JPH11282846A - Machine translation apparatus and computer-readable recording medium recording machine translation processing program - Google Patents

Abstract

(57) [Summary] [Problem] By making use of a part of a fixed sentence or absorbing fluctuations of the ordinary and honorific forms, a fixed sentence can be used even if it is not a strict input sentence. Provided is a machine translation apparatus capable of obtaining a translation result sentence with high accuracy and reducing the load of post-editing. SOLUTION: It is detected whether or not a fixed headline sentence exists in an input sentence (S400), and whether or not the input sentence has a utilization part (S410). When the utilization part includes a utilization part, a morpheme replaceable for each utilization part. It is detected whether or not the condition of the solution of the analysis is satisfied (S430). If the condition is satisfied, the sentence is replaced with a fixed form translation (S420), and the translation process is terminated. If the fixed phrase entry is not included, the normal translation process is continued.

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 for translating an original sentence composed of a source language into a target language and a computer-readable recording medium recording a machine translation processing program. The present invention relates to a technique for appropriately translating a fixed sentence even if it is used.

[0002]

2. Description of the Related Art Conventionally, in this type of machine translation apparatus, a plurality of types of dictionaries are generally prepared. In a translation process, a dictionary most suitable for a text to be translated is used. Is configured.
Here, as the type of the prepared dictionary, for example,
A wide range of words and phrases are registered in general translations and grammar, and a basic dictionary stored in advance by a machine translation device and a specialized dictionary in which words and translations used only in a specific field are registered And a user dictionary that the user can independently edit. Specify the range of the vocabulary of the input sentence that the user wants to translate, and select and build an appropriate dictionary,
Provides more accurate translation results.

[0003] At the same time, fixed phrase registration is also effective in providing high-level translation results. Even in manual translation processing, it is often necessary to replace a specific phrase with a specific translation, instead of performing literal translation. A simple example is a greeting expression. From a manual to an in-house document, this one sentence should always be this translation,
In order to obtain stable translation accuracy without being affected by the presence or absence of a user dictionary or specialized dictionary, fixed phrase registration is effective. If the fixed phrases are not registered, the fixed phrases, unique expressions, and idioms used by the user are literally translated, so that the translated sentences have no meaning or unnatural translations are generated. The greater these quantities, the more extensive the user has to do post-editing. User dictionaries and fixed phrases that users can freely edit are indispensable for obtaining natural translation results. In particular, fixed phrases are replaced with ideal translations, so it is difficult to find perfect translation accuracy. It is very effective in machine translation systems. Japanese Patent Application Laid-Open No. 7-152767 discloses the following technique as one of the registration methods. That is, a fixed sentence is registered in a pattern file as a pair of an original sentence pattern expressed by a combination of a symbol representing an arbitrary phrase and a fixed phrase and a corresponding translated sentence pattern, and the pattern processing unit inputs the original sentence input from the input unit. Is checked to see if it matches the registered original pattern, and if it matches, a word corresponding to the symbol of the original pattern is cut out. The translation control unit controls the translation unit to translate the extracted words one by one. The translated sentence synthesis unit receives a translated sentence pattern corresponding to the original sentence pattern from the pattern processing unit, and creates a final translated sentence by replacing the translation result of the cut-out phrase received via the translation control unit with the symbol of the translated sentence pattern. . The pattern file is such that the user can easily register and update.

[0004]

However, a fixed sentence can be replaced only when the whole sentence exactly matches in the input sentence (although some wildcards can be used), which is an advantage. However, on the contrary, it is not possible to cope with a subtle fluctuation of expression. In addition to the problem of swaying of notation due to differences in character type and character type, there are also differences in the distinction between the ordinary and the honorific. Furthermore, even fixed phrases such as words expressing the present and past tense and negative sentences are not always used in a completely consistent state. That is,
Fixed phrases and other fixed phrases are used every time a sentence is used. Here, the term “conjugation” refers to a broad concept including not only the original conjugation form but also the addition of auxiliary verbs and the like. For example, if it is a polite body or ordinary body,
Registering fixed phrases in all cases, such as "is,""is," and "is," places a heavy burden on users, and it is open to users unnecessarily. There is a problem that the used free space of the fixed form registration is used, which is inefficient. This applies not only to a case where the source language is not limited to Japanese but also to a foreign language such as English, and there has been a problem that processing as a fixed phrase is not performed because a part of the language has changed.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problem, and is intended to be applied to a case where a user registers a fixed form sentence and utilizes a usable part of the fixed form sentence to transform the form. Is also considered to be the same as the original fixed sentence, provides a machine translation device that can expand the frame to which the fixed sentence fits during morphological analysis, obtain a high-accuracy translation result of the target language, and easily perform post-editing It is intended to be.

[0006]

In order to achieve this object, a machine translation apparatus according to a first aspect of the present invention comprises: an input means for inputting a source language sentence; and a source language input by the input means. Source language sentence storage means for storing sentences, headwords in the source language, part of speech of the headwords, word information including target language translations corresponding to the headwords, and connection information necessary for morphological analysis processing A translation dictionary storing grammatical information including: a fixed phrase sentence which is a sequence of arbitrary phrases in the source language; and a fixed translation sentence which is a sequence of translated words in the target language corresponding to the fixed phrase. A fixed sentence data storage unit storing a plurality of fixed sentence data consisting of pairs of the following, and using the translation dictionary and the fixed sentence data storage unit, the headword and the fixed headword for the source language sentence Morphological analysis that performs morphological analysis by searching A translation unit for translating the source language sentence into a target language sentence, comprising: a stage; and a syntax analysis unit that performs a syntax analysis based on a result of the morphological analysis of the source language sentence given by the morphological analysis unit. Display means for displaying the translated target language sentence, wherein the fixed form data storage means adds and registers new fixed form data or is already registered It has a fixed phrase editing means for correcting and registering the fixed phrase data,
The above-mentioned fixed phrase editing means, when registering the fixed phrase data which is a pair of the newly added or modified fixed phrase and the fixed translation, includes a portion which can be used by the fixed phrase. In addition, a utilization part designation unit for arbitrarily designating the part, a part to be used for the standard heading to be registered, and a part for utilization from the standard heading are excluded. And a fixed sentence dividing unit for dividing the unused portion into a new fixed entry sentence.

[0007] According to the machine translation device having this configuration, since the fixed phrase includes a part to be used, even if the fixed phrase is deformed and becomes different from the original fixed sentence, a part which is not used is newly viewed. By making a sentence, the transformed fixed phrase is detected, and the part that is not used is replaced with a fixed translation as a fixed phrase,
By processing the utilization part in association with this, the processing range of the fixed phrase can be expanded, high-quality translation can be easily obtained, and post-editing can be performed with a slight modification of the fixed phrase. In particular, since the user can specify the parts that he or she wants to use and the parts that he or she does not want to use, the user can decide whether to handle the transformed fixed phrase as a fixed phrase or a non-fixed phrase, as desired by the user. Can be.

[0008] A computer-readable recording medium recording the machine translation processing program according to claim 2 stores a machine translation processing program for machine translation processing for translating a source language sentence into a target language sentence. A recording medium, comprising: inputting a source language sentence to a computer; storing the source language sentence input by the input means; storing a source language sentence; A translation dictionary that stores the part of speech of the headword, word information including a translation of the target language corresponding to the headword, and grammar information including connection information necessary for morphological analysis processing; And a fixed phrase data storage unit that stores a plurality of fixed phrase data, each of which is a pair of a fixed phrase sentence that is a sequence of words and a fixed phrase translation that is a sequence of translated words of the target language corresponding to the fixed phrase. make use of A morphological analysis procedure for performing morphological analysis by searching for the headword and the fixed-form headword for the source language sentence, and a morphological analysis result of the source language sentence given by the morphological analysis procedure Machine translation that has a procedure for parsing based on the parsing based on the translation procedure for translating the source language sentence into a target language sentence, and a display procedure for displaying the translated target language sentence A processing program, in the procedure for storing the fixed form data, adding and registering new fixed form data, and correcting and registering the fixed form data already registered; In the fixed phrase editing procedure, when registering fixed phrase data that is a pair of the newly added fixed phrase and the fixed phrase translation, the fixed phrase includes a part that can be used. In addition, a procedure for specifying a part to be used to arbitrarily specify the part, a part to be used to register the standard entry to be registered in the procedure for specifying the part to be used, and a sentence ending to be used from the standard entry A machine translation program is recorded in a computer-readable manner so as to divide into a part not to be used except for the part and to execute a fixed sentence division procedure in which the part not to be used is set as a new fixed heading.

[0009] According to the computer-readable recording medium storing the machine translation processing program according to this configuration, since the fixed sentence includes a part to be used, the fixed sentence is deformed and differs from the original fixed sentence. Even if it becomes, by making the part that is not used a new fixed heading,
Detects deformed fixed phrases and replaces unused parts with fixed translations as fixed phrases, and executes the associated parts in association with the fixed phrase, thereby expanding the processing range of fixed phrases and providing high-quality translation. Can be easily obtained, and post-editing can be performed by machine translation processing that requires only a slight modification of the fixed phrase.
In particular, because the procedure is such that the user can specify the parts that he or she wants to utilize and the parts that he or she does not want to utilize, the user can decide whether to treat the transformed fixed sentence as a fixed sentence or a non-fixed sentence as desired. Obtainable.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment embodying a machine translation apparatus according to the present invention will be described below with reference to the drawings.
In the present embodiment, a case where translation from Japanese (source language) to English (target language) will be described as an example.

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 includes an input unit 2 including a keyboard or the like for inputting a Japanese original to be translated, various operation instructions, and the like. Dictionary 12 and fixed phrase data storage unit 1 for storing fixed phrase data
And an external storage device 4 including a hard disk, an optical disk, or the like and a drive thereof, and a display unit 6 including a CRT or the like for displaying an input Japanese original and an English translation as a translation result. A display control unit 7 for controlling data to be displayed on the display unit 6 and a printing unit 8 such as a laser printer for printing and outputting an English translation as a translation result through an input / output port 24. Connected to the control unit 10. The control unit 10 includes the CPU 1
8, a RAM 22, and a ROM 20 are connected via a pass line 26, and are also input from the input unit 2 and the external translation device 4 to the pass line 26 via an input / output port 24, and perform a translation process based on the data. Then, control is performed such that the translation result is displayed on the display unit 6 or output from the printing unit 8.

The input unit 2 corresponds to an input means of the present invention, and includes a character key for inputting a character (not shown), an input sentence editing key for starting various processes, a translation execution key, a fixed phrase editing key, Various function keys such as output sentence editing key, end key, print execution key, input sentence and output sentence editing,
A cursor key, an insert key, a delete key, a decision key, a next candidate selection key, and the like, which are used for screen operations and the like displayed on the display unit 6, are provided.

The external storage device 4 corresponds to a translation dictionary and fixed phrase data storage means of the present invention, and includes a translation dictionary 12 and a fixed phrase data storage unit 16 used for translation processing.
And a storage unit. The translation dictionary 12 is the translation dictionary of the present invention, and the fixed phrase data storage unit 16 is a fixed phrase data storage unit of the present invention.

The translation dictionary 1 used by the machine translation device
2, the machine translation apparatus may use not only a basic dictionary stored in an external storage device in advance, but also a specialized dictionary edited for each specialized field, a user dictionary constructed by the user independently, and the like.

FIG. 2 shows an example of the data configuration of the translation dictionary 12 described above. Each piece of dictionary registration information includes a headword 30, a part of speech 32 of each headword, a translated word 34 for each headword, and various dictionary information 36 including connection information and semantic information before and after. Consists of fields.

FIG. 3 shows the fixed form data storage section 16 described above.
2 shows an example of a data configuration stored in the. Each of the fixed sentence data information includes a fixed entry sentence 40, a utilization part 42, and a field including a fixed translation 44. Here, what the user edits from the input unit 2 is the utilization part 4
Those without 2 are the standard headings 40 as they are, and "NIL" indicating that they are not utilized is displayed in the utilization part 42. In the case where there is the utilization part 42, a combination of the standard heading 40 and the utilization part 42 is shown to the user as a standard sentence. By doing so, the user can edit without being aware that the internal data is divided.

Next, the control unit 10 includes a CPU 18, an R
It comprises a well-known microcomputer having an OM 20, a RAM 22, an input / output port 24, and a pass line 26 connecting these components. In the control unit 10, various processes are executed by a program stored in the ROM 20.

For example, a translation / editing process for translating a Japanese original sentence input from the input unit 2 into English, editing the original sentence and the translated sentence, and constructing a fixed sentence data used at the time of translation. For example, a fixed phrase editing process is executed.

The ROM 20 stores various types of information used in a translation process described below as follows.

The control program 202 includes input / output, printing,
This program controls each program and controls the entire device.

The translation program 204 is a program for translating a source language sentence input from the input unit 2 into a target language sentence, and corresponds to the translation means of the present invention. The translation program 204 includes a conversion program 210 and a generation program 212 in addition to a morphological analysis program 206 which is a morphological analysis unit of the present invention and a syntax analysis program 208 which is a syntax analysis unit.

The morphological analysis program 206 stores the fixed phrase data stored in the fixed phrase data storage unit 16 and the translation dictionary 1
While referring to the word information in 2, the source language sentence is divided into words that are morphemes, and at the same time, connection information and the like are held in association with the words, and fixed-form translations are added to the morphological analysis results.

The syntax analysis program 208 receives the result of the morphological analysis and grammatically analyzes the relationship between the morphemes. The conversion program 212 converts words in the source language sentence into words in the target language sentence. The generation program 212
A translated sentence is generated from the converted word based on the grammar of the target language.

The fixed phrase editing program 214 corresponds to the fixed phrase editing means of the present invention, and is a program for adding or modifying data stored in the fixed phrase data storage unit 16. The fixed phrase editing program 214 includes a utilization part designation program 218 and a fixed phrase segmentation program 2
16 are included. The utilization part designation program 218 corresponds to utilization part designation means of the present invention, and designates a utilization part of a fixed phrase.

The fixed phrase segmentation program 216, when a portion permitted to be used in the fixed phrase is designated by the utilization portion designation program 218, removes the portion designated to be utilized and the designated portion from the fixed phrase. It is divided into parts that are not utilized.

The RAM 22 has a source language sentence storage area 22
4. A translation storage area 226 and a work area 222 are secured. This source language sentence storage area 224 corresponds to the source language sentence storage means of the present invention.

The source language sentence storage area 224 stores the source language sentence input from the input unit 2. The work area 222 is a storage area in which a processing target for translation work, fixed sentence data being edited, and the like are temporarily captured and subjected to translation / edit processing. The translated sentence storage area stores translated and edited English translated sentences. The fixed phrase data edited and added in the work area 222 is stored and stored in the fixed phrase data storage unit 16 together with the fixed phrase sentence and the fixed phrase translation corresponding thereto.

Next, the translation / editing process executed by the control unit 10 will be described with reference to the flowcharts shown in FIGS.

As shown in FIG. 4, when the translation / edit processing is executed, first, in step 110 (hereinafter abbreviated as S110; other steps are the same), the user operates the character keys of the input section 2 by the user. It is operated and waits for the input of the Japanese original to be translated. Here, the input of the original text may be configured to read the file of the original text that has already been created. Then, when a Japanese original sentence (hereinafter, input sentence) is input, the process proceeds to S120, and the input sentence is
It is displayed in a predetermined area of the display unit 6.

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 is not pressed (S13
0: NO), it is determined whether or not the translation execution key provided on the input unit 2 has been operated (S150), and if not (S150: NO), the process proceeds to S190. Also,
If the translation execution key is operated (S150: YE
S), executing a translation process for translating a Japanese input sentence into an English translation (hereinafter referred to as an output sentence) (S160), and then pairing the output sentence obtained by the translation process with the input sentence The input sentence and the output sentence are displayed in a predetermined area provided in the display unit 6 while being stored in the RAM 22 (S170) (S180).

The translation process may be performed on all input sentences at once, or may be performed on a specified range of sentences by designating a range in units of sentences.

Next, if the translation execution key has not been operated (S150: NO), it is determined whether or not the fixed phrase editing key provided on the input unit 2 has been operated (S190). S190: NO), and proceeds to S210. If the dictionary edit key is operated (S19)
0: YES), execute the fixed phrase editing process (S20)
0). When the fixed phrase editing process is executed, the user can correct or delete the content of the information registered in the fixed phrase data storage unit 16 or register a new required fixed phrase. Become. The fixed phrase editing program that executes the procedure of S200 corresponds to a fixed phrase editing unit of the present invention.

If the standard sentence edit key is not operated (S190: NO), it is determined in S210 whether or not the output sentence edit key provided on the input unit 2 is operated, and if not, the operation is not performed (S210). : NO), and proceeds to S230. If the output sentence edit key is operated (S2
10: YES), execute output sentence editing processing (S22)
0). Note that, when the output sentence editing process is executed, the user executes the translation process in step S160 to convert the output sentence (English translation) displayed in a predetermined area of the display unit 6 into the input sentence in S140. As with the editing process, the input unit 2
Can be directly corrected using a cursor key, a delete key, an insert key, a character key, etc.

If the output sentence edit key has not been operated (S210: NO), it is determined in S230 whether the end key provided on the input unit 2 has been operated, and if not, it has been operated (S230: NO). For example, other key operations such as a print execution key for outputting the translated output sentence to the printing unit 8 for printing are detected, and a process corresponding to the operated key is executed (S240).

Then, S140, S180, S20
After the processing of 0, S220, and S240, the process returns to S130, and the same processing is repeatedly executed.

If it is determined in S230 that the end key has been operated (S230: YES), the process is terminated.

As described above, in the translation / editing process, after the Japanese original sentence is input, the input sentence editing process, translation process, fixed phrase 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 process is executed by operating the translation execution key, if the obtained output sentence (English translation) is incorrect or unnatural, the input sentence edit key is operated. After executing the input sentence editing process and re-editing the Japanese original sentence that is the input sentence, or operating the fixed phrase edit key to edit the fixed sentence data, operating the translation execution key again The output sentence editing process is executed by causing a translation or operating the output sentence edit key, and the English translated sentence which is the output sentence is directly edited.

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 the RAM 22 or the external storage device 4 as text data. And then operate the end key to end the process.

Next, S2 in the flowchart shown in FIG.
00, the user sets the fixed phrase data storage unit 1
6 will be described with reference to the flowchart shown in FIG. 5 for correcting or deleting the content of No. 6 and for registering a new required fixed phrase.

First, in S510, it is determined whether or not the user has input a standard heading to be edited. If it has been input (S510: YES), it is determined whether or not the input fixed phrase entry has already been registered in the fixed phrase data storage unit 16 (S520). If already registered (S
520: YES), re-edit the fixed phrase sentence or the fixed phrase translation in the fixed phrase data storage unit 16 (S540), and return to S5.
Move to 50. If not registered yet (S52
0: NO), and newly register a fixed entry and a fixed translation (S:
530), it is determined whether or not the registered fixed phrase has a part to be used and the user has instructed the part (S550). If there is no instruction to determine that there is no part to utilize (S550: NO), the process proceeds to S590.

If the user thinks that it can be used,
The part considered to be utilized for the text being edited displayed on the screen is designated using a key input or a mouse or the like (S550: YES), and the utilized part designated by the user is designated by the same processing as morphological analysis. The extracted character string is analyzed to extract a utilized part (S560), and the part of speech of the extracted utilized part is estimated (S570). Specifically, if it ends with "no", it is negated "no", if it ends with "Please", it is an auxiliary verb, if it ends with "i", it is an adjective, "da" or "is" or "is" If it ends with "", the adjective verb is applied, and if it ends with "U", the estimating rule such as verb is applied. After estimating the part-of-speech, the part that is not used is extracted from the fixed headings input by the user, the parts that are not used are extracted, and the parts that are not used are made into new fixed headings, and immediately after that, only the used parts are searched. Such connection information is divided into a standard entry and a utilization part (S580), and S590
Move to. Note that the procedures of S560, S570, and S580 are executed by the utilization part designation program 218, which is the utilization part designation means of the present invention.

The fixed form sentence in which the use portion is not specified and the fixed form sentence in which the use portion is specified and the processing is completed are registered (S590). At the beginning of the fixed phrase entry, connection information that allows connection only after a punctuation mark, a colon, a semicolon, a line feed code, etc., which is a sentence delimiter, is added, and a fixed phrase entry that does not utilize the end of the sentence (S550: In the case of (NO), the end of the sentence is used at the end of the sentence (S550: Y)
ES) is a punctuation mark that is used as a sentence delimiter,
In addition to sentence delimiters such as colons, semicolons, and line feed codes, connection information that allows connection of polite auxiliary words "is" and "masu" is added, and this information is registered in the fixed phrase data storage unit 16 (S590). Prepares for editing of the next headword (S5
10). These pieces of connection information are for preventing a search in the middle of the input sentence during the execution of the morphological analysis.

Next, it is determined whether or not a new headword and part of speech have been input (S510). If no headword and part of speech have been input (S510: NO), this processing is terminated.

Next, S1 in the flowchart shown in FIG.
FIG. 6 shows a translation process executed in step S60 for translating an input sentence (Japanese original sentence) into an output sentence (English translated sentence).
This will be described according to the flowchart shown in FIG.

First, a morphological analysis process is performed, and while referring to the judgment rules stored in advance in the morphological analysis program 206, the grammatical information stored in the translation dictionary 12, and the connection information stored in the translation dictionary 12, The fixed phrase data stored in the fixed phrase data storage unit 16 and word information in the translation dictionary 12 are searched. If a fixed entry can be searched, a fixed translation is added to the morphological analysis result, and if a fixed word in the translation dictionary 12 can be searched, search information such as a translated word or part of speech is added to the morphological analysis result. Execute morphological analysis processing (S
250). When a user dictionary and a specialized dictionary are used, priority is given to the user dictionary, the specialized dictionary, and the basic dictionary in this order. The procedure of S250 is executed by the morphological analysis program 206 which is the morphological analysis means of the present invention.

Next, the result of the morphological analysis processing in S250 is received, and the syntax analysis program 20 in the ROM 20 is received.
8 using a predetermined parsing grammar stored in
An analysis process for analyzing the structure of the Japanese language is performed (S26).
0), upon completion, it is detected whether the input sentence is a fixed entry sentence (S400). If it is a fixed entry (S
400: YES), it is detected whether or not there is a designated utilization part (S410). If there is a designated conjugation part in the fixed phrase entry (S410: YES), the auxiliary verb closed behind the conjugation part of the fixed phrase entry is compared with the result of the morphological analysis actually performed. Preparations for replacement are made by deleting solutions of auxiliary verbs such as "is" and "mas" (S43)
0), a replacement process is performed for the fixed form translation (S420).

In the case of a fixed entry sentence in which the utilization part is not specified (S410: NO), the fixed entry is replaced with a fixed translation as it is (S420), and the translation processing of the input sentence is terminated. The procedure of S420 is the fixed phrase replacement program 2
20. If the input sentence is not a fixed entry sentence (S400: NO), the analysis process is continued to perform other translation processing as in the case of a normal input sentence not using a fixed sentence (S265).

Subsequent to the analysis process, the Japanese structure analyzed by the analysis process is converted into an English structure by using a predetermined conversion grammar stored in the conversion program 210 of the ROM 20, and the individual structures are also converted. A conversion process for replacing words with English is performed (S270). After the conversion process, R
Using a predetermined generation grammar stored in the generation program 212 of the OM 20, an English word order is determined from the English structure converted by the conversion process, an article specific to English is generated, and an input sentence is generated. , An essential sentence is added to the output sentence even if it does not exist, and a process of generating an output sentence composed of an English word string is executed by processing a tense, an aspect, a person, a numerical feature, and the like (S280). . Subsequent to the generation process, the plural words of the noun, the past of the verb, the past participle, etc. are applied to the English word string generated by the generation process of S280 by using a predetermined morpheme generation rule stored in the ROM 20. Then, a morpheme generation process for obtaining an output sentence by generating an inflected form of is performed (S290), and the translation process ends.

Here, the processing performed in the above-described determination processing will be described with reference to example sentences.

For example, as shown in FIG. 3, as an example of always using the same expression without using the parts to be used, "Dear sir, I am grateful to your company for being more and more prosperous."
Is registered as an example.

If the Japanese sentence is not registered as a fixed head sentence, the translation is performed literally in accordance with the principle of machine translation, so that "It is glad that"
Dear, your company are mo
The English sentence "re well and prospering" translates very unnaturally.
This is called "Dear Sirs," according to English habits.
Suppose that it is registered in the fixed form data so that

In step S510 of the standard sentence editing process S200, "Dear sir, I am grateful to you for your continued pleasure" is input as a standard heading. This is assumed to be unregistered (S520: NO), and here, the headword translation "D
ear Sirs, "is input (S530), and S5
At 50, it is determined whether or not a designated portion is designated. in this case,
Since there is no fluctuation in the expression such as “I say,” the fixed headings may be used as they are. Therefore S550:
No, and the process directly proceeds to the registration procedure (S590). In step S590, the standard sentence field 40 and the standard translation field 44 of the standard text data storage unit 16 are registered.
Return to 510. If the next fixed entry is not input (S510: NO), the fixed sentence editing process ends.

In this state, a translation process S160 is executed by the user from the input unit 2 for the input sentence "Dear sir, I am grateful to you for your continued pleasure."

First, in step S250, a morphological analysis process is performed to search for a fixed-form headword, and the morphological analysis result “Dear Sirs, I am delighted that your company is becoming more and more prosperous <standard-form headline> /. , S260, and S400. In S400, the standard entry is searched (S4
00: YES), and proceeds to S410. In S410, since there is no useful part, that is, the end of the sentence is not used (S410: NO), in S420, the sentence is replaced as it is with the fixed form translation, and the translation process for this input sentence ends.

Further, as shown in FIG. 3, as an example including a part to be used, a case where "Nimi-da" is registered will be described. This can be an expression such as "I am Niimi" or "I am Niimi" as the expression of the input sentence in the text to be translated. At the same time, here is a sentence that can be translated by machine translation, but it saves the trouble of inputting "This is the new beauty of Brother Industries" and the desired English sentence "Th
is is NIIMI, Brother Indu
stories, Ltd. Take the example of registration in order to obtain ".

S5 of the fixed phrase editing process S200 shown in FIG.
At 10, "Nimi-da" is input as a standard entry. It is assumed that this is not registered (S520: NO), and then the headword translation “This is NIIMI, Br”
other Industries, Ltd. Is input (S530). Here, the user designates a part to be used (S550).
There is a fluctuation of the expression such as "I am there," so the user indicates the utilization part "da" (S550: Y
ES), this part is extracted (S560), "da" is an auxiliary verb, and the adjective verb type is used, and the auxiliary verb "is"
It is presumed that it can be replaced with (S570), the fixed heading is changed to "Nimi", and the utilized part is set to "Da" (S58).
0), "Niimi" is added to the fixed entry field 40, "Da"
Is registered in the utilization part field 42 and the English sentence is registered in the fixed translation field 44 (S590), and the process returns to S510. If the next fixed heading is not input (S51)
0: NO), and ends the fixed phrase editing process.

In this state, the input sentence "Nimi is." Is input from the input unit 2 by the user, and the translation process is executed (S160).

First, in the morphological analysis processing, the fixed head sentence is searched, and the result of the morphological analysis is obtained as “Niimi <fixed head sentence> / is <auxiliary verb> /. <Phrase point>” (S250), S26
0, and proceed to S400. In S400, since the fixed entry is searched (S400: YES), the process proceeds to S410. In S410, since the use part “da” exists, that is, the end of the sentence is used (S410: YE
S), proceed to.

The result of the morphological analysis, “da” and the utilized part “da” are equivalent, and “da” is deleted as an unnecessary solution (S430), and is replaced with a fixed form translation (S42).
0), the translation process for this input sentence ends.

Note that the present invention is not limited to the above-described embodiment, and various changes can be made.
For example, although the present embodiment has been described by taking as an example the case of translating Japanese into English, translation processing in a series of flows of morphological analysis-syntax analysis-conversion-syntax generation-morpheme generation in machine translation is performed in other languages. Is also the same, and the handling of fixed phrases is not limited to any language. Therefore, it is needless to say that the present invention is applicable to a translation process between different languages.

In the present embodiment, the case where the inflected part is "da" and "is" at the end of the sentence is taken as an example. However, in the present invention, the inflected part means not only the change of the ending part of the original word but also the change. It is a broad concept including the addition of auxiliary verbs. In the example of the present embodiment, Japanese is used as an example of the source language sentence. Therefore, it is common that the part utilizing Japanese is at the end of the sentence. For example, in the case of English, the verb "is" is replaced with "isn" A configuration that allows utilization in cases where “t” or “do” becomes “will do” may be used. In this case, since the utilized portion is in the sentence, the unutilized portion and the utilized portion divided by the utilized portion are each provided with connection information, so that the divided portions are integrally handled, thereby implementing the present embodiment. It is easily understood that the same processing as that of the embodiment can be performed.

Further, the present invention can be applied to a sentence including a plurality of utilized portions by performing the same processing.

In the present embodiment, the analogy of the part of speech of the utilized part and the fluctuation of the possible expression are determined in S43 inside the system.
Although it was decided to be 0, it may be set in a menu of the user interface. For example, if the entry is in the verb imperative form, it is equivalent to "Please". Boilerplate sentences are a rescue for machine translation systems where perfect translation is not possible, but at the same time it is inconvenient if there is too much. If this can be finely controlled by the user, the freedom of the user can be expanded, and the translation accuracy can be improved and post-editing can be reduced.

Further, the machine translation apparatus of this embodiment includes a dictionary editing program, a morphological analysis program, an analysis program, a part of speech determination program, a headword division program, a determination program, a before / after determination program, The word recovery program is stored in the ROM in advance, but the present invention is not necessarily limited to this. For example, a program in which these programs are stored in a floppy disk, CD-ROM, or the like can be read by a reading device, installed, and operated. In addition, a program can be read and operated from an external information processing device using a wired or wireless line. In this case, the floppy disk or CD-R
The OM and the memory of the external information processing device that stores the program constitute the recording medium of the present invention.

[0065]

As is apparent from the above description, according to the machine translation apparatus according to the first aspect of the present invention, since the fixed sentence includes a part to be used, the fixed sentence is transformed, and the input sentence is usually changed. Even if it becomes different from the original fixed sentence that is difficult with machine translation, it can detect the deformed fixed sentence by making the unused part a new fixed heading sentence, Replace it with a fixed translation as a fixed sentence,
By processing in conjunction with this, the processing range of fixed phrases can be expanded, and high-quality translations can be easily obtained. There is an effect that processing can be performed in time.

In particular, the utilization part designation means allows the user to designate which part is to be utilized when the end of the sentence is utilized, and the registration means is provided for utilizing the fixed entry sentence when there is a utilized part. The part not to be used can be registered as the standard sentence data while the part to be utilized is divided as the utilization part and the part to be utilized is divided as the utilization part.

In particular, since the user can designate a part to be used and a part not to be used, the user can specify whether the modified fixed sentence is to be treated as a fixed sentence or a non-fixed sentence. Is obtained. In other words, when registering a fixed phrase, by specifying the part to be used, for example,
The user can specify the range that can be recognized as a fixed sentence, such as allowing the change of the normal state but not further change, and the user has changed as desired, such as accepting a certain change, but not accepting any further change There is an effect that it is possible to handle fixed phrases. As a result, there is an effect that a translation result of the target language with high accuracy can be obtained with less labor.

Further, since the transformed fixed sentence is directly replaced with the corresponding fixed-form translated sentence without performing morphological analysis, the processing can be speeded up. Further, according to the computer-readable recording medium on which the machine translation processing program is recorded, since the fixed sentence includes a part to be used, the fixed sentence is deformed, and the input sentence is converted to a normal machine. Even if it is different from the original fixed phrase that is difficult to translate, the part that is not used can be newly detected as a fixed phrase, so that the transformed fixed phrase can be detected. Can be replaced with a fixed form translation, and the utilized part can be executed in association with this. Therefore, the processing range of fixed phrases can be expanded, and high-quality translations can be easily obtained. Post-editing can be performed with only minor corrections of fixed-line translations, making it extremely easy to process high-quality translations in a short time. This has the effect.

In particular, in the procedure for specifying the part to be used, when the end of the sentence is used, the user can specify which part is to be used. The part that does not utilize is a fixed heading,
The part to be used can be registered as the part to be used in the fixed form data. As a result, when registering a fixed phrase, by specifying the part to be utilized, the frame to which the fixed phrase is adapted during morphological analysis is expanded, and the translation result of the target language with higher accuracy can be reduced with less effort. There is an effect that can be obtained.

Further, since the transformed fixed sentence is directly replaced with the corresponding fixed translation without performing morphological analysis, there is an effect that the computer can execute high-speed processing.

[Brief description of the drawings]

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

FIG. 2 is an explanatory diagram of a data structure of a dictionary according to the embodiment.

FIG. 3 is an explanatory diagram of a data configuration of fixed phrase data according to the embodiment;

FIG. 4 is a flowchart illustrating an operation of the machine translation device of the present embodiment.

FIG. 5 is a flowchart illustrating a flow of a fixed sentence editing process of the machine translator according to the embodiment.

FIG. 6 is a flowchart illustrating a flow of a translation process of the machine translation device of the present embodiment.

[Explanation of symbols]

 Reference Signs List 2 input unit (input means) 4 external storage device 6 display unit (display means) 7 display control unit (display means) 10 control unit 12 translation dictionary (translation dictionary) 16 fixed sentence data storage unit (fixed sentence data storage means) 18 CPU 20 ROM 22 RAM 202 control program 204 translation program (translation means) 206 morphological analysis program (morphological analysis means) 208 syntax analysis program (syntax analysis means) 210 conversion program 212 generation program 214 fixed form editing program (fixed form editing means) 216 Fixed sentence segmentation program (fixed sentence segmentation means) 218 Utilization part designation program (utilization segment designation means) 220 Fixed sentence replacement program 222 Work area 224 Source language sentence storage area (source language sentence storage means) 226 Translated sentence storage area

Claims (2)

[Claims] An input unit for inputting a source language sentence; a source language sentence storage unit for storing the source language sentence input by the input unit; a headword in the source language; a part of speech of the headword; A translation dictionary storing word information including a translation of the target language corresponding to the headword and grammatical information including connection information necessary for morphological analysis processing; and a fixed form which is an array of arbitrary phrases in the source language. A fixed sentence data storage unit storing a plurality of fixed sentence data each comprising a pair of an entry sentence and a fixed translation that is a sequence of translated phrases of the target language corresponding to the fixed entry, the translation dictionary and the fixed sentence A morphological analysis unit that performs a morphological analysis by searching the headword and the fixed headword for the source language sentence using a data storage unit;
Syntactic analysis means for performing syntax analysis based on the result of the morphological analysis of the source language sentence given by the morphological analysis means, translation means for translating the source language sentence into a target language sentence, Display means for displaying the target language sentence, wherein the fixed form data storage means adds and registers new fixed form data, or the fixed form data already registered Has fixed form editing means for correcting and registering the fixed form sentence editing means, when registering the fixed form sentence data consisting of a pair of the newly added or corrected fixed form head sentence and the fixed form translation sentence When the standard heading includes a part that can be utilized, a utilization part designating unit arbitrarily designating the part, and utilizing the standard heading to be registered designated by the utilization part designating unit Part and this It is divided into non-utilized parts excluding the parts to be utilized from the fixed headings,
A machine translation unit comprising: a fixed sentence dividing unit that uses the unused portion as a new fixed entry. 2. A recording medium storing a machine translation processing program for machine translation processing for translating a source language sentence into a target language sentence, comprising: A source language sentence storage procedure for storing the source language sentence input by the input means, and word information including a headword in the source language, a part of speech of the headword, and a target language translation corresponding to the headword. A translation dictionary storing grammatical information including connection information necessary for morphological analysis processing, a fixed-form heading which is an array of arbitrary phrases in the source language, and an object corresponding to the fixed-form heading Searching for the headword and the fixed head sentence for the source language sentence using fixed sentence data storage means storing a plurality of fixed sentence data which are pairs of fixed sentence data which are a sequence of language translation phrases Form that performs morphological analysis A step of parsing, and a step of parsing which performs parsing based on a result of the morphological analysis of the source language sentence given by the step of morphological analysis, translating the source language sentence into a target language sentence And a display procedure for displaying the translated target language sentence, wherein in the fixed form data storage procedure, new fixed form data is added and registered. And a fixed sentence editing procedure for correcting and registering already registered fixed sentence data; and in the fixed sentence editing procedure, the newly added or corrected fixed entry sentence and the fixed translation When registering a set of fixed phrase data, if the fixed phrase includes a usable portion, a procedure for specifying a part to be used to arbitrarily specify the portion; ,Previous A standard sentence segmentation that is divided into a part to be used designated by the procedure for specifying a part to be used and a part not to be used by removing the part to be used from the fixed heading, and the part not to be used to be a new fixed heading. And a computer-readable recording medium on which a machine translation processing program for executing the above steps is recorded.