JPH05120329A - Machine translation device - Google Patents

Abstract

PURPOSE:To obtain a correct translated result by classifying participial modifiers into an amount adjective, verb derivative adjective and material adjective or the like and arranging those adjectives in the prescribed order of the classification when there are plural participial modifiers related to one substantive. CONSTITUTION:The first language sentence inputted from an input part 11 is edited into any suitable form by an editing part 12. At a translation part 13, morpheme analysis is executed by a morpheme analysis part 13a, and syntax analysis is executed by a syntax analysis part 13b. A transformation part 13c transforms the analyzed first language sentence into the second language sentence. Next, a syntax generation part 13d generates the correct syntax based on the grammatical rule of the second language sentence, and a morpheme generation part 13e executes the changes of word forms or the like. In this case, the participial modifiers are classified into the amount adjective, verb derivative adjective and material adjective or the like according to the characters and when there are plural participial adjectives related to one substantive, those participial adjectives are successively arranged in the prescribed order of the classification. The translated result sentence is freely edited by the editing part 12, displayed on a display part 15 and printed out by a printing part 16.

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 machine translating from a first language into a second language using a computer, and more particularly to a translation means for controlling the word order of adnominal modifiers. is there.

[0002]

2. Description of the Related Art Conventionally, in a machine translation device, when translating a noun accompanied by a plurality of adnominal modifiers,
The word order of adjectives was determined by using grammatical rules or semantic features. However, the word order of verb-derived adnominal modifiers has not been clarified. For example, taking a Japanese-to-English translation as an example, a verb-derived adnominal modifier in Japanese is generally translated in English as a relative clause. Furthermore, in this case, in order to simplify the sentence, the pronouns are often omitted and translated using the participle form of the verb. For example, when "added parts" is translated into English, it is translated as "the parts which were added", and further translated as "the added part".

[0003]

However, the participle form of this verb is not always before the modified noun,
Depending on the type of verb, it may be practically placed after the noun. In addition, when there are a plurality of adnominal modifiers that modify the same noun, for example, when a noun with the same determiner and number is modified, such as "this two added parts",
The word order of these words and the verb-derived adnominal modifiers was not uniform and was disjoint. Such inconsistent translation becomes a problem especially when the second language which is the translation target word is a language whose word order is severely restricted such as English. That is, in English, the word order of adnominal modifiers is accurately determined, and a translation that is contrary to the word order results in an unnatural impression to the reader. However, memorizing the word order of all verb-derived modifiers requires a lot of memory capacity and is uneconomical.

The present invention has been made to solve the above-mentioned problems, and when translating a first language having a verb-derivative adjunct modifier, a verb-derivative adjunct modifier is added to the translation dictionary of the second language. The purpose of this is to obtain correct translation results by adding information on the word order of, and using that information to generate a second language sentence according to the grammatical rules of the second language.

[0005]

To achieve this object, a machine translation apparatus of the present invention provides an input means for inputting a first language sentence which is a sentence to be translated, and an input first language sentence. A translation dictionary that stores information necessary for translation into the second language sentence that is the translation target sentence, a translation unit that translates the first language sentence into the second language sentence using the translation dictionary, and the translated first sentence. An output means for outputting a bilingual sentence,
The translation means classifies adnominal modifiers into quantitative adjectives, verb-derived adjectives, material adjectives, etc. according to their properties, and when there are multiple adnominal modifiers related to one noun, the adnominal modifiers are sequentially ordered in a predetermined classification. It has a control means for arranging. Specifically, the translation dictionary has an analysis dictionary that stores information about the first language, a conversion dictionary that translates the first language into the second language, and a generation dictionary that stores information about the second language. ing. Also, the translation means includes an analysis means for analyzing the first language sentence using the dictionary information, a conversion means for converting the analyzed first language sentence into a second language sentence, and a converted first language sentence. And a generating unit for correctly generating the bilingual sentence based on the grammatical rule of the second language. More specifically, the generation dictionary has information about the word order of adnominal modifiers, and the generation means has grammatical rules about the word order of adnominal modifiers.

Further, in order to achieve this object, the machine translation apparatus of the present invention has an input means for inputting a first language sentence which is a translation target sentence, and an input first language sentence as a translation target sentence. A translation dictionary that stores information necessary for translating into a certain second language sentence, a translation unit that translates the first language sentence into the second language sentence using the translation dictionary, and the translated second language sentence. The translation dictionary stores information on whether the verb-derived adnominal modifier should be before or after the noun to be modified, the adjunct modifier being derived from the verb. The translating means has a control means for allocating the verb-derived adnominal modifiers to the front and back of the body word based on the front-back information of the storage means.

[0007]

The machine translation apparatus of the present invention having the above-mentioned configuration is
A method that determines the word order of modifiers based on generated grammatical rules by using the information in the generated dictionary when parsing a first language with adnominal modifiers and converting it to a second language to generate a second language sentence Is. The user inputs the first language sentence that is the translation target sentence by the input means of the machine translation device of the present invention. The translation dictionary stores information necessary to translate the input first language sentence into the second language sentence which is the translation target sentence. Of the translation dictionaries, the generation dictionary has information about the word order of the adnominal modifier.

The translation means translates the first language sentence into the second language sentence using the translation dictionary. Further, of the translation means, the analysis means analyzes the first language sentence using the dictionary information.
The conversion means converts the analyzed first language sentence into a second language sentence. The generating unit correctly generates the converted second language sentence based on the grammatical rule of the second language. Here, the generation means has grammatical rules regarding the word order of the adnominal modifiers. The translation means classifies the adnominal modifiers into quantitative adjectives, verb-derived adjectives, material adjectives, etc. according to their properties, and when there are a plurality of adnominal modifiers related to one noun, the adnominal modifiers are sorted in a predetermined order. Are arranged in sequence. The output means outputs the translated second language sentence.

Further, the translation dictionary of the machine translation device of the present invention stores information about adnominal modifiers derived from verbs, which verb-derived adnominal modifier should be before or after the noun to be modified. ing. Then, the translating means sorts the verb-derived adnominal modifiers before and after the wording based on the anteroposterior information in the storage means.

[0010]

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 schematically showing the hardware configuration of a machine translation device according to the present invention. The machine translation device 1 includes a data processing unit 10 configured by a computer, an input unit 11 configured by a keyboard and the like, and a C
The display unit 15 includes an RT and the like, and the printing unit 16 includes a scanning printer and the like. And
In the data processing unit 10, an editing unit 12, which is realized as a software program using the functions of a computer,
A translation unit 13 and a translation dictionary 14 stored in a hard disk, an optical disk, or the like are built in.

Next, the basic operation of this machine translation device will be briefly described. The first language sentence input from the input unit 11 is edited by the editing unit 12 into an appropriate form as needed, and then sent to the translation unit 13. The editing unit 12 is not directly related to the present invention, and thus detailed description thereof is omitted. FIG. 2 is a flowchart schematically showing the configuration of a system program for realizing the function of the translation unit 13. The first language sentence sent to the translation unit 13 is morphologically analyzed by the morphological analysis unit 13a (S21), and further syntactically analyzed by the syntax analysis unit 13b (S22).
It is sent to the conversion unit 13c.

The converting unit 13c converts the analyzed first language sentence to the second language sentence (S23). The second language sentence converted by the conversion unit 13c is the syntax generation unit 1
After the correct syntax is generated in 3d based on the grammatical rule of the second language sentence (S24), the morpheme generating unit 13e changes the word form and the like to complete the process (S25).
The processing in the translation unit 13 described above is performed in the translation dictionary 1
4 is performed with reference to the word information stored in 4.
Further, the obtained translation result sentence is freely edited by the editing unit 12 and displayed on the display unit 15 or the printing unit 16
Printed out by.

In the system configured as described above, when the process of the present invention is performed, that is, when the first language sentence having the verb-derived adnominal modifier is translated into the second language sentence, the generation information and the syntax generation of the translation dictionary 14 are performed. The process of determining the word order of modifiers based on the grammatical rule of the copy will be described below by taking the translation from Japanese to English as an example. For example, a case will be described in which the Japanese words "many added iron parts" and "these two specified parts" are input from the input unit 11 in FIG. The input sentence sent to the translation unit 13 includes the morphological analysis unit 13a and the syntax analysis unit 13
b, the conversion unit 13c converts the intermediate structure as shown in FIGS. 3 and 4 (S21, S22, S2 in FIG. 2).
3). That is, in FIG. 3, the morphological analysis unit 13
a, the syntactic analysis unit 13b decomposes the Japanese sentence "many additional iron parts" into parts of speech,
For parts of speech that are changing, such as "added",
It is converted into its basic form and arranged. Next, the conversion unit 13
c is translated into an English word having the same part of speech. The same applies to “these two parts designated”.

FIG. 5 to FIG. 8 are flow charts showing in detail the processing in the syntax generation unit 13d which performs the processing (S24 in FIG. 2) of generating the second language sentence from this intermediate structure. The flow of S24 of FIG. 2 will be described below with reference to these figures. FIG. 5 is a diagram showing the overall flow of the syntax generation processing in S24. The intermediate structure that has undergone pre-processing such as compound word processing (S51), verb processing (52), noun phrase processing (S
53), the adverb processing is received (S54), and the processing proceeds to post-processing such as generation of a period (S55). The process of generating the noun phrase in S53 is further shown in detail in FIG. Number processing (S61), possessive processing (S62), adjective processing (S
63), relative clause processing (S64), prepositional phrase processing (S6)
5) and adverb processing (S66). However, these steps are processed only when necessary depending on the content of the first language sentence.

FIG. 7 is a diagram showing in detail the flow of adjective processing in S63, and FIG. 8 is a diagram showing in detail the flow of processing of the relative clause in S64. 9 and 10 show an example of specific storage of a part of the translation dictionary 14 shown in FIG. 9 shows the contents of the English adjective dictionary and FIG. 10 shows the contents of the English verb dictionary.

First, "a lot of added iron parts"
Will be described as an example. The input sentence converted into the intermediate structure as shown in FIG. 3 proceeds to the noun phrase generation process of S53 and undergoes the adjective process of S63. In the case of English, adjectives are often statistically arranged in the order of quantitative adjectives, verb-derived adjectives, and material adjectives, so here, conversion is performed in that order. However, since the verb-derived adjective is classified into a verb as a part of speech in the morphological analysis process (S21) and the syntactic analysis process (S22), it is first generated as a relative clause using a relative pronoun, and later, a relative pronoun, etc. Is often omitted and often becomes a verb derivation adjective.

That is, the information on the types of adjectives is obtained from the English adjective dictionary shown in FIG. 9, and according to the flow of FIG.
The quantity adjective many and the material adjective iron in this order
In other words, it is generated as "many iron parts" (S71, S
72). After that, the processing proceeds to the relative clause processing of S64, and S of FIG.
After the relative clause is generated in 81, it is determined in S82 whether or not the relative clause can be omitted. If the relative clause can be omitted, it is confirmed in S83 whether or not there is verb-derived adjective postfix information by referring to the English verb dictionary. To do.

That is, in this example, first, the relational clause "many
iron parts which was added ”was generated (S8
3), of which "which was" can be omitted, so "a"
Although the verb dictionary of "dd" is referred to, since "add" has no verb derivative adjective postfix information (S83), it is at the same position as the verb derivative adjective in Fig. 7, that is, in front of the noun, between the quantity adjective and the material adjective And the participle form of the verb, that is, “ad
ded ”is generated (S85). Therefore, the“ many added iron parts ”is defined as“ many added iron parts ”.
Is translated in the order of words.

Next, "these two parts are designated"
Will be described as an example. The input sentence converted into the intermediate structure as shown in FIG. 4 proceeds to the noun phrase generation process of S53, undergoes the number processing of S61 and the adjective process of S63 shown in FIG. 6, and is generated as “these2 parts”. .. Then S
Proceeding to the relational clause processing of 64, after the relational clause is generated in S81 shown in FIG. 8, it is confirmed in S82 whether the relational clause can be omitted. Here, “these 2 parts which wasindicate
Since "which was" of d "can be omitted," indicate "
The verb dictionary is referred to (S83). As shown in FIG. 10, since “indicate” has relative clause postfix information, the participle of the verb is generated after the modified noun (S8
4), the translation result "these 2 parts indicated" is obtained.

When there are a plurality of adjectives related to one noun, the position where the verb-derived adjective should be placed is processed after the quantitative adjective and before the material adjective.
Second language sentences in a statistically correct order can be generated. In this way, since the noun context information regarding the verb derivation adjective is given to the verb, it is possible to generate a sentence in which the verb derivation adjective is placed at the correct position before and after the noun.

In this embodiment, the case where Japanese is translated into English has been described, but the present invention can be applied to the case of translating into German, French, Russian, etc. and back translation.

[0022]

As is apparent from the above description, according to the present invention, a generation dictionary is used when a first language sentence having a verb-derived adnominal modifier is analyzed, converted into a second language sentence, and generated. Since the word order of the modifiers is determined based on the generated grammar rules using the information of, it is possible to obtain the statistically most frequent word order translation result for the second language sentence that is the translation target sentence. , The user does not need to correct the word order of adjectives by post-editing, and the whole translation time can be shortened.

Further, since the verb stores the before-and-after information for the noun of the verb-derived adnominal modifier and positions the verb-derived adjective based on that, statistically regarding the second language sentence as the translation target sentence. It is possible to obtain the translation result of the most frequent word order, and the user does not need to correct the word order of the adjectives by post-editing, and the overall translation time can be shortened.

[Brief description of drawings]

FIG. 1 is a block diagram schematically showing a hardware configuration of a machine translation device that is an embodiment of the present invention.

FIG. 2 is a flowchart showing an overall flow of a translation process which is an embodiment of the present invention.

FIG. 3 is an explanatory diagram showing an intermediate structure of one sentence in the Japanese-to-English translation according to the embodiment of the present invention.

FIG. 4 is an explanatory diagram showing an intermediate structure of another sentence in translation from Japanese to English according to an embodiment of the present invention.

FIG. 5 is a flowchart showing the overall flow of syntax generation processing.

FIG. 6 is a flowchart showing a flow of noun phrase generation processing.

FIG. 7 is a flowchart showing a flow of processing for generating an adjective that modifies a noun.

FIG. 8 is a flowchart showing a flow of relative clause generation processing.

FIG. 9 is an explanatory diagram showing a part of stored contents of an English adjective dictionary used in an embodiment of the present invention.

FIG. 10 is an explanatory diagram showing a part of stored contents of an English verb dictionary used in an embodiment of the present invention.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Machine translation device 10 Data processing unit 11 Input unit 13 Translation unit 13a Morphological analysis unit 13b Syntax analysis unit 13c Conversion unit 13d Syntax generation unit 13e Morphological generation unit 14 Dictionary 15 Display unit 16 Printing unit

Claims (2)

[Claims] 1. An input unit for inputting a first language sentence which is a translation target sentence, and information necessary for translating the input first language sentence into a second language sentence which is a translation target sentence. A translation dictionary, a translation means for translating the first language sentence into a second language sentence using the translation dictionary, and an output means for outputting the translated second language sentence, wherein the translation means , Adnominal modifiers are classified according to their properties into quantitative adjectives, verb-derived adjectives, material adjectives, etc., and when there are a plurality of adnominal modifiers related to one noun, the adjunct modifiers are sequentially arranged in the predetermined classification order. A machine translation device having a control means. 2. An input means for inputting a first language sentence which is a translation target sentence, and information necessary for translating the input first language sentence into a second language sentence which is a translation target sentence. A translation dictionary, a translation means for translating the first language sentence into a second language sentence using the translation dictionary, and an output means for outputting the translated second language sentence, wherein the translation dictionary is , Regarding adnominal modifiers derived from verbs,
The verb-derived adnominal modifier has storage means for storing information about whether it should be before or after the noun to be modified, and the translating means stores the verb-derived adnominal modifier before and after the information in the storage means. A machine translation device having a control means for allocating before and after the wording based on the above.