresponding to the matching example, the bilingual word of the Japanese word included in the phrase or section is retrieved from the bilingual dictionary (S6), and the retrieved bilingual word is a foreign language. It is converted to the determined other part-of-speech form using a dictionary, and the converted other part-of-speech form is set as a parallel translation word of the phrase or clause (S7).

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION The present invention relates to a machine translation device, and more particularly to
The present invention relates to a machine translation device that translates Japanese sentences into foreign language sentences.

[0002]

2. Description of the Related Art Conventionally, a machine translation device, particularly a machine translation device for translating a Japanese sentence into a foreign language sentence, has a translation selecting method for replacing a Japanese word with a foreign word word by word. , Describe co-occurrence information in the bilingual dictionary,
Based on this, there was a method to replace Japanese phrases and clauses with one foreign word.

[0003]

However, in a machine translation device that uses a word-by-word translation method for replacing words with words, it is not possible to replace phrases and clauses in Japanese sentences with one foreign word, which is redundant. There is a problem that a foreign language sentence is generated. For example, in English, the phrase "attractive" can be expressed as a single word "attractive", so that a phrase or clause in a Japanese sentence that contains a Japanese word (here, "attractive") is In some cases, the foreign word corresponding to the word word (in this case, "attraction") can be expressed in another part-of-speech form (in the above example, the adjective form). However,
In a machine translation device that uses a word-for-word translation method that replaces words with words, a Japanese sentence that says "attractive" is usually translated as "~ that has attraction".

Further, it is difficult for a machine translation device that replaces a Japanese phrase or clause with one word in a foreign language using the conventional co-occurrence information to cope with the variety of expressions of the phrase or clause. There's a problem. For example, co-occurrence information or the like needs to be increased as parallel translation information of “charm” in a parallel translation dictionary in order to support expressions such as “full of charm” and “full of charm”. Doing the same thing for all words has the drawback that the bilingual dictionary becomes too large.

The present invention has been made in order to solve the above problems, and it is possible to replace a Japanese phrase or clause with a single word in a foreign language without increasing the information added to each word in the bilingual dictionary. It is an object of the present invention to provide a machine translation device capable of appropriately selecting a translated word for the expression by.

[0006]

In order to achieve the above object, the present invention provides a Japanese input means for inputting a Japanese sentence,
A Japanese analysis unit that analyzes the Japanese sentence input by the Japanese input unit using a Japanese dictionary and outputs a Japanese intermediate expression; and a Japanese translation unit that outputs the Japanese intermediate expression output by the Japanese analysis unit. A conversion means for converting into a foreign language intermediate expression using the meaning expression example database; and a foreign language sentence generation means for generating a foreign language sentence from the foreign language intermediate expression obtained by the conversion means using a foreign language dictionary, The word meaning expression example database includes an expression example in which the meaning of another part of speech of a foreign language word is expressed in the form of a phrase or clause including a Japanese word that is associated with the foreign language word in the bilingual dictionary. If the phrase or clause in the Japanese intermediate expression matches any of the expression examples included in the meaning expression example database, the converting means determines the other part-of-speech form corresponding to the matching example. In addition, the bilingual word of the Japanese word included in the phrase or clause is extracted from the bilingual dictionary, the extracted bilingual word is converted to the other part-of-speech form determined using the foreign language dictionary, and the conversion is performed. There is provided a machine translation device having a part-of-speech form as a parallel translation of the phrase or clause.

[0007]

[Function] The Japanese analysis unit analyzes the Japanese sentence input by the Japanese input unit using a Japanese dictionary and outputs it as a Japanese intermediate expression, and the conversion unit translates the Japanese intermediate expression into a parallel translation. The foreign language intermediate representation is converted using the dictionary and the word meaning expression example database, and the foreign language sentence generation means generates the foreign language sentence from the foreign language intermediate representation using the foreign language dictionary.

In addition, the word sense expression example database represents the other part-of-speech word sense of a foreign language word in the form of a phrase or clause including a Japanese word that is associated with the foreign language word in the bilingual dictionary. When converting a Japanese intermediate expression into a foreign language intermediate expression, the conversion means includes an expression example and whether the phrase or section in the Japanese intermediate expression matches any of the expression examples included in the meaning expression example database. Whether or not another part-of-speech form corresponding to the matching example is determined, the bilingual word of the Japanese word included in the phrase or clause is extracted from the bilingual dictionary, and the retrieved bilingual word is extracted in the foreign language dictionary. Is used to convert to the determined other part-of-speech form, and the converted other part-of-speech form is used as a parallel translation of the phrase or clause.

[0009]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings.

FIG. 1 is a block diagram showing a schematic configuration of a machine translation device according to an embodiment of the present invention.

In the figure, reference numeral 1 is a Japanese input section which is composed of a keyboard or the like and is used for inputting a Japanese sentence to be translated. 2 is a disk device for storing various dictionaries and example database of meaning representations;
Is a memory for temporarily storing various data.

A CPU 4 controls the operation of the apparatus of this embodiment according to a control program stored in the control memory 5.
(Central controller). This control memory 5 is a ROM
It may be a (read only memory) or a RAM (random access memory). Reference numeral 6 is a display for displaying a Japanese sentence input by the Japanese input unit 1 or a translated sentence of the Japanese sentence, and 7 is a bus for transmitting various data.

Further, as shown in FIG. 2, the disk device 2
Stores a Japanese dictionary 2a, a bilingual dictionary 2b, a word sense expression example database 2c, a foreign language dictionary 2d, etc., and the CPU 4 has a Japanese analysis unit 4a, a conversion unit 4b, a foreign language sentence generation unit 4c, etc. It is composed by.

Next, the operation of the apparatus of this embodiment will be described using an example of translating a Japanese sentence into an English sentence.

FIG. 3A is a diagram showing an example of various expression examples set in the word meaning expression example database 2c of FIG. 2. For example, the database 2c-1 is a general notation form of Japanese nouns. Is N, an English noun corresponding to the N is Ne, an adjective form of the Ne is Ae, and various meanings of the Ae are typified by using the N.
It is a collection of 3, ... Also, database 2
In c-2, V is the general notation of Japanese verb, Ve
Represents the English verb corresponding to V, and ADVe represents the adverbial form of Ne in the database 2c-3.

First, when a Japanese sentence 8 "She is an attractive woman" shown in FIG. 3B is input from the Japanese input unit 1, the Japanese analysis unit 4a shown in FIG. The sentence 8 is parsed into a Japanese intermediate expression 9 as shown in the tree diagram of FIG. 3B, and the clause “attractive” in the sentence 8 is replaced by the noun “female”. It is determined that the section is related to. Since this clause is a solidarity modifier, there is a possibility that it can be converted into one English adjective. Therefore, when translating this section, the conversion unit 4b determines whether or not the entire section can be expressed by one English adjective, before selecting the noun "charm" or the verb "arru". The database 2c-1 is searched and checked.

That is, since "charm" in the above section "charm" is a noun, the "charm" is replaced with the general notation N of the Japanese noun, and the search character string "N is" is replaced. create. Next, the word meaning expression example database 2c-1 is searched to determine whether or not there is an expression example that matches the search character string “with N”. Here, the search character string "N
"Is present" means expression example 3 in the database 2c-1.
The phrase “Attractive” is a Japanese noun N
It is determined that the adjective form Ae of the English noun Ne corresponding to (“charm”) can be expressed, and as shown in FIG. 3C, the English noun “at” corresponding to the noun “charm” is extracted from the bilingual dictionary 2b.
traction ", and the adjective form" at "of the noun" attraction "in the foreign language dictionary 2d (English dictionary 2d in this embodiment).
"tractive" is obtained, and the adjective form "attractive" is translated into the phrase "attractive".

Also, the Japanese intermediate expression 9 shown in FIG.
The other elements in the above are also translated into corresponding English by the conversion unit 4b and converted into foreign language intermediate expressions corresponding to the Japanese intermediate expressions 9 and foreign language generation is performed based on the foreign language intermediate expressions. The section 4c further uses the foreign language dictionary 2d to
e is an attractive lady. "

Next, using the flowchart (FIG. 4),
The procedure of translation processing of phrases and clauses in Japanese sentences including the above-described translation processing will be described.

First, from the Japanese intermediate representation of the Japanese sentence input by the Japanese input unit 1, a phrase or section J (W
j) (for example, "attractive") is extracted (step S1). Here, Wj represents a specific Japanese word (for example, the noun “charm”).

Next, the Wj is replaced with a general notation form Xj of a Japanese word (if the word is a noun, for example, N in FIG. 3A), and a search character string J (Xj) (for example, N) is replaced. ,
"With N") is created (step S2). Next, the word meaning expression example database 2c is searched (step S3),
It is determined whether there is an expression example that matches the search character string J (Xj) (step S4). If there is no expression example that matches the search character string J (Xj), J (Wj) is verbatim translated by a normal translation process (step S5), and this processing ends. In step S4, if there is an example that matches the search character string J (Xj), the other part-of-speech form corresponding to the example is determined (for example, since "with N" is a noun modifier, another part-of-speech). The form is determined to be an adjective form, and a foreign word Wf (eg, English noun “attraction”) corresponding to the Japanese word Wj (eg, noun “charm”) is retrieved from the bilingual dictionary 2b (step S6). ), The other part-of-speech form Yf of this Wf
(For example, the adjective form "attrac" of the English noun "attraction"
tive ”) is taken out from the foreign language dictionary 2d, and the Yf is taken as the translated word of J (Wj).

As described above, in this embodiment, the bilingual dictionary 2b is used.
In, without adding co-occurrence information etc. to each word,
In the phrases and clauses of the Japanese sentence, those that can be replaced by one word such as an adjective or adverb in a foreign language are determined and converted, so that the phrase or phrase of the Japanese sentence can be converted without increasing the translation dictionary 2b. It is possible to select an appropriate translated word for a clause.

Further, in the above-described embodiment, one other part-of-speech correspondence pattern (for example, a conversion pattern from an English noun Ne corresponding to a Japanese noun N to its adjective form Ae, or a Japanese verb V). For example, a conversion pattern from an English verb Ve to its adjective form Ae) is provided with one meaning expression example database (for example, database 2c-1 or database 2c-2 in FIG. 3A). Was there.

For example, the database 2c-1 is a collection of expression examples representing the meaning of "having the property and ability of N". However, when it is considered that the semantic relationship between a word and other part-of-speech forms, such as an English noun and its adjective form, is classified into multiple types,
A plurality of word meaning expression example databases corresponding to each may be prepared.

For example, if the English noun Ne corresponding to the Japanese noun N ("meaning") is "meaning", then the adjective Ae is set to "meaningless" (no meaning adjective). Consider the case where the above Ae has the meaning of "not having the property or ability of N". On the other hand, from the Ne, the adjective Ae
Even in the conversion pattern to, separately from the database 2c-1,
In FIG. 3, a word meaning expression example database such as the database 2c-4 shown in FIG. In this case, in the foreign language dictionary 2d, there may be a plurality of types of links that associate words with other parts of speech. In the example described above, two types of links, "Ne → Ae (adjective)" and "Ne → Ae (negative adjective)" may be prepared.

By doing so, it is possible to replace various Japanese expressions represented by phrases and clauses with a single word such as an adjective or an adverb in a foreign language without increasing the size of the bilingual dictionary 2b. It is possible to judge and convert anything.

In the above embodiment, the case where the foreign language is English is shown, but the present invention is not limited to this, and similar processing can be applied to German, French, Spanish, Russian and the like. it can.

[0028]

As described above, in the machine translation device of the present invention, the meaning of another part of speech of a foreign language word includes a Japanese word that is associated with the foreign language word in the bilingual dictionary. A database of semantic meaning examples including expression examples expressed in the form of phrases and clauses is provided. When converting Japanese intermediate expressions to foreign language intermediate expressions, phrases and clauses in Japanese intermediate expressions are
Judge whether or not it matches any of the expression examples included in the meaning expression example database, determine the other part-of-speech form corresponding to the matching example, and determine the parallel word of the Japanese word included in the phrase or section. The foreign word is extracted from the bilingual dictionary, converted into the other part-of-speech form determined using a foreign language dictionary, and the converted other part-of-speech form is used as the bilingual word of the phrase or clause. It is possible to judge and convert Japanese phrases and clauses that can be expressed by, without significantly increasing the information to be written in the bilingual dictionary, and to translate simple foreign language sentences with few redundant expressions.

[Brief description of drawings]

FIG. 1 is a block diagram showing a schematic configuration of a machine translation device according to an embodiment of the present invention.

FIG. 2 is a diagram showing details of a disk device 2 and a CPU 4 of a machine translation device according to an embodiment of the present invention.

FIG. 3 is a diagram showing an example of a translation operation by a machine translation device according to an embodiment of the present invention.

FIG. 4 is a flowchart showing a flow of control processing in a translation operation by the machine translation apparatus according to the embodiment of the present invention.

[Explanation of symbols]

 1 Japanese input unit (Japanese input unit) 2 Disk device 2a Japanese dictionary 2b Bilingual dictionary 2c Meaning expression example database 2d Foreign language dictionary 4 CPU 4a Japanese analysis unit (Japanese analysis unit) 4b Conversion unit (conversion unit) 4c Foreign language sentence generation unit (foreign language sentence generation means)

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Kazue Kaneko 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Minor Fujita 3-30-2 Shimomaruko, Ota-ku, Tokyo Canon Inc. (72) Within the corporation

Claims (1)

[Claims] 1. A Japanese input means for inputting a Japanese sentence,
A Japanese analysis unit that analyzes the Japanese sentence input by the Japanese input unit using a Japanese dictionary and outputs a Japanese intermediate expression; and a Japanese translation unit that outputs the Japanese intermediate expression output by the Japanese analysis unit. A conversion means for converting into a foreign language intermediate expression using the meaning expression example database; and a foreign language sentence generation means for generating a foreign language sentence from the foreign language intermediate expression obtained by the conversion means using a foreign language dictionary, The word meaning expression example database includes an expression example in which the meaning of another part of speech of a foreign language word is expressed in the form of a phrase or clause including a Japanese word that is associated with the foreign language word in the bilingual dictionary. If the phrase or clause in the Japanese intermediate expression matches any of the expression examples included in the meaning expression example database, the converting means determines the other part-of-speech form corresponding to the matching example. In addition, the bilingual word of the Japanese word included in the phrase or clause is extracted from the bilingual dictionary, the extracted bilingual word is converted to the other part-of-speech form determined using the foreign language dictionary, and the conversion is performed. A machine translation device having a part-of-speech form as a parallel translation of the phrase or clause.