JPH11328176A - Machine translation apparatus and computer-readable recording medium recording machine translation processing program - Google Patents

Abstract

(57) [Summary] [Problem] Even when there is no translation corresponding to a dictionary, if this word is necessary for syntax generation, the syntax is generated without deleting the word, thereby improving accuracy. To provide a machine translation device capable of high machine translation processing. SOLUTION: When a flag t attached to an untranslated word is detected for an untranslated word in a first language that does not have a corresponding translated word in a second language (S600: YES), the part-of-speech such as a classifier is determined (S610). The type is determined (S66
0, S670), and because of the relationship between the character strings before and after the connection, if this non-translated word is necessary for syntax generation, it is not deleted (S650: YES, S630: NO, S630: N).
O) Those that can be deleted after a certain processing are processed (S
680, S640) and delete it (S690), and perform a normal syntax generation process (S700) for the word.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation apparatus and a machine translation processing program, and more particularly, to a technique for deleting a non-corresponding phrase at the time of syntax generation after confirming whether or not it can be deleted.

[0002]

2. Description of the Related Art Conventionally, in a machine translation device for translating a source language into a target language, a morphological analysis is performed by searching the source language using a dictionary held by the system, and performing a morphological analysis. In general, a method of converting a word into a target language according to a translation for each word stored in a dictionary is used.

However, between two languages having different cultural backgrounds, translated words do not always correspond one-to-one for each word. For example, in Japanese, the number of objects is represented by a classifier such as "three pencils". However, in English, for example, it is expressed without using a classifier such as "three pencils". In this case, the classifier "book"
Because there is no translation for English, the dictionary for the Japanese morphological analysis system may have a translation for "book", for example, blank, or may indicate that there is no translation by some kind of symbol. ing. Therefore, if the classifier "book" appears, it is processed without translation in the conversion process. Similarly, the word “type” in the case of “automatic cutting blade” is also referred to as “a automati” in English.
c cutting blade ",
Since there is no translation corresponding to "type", the existence of the word "type" may be deleted, and this may be ignored and translated.

[0004]

However, nouns such as "types" and adverbs usually have no corresponding translation, and do not appear in the text of the target language in the conversion process. When this type is used, such as "the environment in which this type operates", since "this type" is the subject of this sentence, it is assumed that there is no uniform translation for "type". If a syntax is generated by deleting it before and ignoring its existence, there is a problem that proper translation cannot be performed because the subject is missing in the syntax and the target language sentence is not established.

The present invention solves the above-mentioned problem. Even when there is no translation corresponding to a dictionary, this word is not uniformly deleted before generating a syntax. To provide a machine translation device capable of performing machine translation processing with higher accuracy by generating syntax by utilizing information necessary for syntax generation added to the word in morphological analysis without deleting the word. With the goal.

Another object of the present invention is to provide a computer-readable recording medium in which a program for causing a computer to execute the above-described machine translation processing is recorded.

[0007]

According to a first aspect of the present invention, there is provided a machine translation apparatus, comprising: an input unit for inputting a first language sentence composed of a source language;
The part of speech of the headword, a translation of the headword or an untranslated word without a corresponding translation, a display to that effect, a dictionary storing grammar information, and the input first language sentence, Translation means for generating and translating a second language sentence in the target language using a dictionary, and at the time of generating the second language sentence,
A second language sentence is established when the non-translated word is detected and translated without generation of a headword of the non-translated word detected by the non-translated word detecting means for detecting the non-translated word; A deletion confirming means for confirming whether or not the non-translated word is deleted even when the second-language sentence is established without the generation of the headword of the non-translated word. Consists of

In the machine translation apparatus having this configuration, even if there is no translation corresponding to the dictionary, this word is not uniformly deleted before generating the syntax, and it is determined whether the word is necessary for generating the syntax. If it is necessary for generation, the word is not deleted, so that even if there is no corresponding translated word, a translated sentence can be output without breaking the syntax.

[0009] In the machine translation apparatus according to the second aspect of the present invention,
In addition to the configuration of the machine translation device according to claim 1, when the headword of the dictionary, the part of speech of the headword, and a translation of the headword or an untranslated word without a corresponding translation are displayed. And dictionary editing means for editing grammar information.

[0010] With the machine translation apparatus having this configuration, the contents of the dictionary can be more easily used in the machine translation apparatus of claim 1, and the function of the machine translation apparatus can be further enhanced.

[0011] In the machine translation apparatus according to the third aspect of the present invention,
In addition to the configuration of the machine translation device according to claim 1 or 2, if the untranslated word is translated by deleting the untranslated word without generating a headword of the untranslated word in the deletion confirming unit, When it is determined that the bilingual sentence does not hold and the deletion is not performed by the deletion unit, a warning unit that warns that the intranslated word is not deleted is provided.

In the machine translation apparatus having this configuration, the user can understand that the untranslated word remains in the translated sentence without being deleted, so that attention can be given to whether or not a correct translation has been performed.

[0013] In the machine translation apparatus according to the fourth aspect of the present invention,
In addition to the configuration of the machine translation device according to any one of claims 1 to 3, the warning unit displays a headword in the first language at a corresponding position in the generated translation in the second language. It is characterized by becoming.

In the machine translation apparatus according to this configuration, the untranslated word that has not been deleted is intuitively located at what position and what content, and subsequent editing is facilitated.

[0015] In the machine translation apparatus of the invention according to claim 5,
In addition to the configuration of the machine translation device according to claim 3 or 4, in addition to the warning means, a guidance display means for displaying a guidance screen including a reason for not being deleted and a countermeasure for not being deleted is provided. It is characterized by having.

In the machine translation apparatus according to this configuration, when there is an untranslated word that has not been deleted, information such as a countermeasure is provided, so that subsequent editing becomes easier.

According to a sixth aspect of the present invention, there is provided a computer-readable recording medium storing a machine translation processing program, wherein an input procedure for inputting a first language sentence consisting of a source language to a computer, a headword, and the headword If the part-of-speech of the word and the translated word of the headword or an untranslated word that does not have a corresponding translated word are displayed, and using a dictionary storing grammatical information, the input first language sentence A translation procedure for generating and translating a second language sentence in a language, a procedure for detecting an untranslated word for detecting the untranslated word at the time of generating the second language sentence, and a procedure for detecting an untranslated word. In the case where the non-translated word is deleted and translation is performed without generating a headword of the translated word, a deletion confirmation procedure for confirming whether or not a second language sentence is established, and the deletion confirmation procedure, Headword generation Ku is also characterized in that to execute the deletion of steps to remove the non-translation when the second language sentence established by.

In the recording medium on which the machine translation processing program according to this configuration is recorded, even if the computer does not have a translation corresponding to the dictionary, the word is not uniformly deleted before generating the syntax, and is not necessary for generating the syntax. If it is necessary to generate a syntax, a machine translation process is performed that does not delete the word and outputs a translation without breaking the syntax even if there is no corresponding translation, if necessary. be able to.

According to a seventh aspect of the present invention, in addition to the configuration of the computer readable recording medium storing the machine translation processing program according to the sixth aspect of the present invention, the computer further comprises: A program for executing a dictionary editing procedure for displaying the part of speech of the headword, a translation of the headword or an untranslated word having no corresponding translation, and editing grammar information. And

[0020] In the recording medium storing the machine translation processing program according to this configuration, in the machine translation processing executed by the machine translation processing program according to the sixth aspect, the contents of the dictionary are further made easy to use. The translation accuracy of the processing can be further improved.

According to an eighth aspect of the present invention, in addition to the configuration of a computer readable recording medium having recorded thereon the machine translation processing program according to the sixth or seventh aspect, the non-translated word is stored in the computer. In the step of confirming the deletion, if it is determined that the second language sentence is not established if the translated word is deleted and translated without generation of the headword of the translated word, it is not deleted in the deleting step,
A warning procedure for warning that the non-translated word is not deleted is executed.

In the recording medium storing the machine translation processing program according to this configuration, the machine translation processing is executed so as to inform the user that the untranslated word remains in the translated sentence without being deleted. You can call attention to whether or not.

According to a ninth aspect of the present invention, in addition to the configuration of the computer readable recording medium storing the machine translation processing program according to any one of the sixth to eighth aspects, the warning procedure includes The method is characterized in that a headword of the first language is displayed at a corresponding position of the generated translation of the second language.

In the recording medium on which the machine translation processing program according to this configuration is recorded, the computer displays the untranslated non-translated words in the machine translation processing so that the user can intuitively understand the location and the content of the untranslated words. The subsequent editing is facilitated.

According to a tenth aspect of the present invention, in addition to the configuration of a computer-readable recording medium storing the machine translation processing program according to the eighth or ninth aspect, the recording medium further comprises: On the computer,
A guide display procedure for displaying a guide screen including a countermeasure for the reason why the file was not deleted and the countermeasure for the fact that the file was not deleted is executed.

In the recording medium on which the machine translation processing program according to this configuration is recorded, when a non-translated word is not deleted by the computer, a guidance display for giving information such as a countermeasure is executed in the machine translation process. Subsequent editing becomes easier.

[0027]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a block diagram showing an embodiment of a machine translation apparatus and a machine translation processing program according to the present invention. In this embodiment, an example will be described in which the source language (first language) to be translated is translated into Japanese and the target language (second language) to be translated is translated into English.

First, FIG. 1 is a block diagram showing a schematic configuration of a machine translation apparatus according to the present embodiment. As shown in FIG. 1, the machine translation apparatus according to the present embodiment is used for an input unit 2 such as a keyboard for inputting a Japanese original to be translated, various operation instructions, and the like, and for translation processing. Various dictionaries such as a basic dictionary 12 and a user dictionary 14 are stored, and an external storage device 4 such as a hard disk or an optical disk
And a display unit 6 including a CRT or the like for displaying an input Japanese original sentence or an English translation as a translation result, and a display control unit 7 for controlling data displayed on the display unit 6.
And a translation unit based on data input from the input unit 2 or the external storage device 4, and a printing unit 8 including a laser printer or the like for printing and outputting an English translation or the like as a translation result, The translation result is displayed on the display unit 6,
The control unit 10 controls the output of the printing unit 8 and the like.

The input unit 2 includes a character key (not shown) for inputting characters, an input sentence edit key for starting various processes, a translation execution key, a user dictionary edit key, a registration key, a delete key, and a cancel key. Keys, various function keys such as an output statement edit key, an end key, and a print execution key; cursor keys used for editing input statements and output statements; and for operating the screen displayed on the display unit 6; An enter key, a next candidate selection key, and the like are provided.

The external storage device 4 stores a basic dictionary 12 and a user dictionary 14 used for translation processing.

The dictionary used by the machine translator is a basic dictionary 12 which is a system dictionary which the machine translator has in advance as a part of the system, and is also edited for each specialized field such as medical terms. There is a specialized dictionary (not shown) and a user dictionary 14 that the user independently builds. The specialized dictionary and the user dictionary 14 are arbitrarily used, and the basic dictionary 12 alone can operate as a system. When the user dictionary 14 and the specialized dictionary are used, they are stored in the external storage device 4 and used. Further, the specialized dictionary may be configured such that a plurality of dictionaries can be simultaneously present for each of various specialized fields, and the user dictionary 14 can be simultaneously present for each user or each of the fields of use. In the present embodiment, a description will be given in a situation where only one type of the basic dictionary 12 and the user dictionary 14 is used for convenience.

FIG. 2 shows an example of the data structure of the basic dictionary 12. Each dictionary registration information has a headword 30
For a part of speech 32 of each headword, a translation 34 for each headword, and a word having no translation in the second language corresponding to the headword, a no-flag t is set in the no-flag 36 area. It is used for a syntax generation process described later. here,
In the present application, a word having no translation in the second language corresponding to the headword in the first language is referred to as “untranslated word” for convenience.
Further, the basic dictionary 12 includes dictionary information 38 necessary for other translations.

Note that the user dictionary 14 has basically the same configuration as the data configuration of the basic dictionary 12. However,
While the contents of the basic dictionary 12 cannot be changed by the user without permission, the contents of the user dictionary 14 can be operated by the user operating the user dictionary editing key. It is possible to register untranslated words in the same way. Note that a configuration may be adopted in which untranslated words can be registered only in the user dictionary. In the morphological analysis process (FIG. 6: step 250 (hereinafter, step is abbreviated as S and abbreviated as S250)), if registered in the user dictionary 14, the basic dictionary 12
, The content of the user dictionary 14 is read out.

Next, the control unit 10 will be described. The control unit 10 is configured by a known microcomputer including a CPU 18, a ROM 20, a RAM 22, an input / output port 24, and a bus line 26 connecting these components. Then, in the control unit 10, various processes are executed by the following programs stored in the ROM 20.

The control program 202 controls input / output of this embodiment and each program to control the whole.

The translation program 204 comprises a morpheme analysis program 206, a syntax analysis program 208, a conversion program 210, a generation program 212, and a morpheme generation program 214, and converts a Japanese sentence, which is a source language sentence input to the machine translation device, into This is a program for translating into English sentences that are target language sentences.

The morphological analysis program 206 searches the Japanese sentence as the first language using the basic dictionary 12 and the user dictionary 14 for the same character string. The dictionary information of the column header is read and added to the character string. In this way, the words are decomposed into words, and at the same time, information on the words is stored for later translation processing.

The syntactic analysis program 208 is a program for describing a Japanese sentence, which has been subjected to morphological analysis into words, based on grammatical information described in a dictionary, in a modification relation in a tree structure.

The conversion program 210 converts words in the source language into the target language, and here converts Japanese into the corresponding English.

The generation program 212 generates a syntax in which an input sentence described in a tree structure and converted into English is reconstructed into an English sentence according to the grammar of the target language.

Further, the morpheme generation program 214 performs morpheme generation for adjusting details in accordance with the grammar and generating a final translated sentence.

The untranslated word detection program 215 converts the “untranslated word” of the input sentence morphologically analyzed in the translation generation processing (FIG. 6: S280), that is, the translated word of the second language corresponding to the first language into the dictionary. For words that do not have the corresponding translated word no-flag “t” instead of the translated word in the morphological analysis, an untranslated word that does not have a corresponding translated word is detected by detecting this no-flag t (FIG. 5: S600).

The non-translated word deletion confirmation program 216 generates a syntax when the non-translated word is deleted according to the procedure shown in FIG. 5 described later (FIG. 6: S280).
Judge whether an appropriate second language sentence can be established, specifically, whether a tree structure can be formed, and determine that the second language sentence can be established even if the headline of the intranslated word is deleted Only when this is done, the deletion is permitted (FIG. 5: S610 to S680).

The non-translated word deletion program 217 deletes the non-translated word headword when the non-translated word deletion confirmation program 216 allows the deletion of the head word of the non-translated word in the generation process (FIG. 6: S280). Figure 5: S69
0) It is a program.

If the untranslated word is not deleted, it remains as the heading in the first language even after the conversion process (FIG. 6: S270), and is further deleted by the untranslated word deletion program 217. If not, it will be present in the second language as it is as the headword of the notation in the first language.

The user dictionary editing program 218 is a program for editing the contents of the user dictionary 14.

The non-deletion warning program 219 displays the non-translated word when the non-translated word deletion confirmation program 216 makes the deletion impossible, and warns the user that the deletion could not be performed.

The guide display program 220 is for displaying a guide for the cause of the undeleted headword warned by the undeleted warning program 219 and what to do with the user. .

The RAM 22 stores an original Japanese sentence input from the input unit 2 and an English translation translated and edited by the translation and editing processing.

The work area 222 is an area for temporarily storing character strings and parameters for various processes.

The source language sentence storage area 224 is an area for storing the input source language sentence in the form of text data.

The translated sentence storage area 226 is an area for storing translated target language sentences in the form of text data.

The input / output port 24 is an interface through which the control unit 10 inputs and outputs data and the like of the input unit 2, the display unit 6, the printing unit 8, and the like. The interface serves as an interface for inputting and outputting, and inputting from various input devices such as a voice recognition device, and can constitute the input means of the present invention in this sense.

The bus line 26 includes an address bus, a data bus, and a control bus for exchanging data of each device.

Next, the translation and editing processing executed by the control unit 10 according to the flowcharts shown in FIGS. 3 to 5 will be described.

FIG. 3 is a flowchart showing an outline of the translation and editing processing executed by the control unit 10. As shown in FIG. 3, when the translation and editing processes are executed, first, the user operates a character key or the like of the input unit 2 (S110), and waits for input of a Japanese original text to be translated. . Here, the input of the source sentence may be configured to read the source language sentence storage area 224 that has already been created. Then, when a Japanese original sentence (hereinafter, input sentence) is input, the input sentence is displayed in a predetermined area of the display unit 6 (S12).
0).

Next, it is determined whether or not the input sentence editing key provided on the input unit 2 has been operated (S130). If not (S130: NO), the process proceeds to S150. If the input sentence edit key is operated (S13
0: YES), execute the input sentence editing process (S14)
0). When the input sentence editing process is executed, the user inputs the input sentence displayed in a predetermined area of the display unit 6 to the input unit 2.
Can be directly corrected using the cursor key, delete key, insert key, character key, etc. If the input sentence edit key has not been pressed, it is determined whether or not the translation execution key provided on the input unit 2 has been operated (S
150), if not operated (S150: NO),
The process proceeds to S190, and if the translation execution key is operated (S150: YES), a translation process for translating a Japanese input sentence into an English translation (hereinafter referred to as an output sentence) is executed (S160). The output sentence obtained by the translation process is stored in the translated sentence storage area 226 of the RAM 22 as a pair with the input sentence (S170), and the input sentence and the output sentence are respectively displayed in predetermined areas provided on the display unit 6. Yes (S18
0).

The translation process may be performed on all input sentences at once, or may be performed by specifying a range for each sentence and only for the specified range of sentences.

If the translation execution key has not been pressed (S
150: NO), it is determined whether or not the user dictionary editing key provided on the input unit 2 has been operated (S190). If not (S190: NO), the process proceeds to S210. If the dictionary edit key is operated (S19)
0: YES), execute the user dictionary editing process (S20)
0). When the user dictionary editing process is executed, the user can correct or delete the contents of the information registered in the user dictionary 14 or register a new required headword.

In S210, it is determined whether or not the output sentence editing key provided in the input unit 2 has been operated. If not (S210: NO), the process proceeds to S230. If the output sentence edit key is operated (S210: Y
ES), an output sentence editing process is executed (S220). still,
When the output sentence editing process is executed, the user proceeds to S160.
The output sentence (English translation) displayed in a predetermined area of the display unit 6 by executing the translation process in
Similarly to the input sentence editing process of 40, it is possible to directly make corrections using a cursor key, a delete key, an insert key, a character key and the like provided on the input unit 2.

If the output sentence edit key is not pressed (S
210: NO), it is determined whether or not the end key provided on the input unit 2 has been operated (S230), and if not (S230: NO), another key, for example, a translated output sentence is A print execution key or the like that is output to the printing unit 8 and printed.
Other key operations are detected, and a process corresponding to the operated key is executed (S240).

That is, S130: NO, S150: NO,
S190: NO, S210: NO, S230: NO Waiting for input to wait for operation of any one of the input sentence edit key, translation execution key, user dictionary edit key, output sentence edit key, and end key in each step. A loop is formed.

Then, the input sentence editing process (S14)
0), display of input sentence and output sentence (S180), user dictionary editing process (S200), output sentence editing process (S220),
After other key processing (S240) is processed, S130
Returns to the input standby state, and repeats the same processing.

If the end key is operated (S230:
YES), this process ends (END).

As described above, in the translation and editing processes, after the Japanese original text is input, the input text editing process, translation process, user dictionary editing process, and the like are performed in accordance with the operation of various keys provided on the input unit 2. And various processes such as an output sentence editing process. Then, if the translation execution key was operated and the translation process was executed, the obtained output sentence (English translation)
If the input is incorrect or unnatural, operate the input sentence edit key to execute the input sentence editing process, and then re-edit the Japanese original sentence, or edit the user dictionary. After the user dictionary 14 is edited by operating the keys, the translation execution key is operated again to re-translate, or the output sentence editing key is operated to execute the output sentence editing process, and the output sentence English Edit the translation directly.

If the translation result desired by the user can be obtained by repeating such processing, the print key is operated to output the translation result from the printing unit 8 or to save it in the external storage device 4 as text data. Then, the end key is operated to end the processing.

Next, according to the flowchart shown in FIG. 4, this is executed in S200 in the flowchart shown in FIG. 3, and the user corrects or deletes the contents of the user dictionary 14, or registers a new necessary headword. The user dictionary editing process to be performed will be described.

First, editing is started, and a user dictionary editing key is operated. Since the processing is not ended here (S500: N
O) On the screen for inputting the headword and part of speech, wait for the user to input the headword and part of speech that the user wants to edit (S51).
0). If the headword and the part of speech are input (S510: YE
S), it is determined whether the input headword is already registered in the user dictionary 14 (S520). If already registered (S520: YES), the user dictionary 14
In step S540, a headword, a translated word, or other dictionary information is read in (S540), and if not already registered (S520: NO), a template for editing a new headword is prepared (S530). . These data are called up, or the translated word or the corresponding translated word no flag t and other dictionary information are edited by the user in the headword for which the template is prepared (S550).
It is determined whether editing of all necessary dictionary information has been completed for each part of speech (S560), and if not completed (S560).
(560: NO), the process returns to S550, and the translation, the flag t, and the dictionary information are input again, and if the input is completed (S56).
0: YES), all information is registered in the user dictionary 14 (S570), and the process returns to the state of waiting for input again in preparation for editing of the next headword (S500: NO, S510:
NO).

In S510, when the headword and the part of speech are not input (S510: NO), the end key is operated (S500: YES) to end this processing (E).
ND).

Next, according to the flowchart shown in FIG. 6, the translation process executed in S160 in the flowchart shown in FIG. 3 to translate an input sentence (original sentence in Japanese) into an output sentence (translated in English) will be described.

First, when the translation execution key is operated, S2
At step 50, a headword is searched for with reference to predetermined judgment rules in the morphological analysis program 206 and headwords, parts of speech, and dictionary information in the basic dictionary 12 and the user dictionary 14. A morphological analysis process is performed to add information such as translations and parts of speech to the morphological analysis results for the headwords searched from the basic dictionary 12 or the user dictionary 14. If the user dictionary 14 and the specialized dictionary are used, priority is given to the user dictionary 14, the specialized dictionary, and the basic dictionary 12 in this order.

At S260, the result of the morphological analysis processing at S250 is received, and the analysis processing for analyzing the Japanese structure is executed by using a predetermined analysis grammar stored in the syntax analysis program 208 of the ROM 20. The relationship between the character strings divided into words is analyzed by the analysis, and the relationship between the sentences is constructed by a tree structure.

In S270, a conversion process for converting a Japanese word analyzed by a syntax analysis process into an English translation using a predetermined conversion grammar stored in the conversion program 210 of the ROM 20 is executed.

At S280, the word order of English is determined from the English structure converted by the conversion process, using a generation grammar predetermined by the generation program 212 stored in the ROM 20.

In S290, the morpheme generation program 214
Allows you to generate more English-specific articles, supplement required subjects in output sentences even if they do not exist in the input sentence,
Morphological generation processing for processing the tense, aspect, person, numerical features, and the like is performed to generate an output sentence composed of an English word string as a final translated sentence, and the translation processing ends.

FIG. 5 is a flowchart showing the generation process in S280 of FIG. 6 in more detail.

First, when the conversion process (FIG. 6: S270) is completed, the process proceeds to the generation process (FIG. 6: S280). First, the non-flag = t is detected (S600). If no flag = t is not detected (S600: NO), it is not necessary to determine whether deletion is possible, and other generation processing (S700) is immediately performed. The other generation processing is performed by a well-known generation processing, and a description thereof will be omitted. On the other hand, if no flag = t is detected (S600: YES), that is, in the case of an untranslated word, if there is no corresponding translation word, the part-of-speech of the untranslated word which is a headword with no flag = t is referred to the result of morphological analysis. Then, it is determined whether it is a classifier (S610). Here, if the input word is a classifier such as "-sheet" or "-book" (S61)
0: YES), it is determined whether the front of the classifier is a number (S620), and if it is a number (620: YES), it is immediately deleted (S690).

If it is not a number (S620: NO),
Next, it is determined whether the front is an adjective (S630). This assumes a case in which Japanese is an article number such as "number" and is translated into an adjective such as "several" in English. If the front is an adjective (S630: YES), the front adjective is set as a pronoun (S640).

Next, if the untranslated word is not a classifier (S610: NO), it is determined whether there is a modifier in front (S650), and if there is a modifier (S650: YE).
S) If the headword is deleted, the syntax may be broken. Therefore, the headword is sent to other generation processing without deletion (S700).

If there is no modifier (S650: NO), it is determined whether or not the part of speech of the untranslated word is a noun (S660). If it is not a noun (S660: NO), it is deleted as an adverb or the like that can be deleted (S690). If it is a noun (S660: YES), it is further determined whether or not the headword is a main noun (S670). Here, the "main noun" is a noun that is the center of meaning in a compound noun and is usually located at the last part. Therefore, the last element of a compound noun can be determined as a main noun. If it is not a main noun, it can be determined that there is no significant role in meaning (S670: NO), and it is deleted (S6).
90). On the other hand, if it is a main noun (S670: YE
S) When the main noun is gone, the information as the compound word is gone. Therefore, it is necessary to move the compound noun before the deletion, and the information of the main noun is moved to the immediately preceding noun (S680). If the information of the main noun is transferred to the preceding noun, even if it is the main noun, it has no meaning to be continued and is deleted (S690).

For the words that have not been deleted, syntax generation processing is continuously performed (S700), a tree (tree) structure is described, the generation processing (FIG. 6: S280) ends, and morpheme generation processing (FIG. 6: The process proceeds to S290). In the morpheme generation process (FIG. 6: S290), the morpheme generation program 214 stored in the ROM 20 uses a morpheme generation rule that is predetermined, and the English word string generated by the generation process (FIG. 6: S280) is generated. On the other hand, a morpheme generation process (FIG. 6: S290) for generating an output sentence by generating an inflected form such as a plural form of a noun, a past verb, a past participle, etc., and ends the translation process (FIG. 3: S160) I do.

Hereinafter, the processing of the machine translator according to the present embodiment will be described using a specific example along the flowchart shown in FIG. Here, the input sentence is "Set the automatic cutting blade"
The description will be made with reference to the “type” in the case of. In this case, when the morphological analysis processing (S250) is performed, "automatic (prefix) / type (noun) / cut (noun) / blade (noun) /
(Verb) / set (verb) ".

This morpheme is composed of the compound noun “auto / type /
The relationship of the syntax that "cut / blade" exists is a tree
It is described in a structure (S260).

In the next conversion process (S270),
"Automatic (automatic) / no flag = t (type) /
"cutting (cutting) / blade (blade)". However, the language in parentheses indicates the language before conversion.

Here, the generation procedure will be described with reference to the detailed flowchart of S280 of FIG. 6 shown in FIG.
Since the “type” has the non-flag t without a corresponding translation added thereto, “YES” is determined in S600, and it is determined whether there is a classifier next (S610). Because it is not a classifier (S6
10: NO), the presence or absence of the modifier is determined (S65).
0), since there is no modifier (S650: NO), it is determined whether the word is a noun (S660). In this example, as described above, since the result of the morphological analysis is added if the compound noun is a compound noun (S660: YES), it is determined whether or not it is the main noun (S670). The main noun refers to a noun that is central in meaning in a compound noun, and in this example, “blade / blad”
"e" is the main noun. Therefore, the “type” is not a main noun, and thus is determined to be NO in S670 and is deleted (S690). By the above processing, "type"
Is not generated. As a result, in this example, an article or the like is added and “An automatic cutt” is added.
ing blade is set. "
It is generated by the morpheme generation procedure (S290).

On the other hand, similarly, when the "type" is used,
When used in the context of "the environment in which this type operates", the treatment is different. Hereinafter, a similar description will be given. First, when morphological analysis (S250) is performed, morphemes such as "this (adnominal) / type (noun) / (particle) / operating (verb) / environment (noun)" are analyzed.

FIG. 8 is an example of a description in a tree structure after the syntax analysis and the syntax generation. When the syntax analysis (S260) is performed based on the result of the morphological analysis, the fact that the "type" is modified by the adjectival "this" and "this" and "type" constitute the main part, and "operation" The syntax that modifies the "environment" is analyzed, and described in a tree structure as shown in FIG. 8A. This is converted (S2
70) Then, “this (this) / no flag = t
(Type) / work (operate) / circumstan
ce (environment) ". This structure is shown in the figure,
The result is as shown in FIG.

The next generation process will be described with reference to the flowchart of FIG. 5. Since there is no flag = t, S600 is determined to be YES and is not a classifier (S610: N
O), since it has been modified to the adverb "this", S650
Then, it is determined as YES, and other generation processing is performed without being deleted (S700), and the processing here ends (EN
D), the words are arranged in word order according to the English grammar (S280), and the final English sentence is generated by the processing of morpheme generation (S290), but the "type" is not deleted, but it corresponds to this. Since there is no translation, "type" is displayed as it is in the English sentence.

FIG. 7 is a diagram showing a screen display showing an example of a message for calling attention and a guidance display by a help display, which is displayed when an intranslated word is not deleted. That is, as shown in FIG.
nice that this [type] works. To get the English sentence. As shown in this figure, when "type" is deleted, it is unnecessary and is not displayed at all, but if not deleted, for example, it is surrounded by a frame,
It is desirable to give a warning of post-editing by shading or blinking.In addition, by displaying a message that calls attention, in order to not forget the processing, and to suggest further processing to be performed It is desirable to provide guidance by displaying help.

Next, a description will be given of an example of an input sentence "Gully", which means that an abnormal sound is generated.
First, when the morphological analysis (S250) is performed, the morpheme is decomposed into morphemes of "garigari (adverb) / abnormal sound (noun) / (particle) / occurrence (verb)". Next, based on the results of this morphological analysis, the adverb “garrigari” modifies the verb “occurs”, and the noun “abnormal sound”, which is a noun, becomes the subject of the verb “occurs”. (S
260), and this is stored as a tree structure relationship. Then, a conversion process (S270) is performed based on the translated word information of the morphological analysis result, and “no flag = t (gully)
/ Noise (abnormal sound) / come out (occurs) ". This is sent to the syntax generation processing (S2
80), an English sentence is generated. However, since “Garigari” has no flag = t and no translation exists, S600 is determined as YES, and “Garigari” is an adverb and not a classifier (S61).
0: NO), it is then determined whether there is a word that modifies “garigari” (S650), and there is no such modifier from the syntax generation result (S650: NO), and “garigari” is an adverb. Is not a noun (S66)
0: NO) is deleted (S690: NO). Then, in the next morpheme generation processing (S290), "garigari" has already been deleted, so the final English sentence "A noisec"
Omes out "is generated. In this case, the indication "Garigari" does not appear at all in the final translation English.

Finally, "two types of automatic cutting blades"
If you explain the sentence containing the word "type" as an example,
First, in the morphological analysis (S250), morphemes such as "2 (number) / type (number) / (particle) / automatic (prefix) / cut (noun) / blade (noun) / type (noun) /" Decomposed. Then, the result of this syntax analysis (S26
0), the syntax that "two types" modify the compound noun "auto / cut / blade / type" is analyzed. In the conversion process (S270), these are converted on the basis of the translated sentence information that is not possible in the result of the morphological analysis, and "two (2) / kind (type)
/ Automatic (automatic) / cutting (cutting) / blade (blade) / no flag = t (type) ", and the syntax is analyzed in the next syntax generation processing (S280). First, for the “type” with no flag = t, S600
Is determined to be YES, it is next determined whether or not it is a classifier (S610). In rare cases, “type” is also used as a classifier, but in this case, it is determined as a noun by morphological analysis and is not a classifier (S610: N
O) Then, it is determined whether or not there is a modifier for modifying the “type”, and the compound word “automatic / cutting / b” is determined.
There is "two / kind" that modifies "
Since there is no modifier for modifying "blade", it is determined here that there is no modifier (S650: NO), and it is determined whether or not it is a noun, but since "type" is a noun (S660: YE).
S), it is further determined whether or not the subject is a main noun (S670).
In this case, the “type” is “automatic / cutt”.
Although the compound noun is formed together with "ing / blade", "type" is the main noun since "type" is the center of the compound in terms of meaning (S670: YES).

In an actual system, the number of steps required to perform a semantic analysis increases, so that even a simple determination that the last word of a compound noun is the main noun is considered to be practically acceptable. It is. Then, the “type” determined as the main noun is obtained by transferring the information obtained by the morphological analysis to the front noun, in this case, “blade” (S680), and then deleted (S690). Then, the syntax generation processing is continuously performed (S280). Here,
The syntax follows the English grammar by exchanging qualifications. In other words, the Japanese qualification / modification relation "N1 (no) N
"2" to "N2 (of) N1" in English. However, in this example only, since “kind” is used, the order is not changed as an exception.

After that, a plurality of changes are made according to English grammar.
Generation of an article, morpheme generation such as personal change, tense change, etc. is performed (S290) to generate an English sentence as the final translation. As a result, "Two Kinds of Automatic"
cutting blades ".

As is apparent from the above specific examples, the machine translation apparatus according to the present embodiment does not unnecessarily translate words having no corresponding translation.
On the other hand, instead of just having a translation, we don't delete it all the time; instead, we use the word in detail and reflect it in the translation, and if necessary, send a message that alerts the user. The message is posted to urge the user to check, so that the user does not notice the result of the process performed in a place where the user cannot see, so that the translated sentence desired by the user is not undesired.

In this embodiment, the control program 2
02 functions as the input means of the present invention together with the input unit.
Reference numeral 2 indicates that the CPU 18 causes the computer to execute the input procedure of the present invention. Further, the basic dictionary 12 corresponds to the dictionary of the present invention. The translation program 204 is executed by the CPU 18
Function as the translation means of the present invention, and cause the computer to execute the translation procedure of the present invention. The morphological analysis program 206 causes the CPU 18 to function as the morphological analysis means of the present invention, and causes the computer to execute the morphological analysis procedure of the present invention. The untranslated word detection program 215 causes the CPU 18 to function as the untranslated word detection means of the present invention, and executes the procedure of untranslated word detection.
The non-translated word deletion confirmation program 216 causes the CPU 18 to function as the deletion confirmation means of the present invention, and executes the procedure of the deletion confirmation. The non-translated word deletion program 217 constitutes the deletion means of the present invention, and causes the deletion procedure to be executed. The user dictionary editing program 218 causes the CPU 18 to function as a dictionary editing unit of the present invention, and to execute a dictionary editing procedure. The undeleted warning program 219 causes the CPU 18 to function as a warning unit of the present invention, and executes a warning procedure. The guidance display program 220 causes the CPU 18 to function as guidance display means of the present invention, and to execute a guidance display procedure.

The present embodiment can be modified as follows. For example, languages are not limited to Japanese and English sentences.

If the translated word must not be deleted, the headword is generated as it is this time, but it is also possible for the user to register a translated word used only in such a case. Until now, user dictionaries always generate constant translations anytime, anywhere.However, giving users control to generate or not translate translations according to the situation where headwords are used like this If this is done, the freedom of the user will be expanded, and the translation accuracy can be improved and post-editing can be reduced.

As described above, the present invention has been described with reference to one embodiment. However, it is needless to say that the present invention is not limited to the above-described embodiment, and that various modifications can be made without departing from the scope of the claims. No.

[0099]

As is apparent from the above description, claim 1
In the machine translation apparatus according to the invention, even if there is no translation corresponding to the dictionary, the word is not uniformly deleted before generating the syntax, and it is determined whether the word is necessary for generating the syntax. When necessary, there is an effect that the translated sentence can be output without deleting the word and without breaking the syntax even if there is no corresponding translated word. Therefore, it is possible to reduce the possibility that an erroneous translation is caused due to a uniform deletion and the meaning is not understood, thereby improving the translation accuracy.

According to the machine translation apparatus of the second aspect,
In the machine translation device according to the first aspect, there is an effect that the contents of a dictionary that utilizes the function of the translator can be obtained.
Therefore, it is easy to use, the function of the machine translation apparatus can be further enhanced, and the translation can be performed with higher accuracy.

In the machine translation apparatus according to the third aspect,
There is an effect that the user can understand that the untranslated word remains in the translated sentence without being deleted. Therefore, if the translation is not deleted, the user will be alerted, and the user will not know that the untranslated word has not been deleted, and the translation will not be different from the translation intended by the user. There is an effect that the accuracy of the image can be improved.

In the machine translation apparatus according to the fourth aspect,
This has the effect of intuitively knowing the location and content of the untranslated words that have not been deleted. Therefore, there is an effect that the translated text translated in the subsequent editing is easily corrected.

The machine translation apparatus according to claim 5 has the effect that when there is an untranslated word that has not been deleted, information such as a countermeasure to be taken thereafter is given. That is, if the reason why the data is not deleted is known, the dictionary can be corrected, or the subsequent editing can be easily performed even if the knowledge is small, and a high-precision machine translation result can be obtained if the given countermeasures are taken. It has the effect of being able to do so.

In a computer-readable recording medium on which the machine translation processing program according to the present invention is recorded, even if the computer does not have a translation corresponding to the dictionary, the word is uniformly read before generating the syntax. Judge whether it is necessary for syntax generation without deleting, and if necessary for syntax generation, do not delete the word and output the translation without breaking the syntax even if there is no corresponding translation There is an effect that it is possible to execute a machine translation process that can perform the above. In particular, there is an effect that a general-purpose computer can execute a machine translation process using a recording medium on which this program is recorded, without using a dedicated machine as exemplified in the embodiment.

According to the recording medium of the present invention, in addition to the effects of the machine translation processing executed by the machine translation processing program of the sixth aspect, the contents of the dictionary are made easier to use. There is an effect that the accuracy can be further increased.

According to the recording medium of the eighth aspect, there is an effect that a machine translation process can be executed so as to warn a user that a non-translated word remains in a translated sentence without being deleted in a computer. Therefore, there is an effect that it is possible to urge the user to check whether or not an unintended translation has been performed and whether or not a correct translation has been performed.

According to the recording medium of the ninth aspect of the present invention, the machine translation processing is performed by the computer so that the user can intuitively recognize from the screen the location and content of the untranslated words that have not been deleted. There is an effect that the procedure displayed in is executed. For this reason, there is an effect that the subsequent editing of the translation by the user is facilitated.

In the recording medium according to the tenth aspect of the present invention, when there is an untranslated word that has not been deleted in the computer, a warning indicating that the untranslated word has not been deleted is sent to the computer together with information on the cause and subsequent measures. The effect is that the given machine translation process is executed. Therefore, there is an effect that post-editing of a translated sentence or editing of a dictionary is further facilitated.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a schematic configuration of a machine translation device according to an embodiment.

FIG. 2 is a diagram showing an example of a data configuration of a basic dictionary 12.

FIG. 3 is a flowchart showing an outline of a translation and editing process executed by a control unit 10;

FIG. 4 is a detailed flowchart of a user dictionary editing process S200.

FIG. 5 is a detailed flowchart of generation processing S280 of translation processing S160.

FIG. 6 is a flowchart showing a translation process executed in S160 to translate an input sentence (a Japanese original sentence) into an output sentence (an English translated sentence).

FIG. 7 is a diagram showing a screen display showing an example of a message for calling attention and a guidance display by a help display, which is displayed when an intranslated word is not deleted.

FIG. 8A is a diagram showing a tree structure in parsing. (B) is a diagram illustrating a tree structure in syntax generation.

[Explanation of symbols]

 2 input unit 4 external storage device 6 display unit 7 display control unit 8 printing unit 10 control unit 12 basic dictionary 14 user dictionary 18 CPU 20 R0M 22 RAM 24 input / output port 26 bus line 202 control program 204 translation program 206 morphological analysis program 208 Syntax analysis program 210 Conversion program 212 Generation program 214 Morphological generation program 215 Non-translated word detection program 216 Non-translated word deletion program 217 Non-translated word deletion program 218 User dictionary editing program 219 Undeletion warning program 220 Guidance display program 222 Work area 224 Source language sentence Storage area 226 Translation storage area

Claims (10)

[Claims] 1. An input means for inputting a first language sentence consisting of a source language, a headword, a part of speech of the headword, and a translation of the headword or an untranslated word if there is no corresponding translation. And a dictionary storing grammar information, a translation unit configured to generate and translate the input first language sentence into a second language sentence in the target language using the dictionary, and the second language sentence When generating, the translated word is detected by the translated word detecting means for detecting the translated word, and if the translated word is deleted and translated without generating the translated word headword detected by the translated word detection means, the second language sentence Deletion confirming means for confirming whether or not holds, and in the deletion confirming means, a deleting means for deleting the untranslated word when the second language sentence is established without generation of the headword of the untranslated word. Machine translation device equipped with 2. A dictionary editor for editing a headword of the dictionary, a part of speech of the headword, a translation of the headword or an untranslated word having no corresponding translation, and a grammar information. 2. The machine translation apparatus according to claim 1, further comprising means. 3. The method according to claim 1, wherein the second language sentence is not established if the untranslated word is deleted and translated by the deletion confirmation unit without generating the untranslated word headword. 3. The machine translation apparatus according to claim 1, further comprising a warning unit that warns that the intranslated word is not deleted when the translation is not deleted. 4. The method according to claim 1, wherein the warning unit displays a headword in the first language at a corresponding position of the generated translation in the second language. A machine translation device according to any of the above. 5. A guidance display means for displaying a guidance screen including a reason for not being deleted and a countermeasure for not being deleted in addition to the warning means. 5. The machine translation device according to 4. 6. An input procedure for inputting a first language sentence consisting of a source language to a computer, a headword, a part of speech of the headword, and a translation of the headword or an untranslated word without a corresponding translation. In the case, using a dictionary that stores a display to that effect and a grammar information, the input first language sentence is translated into a second language sentence in the target language by translating the second language sentence. At the time of generating a language sentence, the steps of detecting the intranslated word to detect the intranslated word, and when translating by deleting the untranslated word without generating a headword of the untranslated word detected by the procedure of the untranslated word detection,
A step of confirming whether or not a second language sentence is established; and a step of confirming the deletion, wherein the step of confirming deletion is performed when the second language sentence is satisfied without generation of a headword of the untranslated word. And a computer-readable recording medium storing a machine translation processing program for executing a deletion procedure. 7. The computer according to claim 1, wherein the headword of the dictionary, the part of speech of the headword, a translation of the headword or an untranslated word having no corresponding translation, and grammatical information are displayed. 7. The computer-readable recording medium according to claim 6, further comprising a program for executing a dictionary editing procedure for editing. 8. The computer according to claim 1, wherein the non-translated word is determined to not be a second language sentence if the non-translated word is deleted and translated without generating a headword of the non-translated word in the deletion confirmation procedure. 8. A machine translation processing program according to claim 6, wherein a warning procedure for warning that the intranslated word is not deleted is executed if the deletion is not performed by the deletion procedure. Computer readable recording medium. 9. The warning procedure according to claim 6, wherein the warning procedure is a procedure of displaying a headword of the first language at a corresponding position of the generated translation of the second language. A computer-readable recording medium that records the machine translation processing program according to any one of the above. 10. A guide display procedure for causing the computer to display a guide screen including a reason for not being deleted and a countermeasure for not being deleted, together with the warning procedure. A computer-readable recording medium recording the machine translation processing program according to claim 8.