JP3405491B2 - Machine translation equipment - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation device,
Particularly, it relates to a display function of the processing content.

[0002]

2. Description of the Related Art When translating an input original sentence with a machine translation device, in addition to using a basic language dictionary, a user registration dictionary, a technical term dictionary, a learning dictionary, etc. are used together in order to perform more appropriate translation. And translate it.

Of these, the dictionaries that the user registers using his / her own knowledge are the user registration dictionaries and the learning dictionaries. Generally, the user registration dictionaries are unregistered words or basic words not registered in the basic word dictionary. The word that should be given priority over the dictionary is registered, and the user himself / herself registers the information necessary for the translation intended by the user, so the registered content is clear to the user. Therefore, it is relatively easy to determine whether to use a certain user-registered dictionary for translation.

On the other hand, the learning dictionary stores the dictionary information used when translating a certain sentence and preferentially uses the learned dictionary information for the subsequent translation of the sentence, or
This is for preferentially outputting the translated word selected by the user in advance (see Japanese Patent Laid-Open No. 61-260367), and the learning dictionary holds the information of the basic word dictionary for the learned word. ing. In other words, the learning dictionary is more complete than the user-registered dictionary as dictionary information used for machine translation.

[0005]

However, the conventional machine translation system has the following problems.

(1) There has been no means for confirming the entire contents of the learning dictionary by listing the contents of the learning dictionary, so that the user can reuse the learning dictionary that he has made when he reuses the learning dictionary. Could not decide whether to use. For this reason, the person who created the learning dictionary had no choice but to judge the contents of the learning dictionary by inferring it from the memory at the time of learning or the name given to the learning dictionary.

If the contents of the learning dictionary cannot be guessed, it is necessary to give up the reuse of the learning dictionary and perform a wasteful work of creating a new learning dictionary even if the contents of learning are the same. did not become.

[0008] Since the learning dictionary stores dictionary information in the basic word dictionary of the translated words used at the time of translation, the learning dictionary is such that the user himself / herself changes the priority of the translated words in the basic word dictionary with abundant dictionary information. Although it is a means that can translate a desired translated word, there is a problem that the contents of the learning dictionary are not well understood. Therefore, the availability of the learning dictionary cannot be determined, and the user cannot effectively utilize the learning dictionary. Therefore, it is necessary to easily confirm the contents of the learning dictionary in order to determine whether or not the learning dictionary can be used.

(2) Both the user-registered dictionary and the learning dictionary are the same in which the user's knowledge is registered and used for translation, but since the character of the dictionary is different as described above, higher quality translation is possible. In order to obtain results, it is important to make effective use of the characteristics of each dictionary.

Since the user-registered dictionary is created mainly for the purpose of supplementing the words lacking in the basic word dictionary, it is easy for the user to understand what kind of document should be used for translation, and effective re-reading is possible. Although it can be used, if the learning dictionary is a dictionary created by another person, it is difficult to know what kind of document is effectively used by just looking at the contents of learning. Therefore, in order for a plurality of users to effectively reuse the learning dictionary, it is necessary to have a criterion for judging what document a certain learning dictionary should be used for. Moreover, it is necessary to have a criterion for properly using a plurality of learning dictionaries.

(3) Further, the learning dictionary cancels learning when the content of learning a certain translated word is inappropriate,
The learning once performed can be invalidated by learning another translation. However, because there was no way to know what kind of intention the learning that seems to be inappropriate was made, it was possible to cancel the learning that seems to be inappropriate by the user's sensory judgment or to learn another translated word. By doing so, it was happening that learning that was effective and versatile as a whole was inadvertently canceled. Therefore, for any learning dictionary, there is a need for a criterion for judging the validity of learning for a certain translated word.

(4) Whether or not a translated word once learned is general and appropriate learning in a certain type of document depends on the translation result when another sentence in which the learned word appears is translated. You may not know if you do not know whether or not. In particular, it is necessary to cancel or change the content once learned when another sentence is translated and the result is not the desired translation, and when changing the content of learning, the validity of learning is changed. Need to judge. However, it is cumbersome for the user to search the learning dictionary for information to judge the validity of a word that needs to be changed in learning, and it is immediately necessary to search for such a specific word. It is desirable to be able to confirm the intention of learning.

(5) The reason why the learning dictionary is not effectively reused is that the learning content is not very clear as described above, and there is a problem of the range of application of learning.

When performing learning, the user performs learning that he or she wants to reflect in subsequent translations in consideration of the influence of certain learning on the translation of other sentences, the versatility of the learning content, etc. If you do not want to influence it, you should avoid learning.

However, in practice, even if the translated word to be learned is not a general-purpose translated word, there are cases where learning is performed.

For example, since a machine translation system is generally provided with a means for easily changing a translated word by selecting from several candidates, the user uses this means for the purpose of correcting the translation result. The translation of a sentence in a document may be changed. In this case, if the changed translated word is not so general, the original learning should not be performed. However, if the document is to be subjected to machine translation again, even if the translation word is corrected by the translation word changing means, if it has not been learned, retranslation causes an inappropriate translation word before translation to be translated again. That is the case, and it becomes necessary to correct the translated word again. In order to avoid such a situation, it is often the case that the user performs local learning that is not general purpose when changing the translation.

Such learning may be effective only for a specific document, local learning such that a specific translated word may be translated only for a specific sentence, or general-purpose learning. This may cause the translated words that have been learned with different application ranges such as simple learning to be mixed in one learning dictionary, making the application range of the learning dictionary extremely narrow and hindering the reuse of the learning dictionary. .

In order to prevent learning with different application ranges from being mixed in one learning dictionary, the translated words to be learned must be limited to those having general versatility. In order to limit the learning to general-purpose learning, the user has no choice but to correct the translated word by searching for the part that finally needs to be corrected and performing the correction by himself / herself.

Further, for example, it is assumed that a learning dictionary created for a certain field is used to translate a certain document included in the field. At this time, by adding effective learning to the document to the learning dictionary, a learning dictionary having a considerably effective and fulfilling content for the document is constructed.

However, the learning dictionary is accordingly
The original versatility is lost and it cannot be reused for translation of other documents.

From these reasons, in order to effectively reuse the learning dictionary, it is necessary to prevent the reuse of the learning dictionary due to the mixture of learnings having different learning application ranges.

The present invention has been made to solve the above problems, and enables effective reuse of a learning dictionary, and machine translation that outputs high-quality translation results by using an appropriate learning dictionary. The purpose is to get the device.

[0023]

A machine translation device of the present invention
Is an input unit for inputting a document including a sentence in the first language.
And a plurality of words in the first language, together with each word
A basic language dictionary storing dictionary information in a second language, and
According to the usage history, the dictionary information is stored as words in the first language.
Learning dictionaries consisting of various learning dictionaries for storing together
The translation dictionary section that has a writing section, and the input sentence in the first language
Translate into a sentence in a second language using the dictionary information of the translation dictionary unit
To display the translation unit and processing information in the translation unit.
And an output display unit for the translation unit, the translation unit
Of the basic word dictionary used to translate this sentence when
The dictionary information is stored in the learning dictionary section for subsequent translation.
The dictionary learning function that preferentially uses these dictionary information.
A machine translation device equipped with a predetermined learning dictionary
And associate the document name of the translated document with each learning dictionary.
Storage to have a learning dictionary history storage unit, thereby, the
The purpose is achieved .

The machine translation apparatus of the present invention converts a sentence in the first language into
An input section for inputting a document including a plurality of first language
A dictionary in a second language corresponding to each word together with the word
Basic word dictionary storing information and use of the dictionary information
For storing with words in the first language, depending on history
A translation dictionary unit having a learning dictionary unit, and the input first language
Sentence of the second language using the dictionary information of the translation dictionary section
Display the translation part to translate and the processing information in the translation part.
And an output display unit for
When you translate a sentence of a word, the basic word used to translate this sentence
The dictionary information of the dictionary is stored in the learning dictionary section, and is used for subsequent translation.
In contrast, a dictionary learning machine that preferentially uses these dictionary information
A machine translation device equipped with a function,
When the dictionary information is stored, the translated sentence in the first language
And at least one of the sentences of the second language as a learning history sentence
Te, have a university <br/>習辞document history storage unit for storing the said learning dictionary section in association with the dictionary specification information, thereby, the object is achieved
To be done .

It is stored by the learning dictionary history storage unit.
Display learning history sentences for a specific translated word during translation
You may have a learning dictionary display control part.

The machine translation apparatus of the present invention converts a sentence in the first language into
An input section for inputting a document including a plurality of first language
A dictionary in a second language corresponding to each word together with the word
Basic word dictionary storing information and use of the dictionary information
For storing with words in the first language, depending on history
A translation dictionary unit having a learning dictionary unit, and the input first language
Sentence of the second language using the dictionary information of the translation dictionary section
Display the translation part to translate and the processing information in the translation part.
And an output display unit for
When you translate a sentence of a word, the basic word used to translate this sentence
The dictionary information of the dictionary is stored in the learning dictionary section, and is used for subsequent translation.
In contrast, a dictionary learning machine that preferentially uses these dictionary information
A machine translation device equipped with a function,
When memorizing dictionary information, normal learning and specific document
Store the dictionary information separately from learning that is effective only
It has a learning dictionary registration unit, which is used when translating a specific document.
Dictionary information stored as learning that is valid only for certain documents
Information and dictionary information stored as the normal learning are given priority.
Used for translation, and when translating other than the specific document, the
Preferentially using only the dictionary information stored for learning
Translate, which achieves the above objectives.

The machine translation apparatus of the present invention converts a sentence in the first language into
An input section for inputting a document including a plurality of first language
A dictionary in a second language corresponding to each word together with the word
Basic word dictionary storing information and use of the dictionary information
For storing with words in the first language, depending on history
A translation dictionary unit having a learning dictionary unit, and the input first language
Sentence of the second language using the dictionary information of the translation dictionary section
Display the translation part to translate and the processing information in the translation part.
And an output display unit for
When you translate a sentence of a word, the basic word used to translate this sentence
The dictionary information of the dictionary is stored in the learning dictionary section, and is used for subsequent translation.
In contrast, a dictionary learning machine that preferentially uses these dictionary information
A machine translation device equipped with a function,
Learn the translated first language sentence when storing dictionary information
As a history sentence, in the learning storage unit in association with the dictionary information.
A learning dictionary history storage unit to be stored, and the dictionary in the learning dictionary unit.
When you remember information, you can
The dictionary information is recorded separately from learning that is valid only for a limited time.
The learning dictionary registration unit to remember and these learning seeds when translating
A learning type discriminating unit for discriminating the type of learning.
If the learning was effective only for one specific sentence
In this case, the translation unit determines that the sentence being translated is stored in the learning dictionary history storage unit.
Only if the learning history sentences stored in the
The dictionary information in the dictionary section is used preferentially for translation, and
Therefore, the above object is achieved.

[0028]

[0029]

[0030]

In the present invention , since the document name of the document translated using any learning dictionary is stored in association with each learning dictionary, the history of use of the learning dictionary can be provided to the user.

In the present invention , the original sentence and the translated sentence at the time of learning are stored in association with the learned translated word. Therefore, the learning history sentence for confirming the learning intention is provided to the user. can do.

In the present invention , since the learning history sentence for a specific word can be displayed during translation, it is possible to provide the user with information that can confirm the intention and validity of learning for a specific word during translation. it can.

[0033] In the present invention, a normal of learning, specific
Only to learn to distinguish between a valid learning to document, since so as to determine them at the time of the translation, using the learning dictionary of different learning content range of applications exists without worrying about the impact of special learning it can.

In the present invention , normal learning and specific learning
Differentiate learnings that are valid only for one sentence and translate them
Since it is decided to distinguish when learning
A learning dictionary that has contents without worrying about the effects of special learning
Available for

[0035]

[0036]

[0037]

[0038]

[0039]

Embodiments of the present invention will be described below with reference to the drawings.

(Embodiment 1) FIG. 1 is a block diagram for explaining a machine translation device according to a first embodiment of the present invention .

In the figure, 101 is a machine translation apparatus of this embodiment. This apparatus 101 includes a first language sentence to be translated, an input unit 1 for inputting various processing instructions, and an input first language sentence. A translation dictionary unit 4 used when translating a sentence in a second language into a sentence in a second language,
It has a translation unit 2 that performs a process of translating into a language and other processes necessary for translation, and the translation unit 2 is equipped with a central processing unit (CPU) for executing the process. Further, the machine translation apparatus 101 outputs the input sentence in the first language, the sentence in the second language as a translation result, the contents of the translation dictionary unit, and the like, and the output display unit 3, and the contents of the learning dictionary. And a learning dictionary display control section 5 for controlling to display on the output display section 3. Here, the translation dictionary unit 4 is composed of a basic word dictionary 6 and a learning dictionary unit 7. The learning dictionary unit 7 has a plurality of learning dictionaries 7a, 7b, 7c each having a different name.

Next, the operation will be described.

FIG. 5 is a flow chart showing the operation of performing the learning dictionary setting process A in the machine translation apparatus of this embodiment. Here, a machine translation apparatus in which the first language is English and the second language is Japanese is taken as an example. explain.

The learning dictionary setting process A includes, as its main process, a process of displaying the learning contents of all the words learned in the learning dictionary.
Will be described in detail.

First, in step A1, when an initial menu screen such as the display screen example of FIG. 12A is displayed on the output display section 3, next, in step A2, the processing items shown in the initial menu screen are displayed. The processing item to be executed is selected and input from the input unit 1.

In step A2a, the translation unit 2 determines whether or not a translation execution processing item has been selected. If the translation execution is performed, the processing B shown in FIG. 6 is performed. On the other hand, if the translation is not executed, the process proceeds to step A3. In step A3, the translation unit 2 determines whether or not the item selected in step A2 is the end of the process. If the end, the process proceeds to step A12 to end the process. CP if not finished
The process of U proceeds to step A4, and the translation unit 2 determines whether the learning dictionary selection is selected in step A2. If it is the learning dictionary selection, the process of the CPU proceeds to step A5, and if not, it proceeds to step 10.

At step A10, the item selected at step A2 is processed, and the process returns to the display of the initial menu screen at step A1.

When the item selected in step A4 is the learning dictionary selection, a list of all the learning dictionaries 7a to 7c existing in the learning dictionary unit 7 is displayed in step A5. Next, in step A6, the user selects a learning dictionary to use from the displayed list of learning dictionaries. Here, an instruction to cancel the learning dictionary selection can be given.

An example of the screen displayed in step A5 is the learning dictionary selection screen shown in FIG. 12 (b). In this screen display example, 01: BUSINESS. gakushu and 02: FOOD.
It shows that there are three learning dictionaries, gasushu and 03: DEFENCE.gakushu.
You will have to choose from one learning dictionary. When not using or when canceling the learning dictionary once selected, an instruction to cancel the selection of the learning dictionary may be given.

At step A7, the user selects whether to display the contents of the selected learning dictionary. If so, the process proceeds to step A8. If you do not display
The process proceeds to step A11, and the translation unit 2 uses the learning dictionary selected in step A6 for translation.
Is set, and the screen returns to the initial menu screen to step A1.

In step A8, the learning dictionary display control section 5 reads the learning dictionary selected in step A6 and sends the contents to be displayed to the user to the output display section 3.
As a result, the contents of the learning dictionary are displayed on the output display unit 3.

An example of the screen displayed at this time is shown in FIG.
It is a learning dictionary content display screen shown in (c).

In this display screen example, BUSINESS. Gakushu
The learning content of the learning dictionary is displayed, boo
It is displayed as the contents of the learning dictionary that words such as k, paper, and task have been learned, and the translated words that have been learned are “ledger”, “newspaper”, and “work”, respectively.

The translated words before learning each translated word are also displayed.

In this example of the display screen, the headline, the part of speech, the translated word, and the translated word before learning, which are stored in the learning dictionary, are displayed, but the number is not limited to these four, and if necessary, Other information related to learning other than these, for example, case pattern information being learned, that is, syntactic information such as case dominance relationship in a sentence at the time of learning can be displayed.

Next, in step A9, the user selects whether or not to use the learning dictionary whose learning contents are being displayed. If so, the translating unit 2 selects the learning dictionary in step A11. It is set so that it can be used for translation, and the process returns to the initial menu screen to step A1.

If the learning dictionary whose contents are being displayed is not used, the processing of the CPU returns to the processing of displaying the list of learning dictionaries at step A5, and at step A6, the user selects another learning dictionary. , Cancel the selection of learning dictionary.

As described above, in the present embodiment, the contents of the learning dictionary are displayed as a list regarding what kind of learning is done in a certain learning dictionary. Therefore, when selecting a learning dictionary to be used for translation, It is possible to present the user with criteria for determining whether or not it is appropriate to translate using a learning dictionary.

(Second Embodiment) FIG. 2 is a block diagram for explaining a machine translation device according to a second embodiment of the present invention .

In the figure, reference numeral 102 denotes a machine translation apparatus of the present embodiment, which is the same as the machine translation apparatus 101 of the first embodiment described above, but also a document name translated by using a certain learning dictionary and its learning. The learning dictionary history storage unit 8 stores, for each learning dictionary, accompanying information such as a translation environment when translation is performed using the dictionary.

Next, the operation will be described.

The operation of this embodiment will be described with reference to the flow charts of FIGS.

FIG. 6 is a flow chart showing an operation of storing the usage history of the learning dictionary in the machine translation apparatus.

This process B is step A2 of process A shown in FIG. 5, and when "01: execute translation" is selected from the selection items on the initial menu screen of FIG. 12A, the process of step A2a is executed. Executed.

First, when the translation processing screen is displayed in step B1 and an instruction for processing is input in step B2, the translation unit 2 branches the processing according to the type of instruction input (step B3).

When the end instruction is input, step B4
Then, the translation unit 2 checks whether or not any character is input. If there is no character input, the process is terminated as it is, and the process of the CPU returns to step A1 of process A.

If there is any character input, in step B5 the translation unit 2 determines whether or not the input translation document (original sentence, translated sentence) has been saved. If saved,
The translation unit 2 checks in step B6 whether or not the learning dictionary is set. If the learning dictionary is set, the process proceeds to step B7, and if not set, the process proceeds to step A1 of process A.

At step B7, the user selects whether or not to save the translated document name as the learning dictionary use history. When the translated document name is stored, the learning dictionary history storage unit 8 stores the translated document name in the learning dictionary history storage area of the learning dictionary in step B8. Next, in step B9, the user selects whether or not to save other accompanying information as a learning dictionary usage history. If not saved, the whole translation process ends, and the process of the CPU returns to the initial menu screen of step A1 of process A. If other incidental information is stored, the learning dictionary history storage unit stores the other incidental information in the learning dictionary storage area in the learning dictionary in step B10, completes the entire translation process, and the process of the CPU is step A. Return to the initial menu screen of A1.

If the translated document is not saved in step B5, the user selects whether to end without saving the translated document in step B11. When the end is selected, the translation process ends, and the process of the CPU returns to the initial menu screen of step A1 of the process A. If not finished, the end instruction is canceled and the process of the CPU performs the translation process of step B1. Return to screen display.

If there is an input for instructing to save the translated document in the processing instruction input in step B2, if the translation document name to be saved is input by the input unit 1 in step B12, the translation unit 2 in step B13 The translated document is saved with this translated document name, and the screen returns to the translation processing screen of step B1.

Further, if there is an input for instructing translation execution in the processing instruction input in step B2, step B1
When character input is performed by the input unit 1 in step 4, step B
At 15, the translation unit 2 translates the input sentence in the first language, and returns to the translation processing screen of step B1.

Further, if there is an input instructing processing other than the above-mentioned three kinds of processing such as calling a translated document in the processing instruction input in step B2, step B16
Then, the processing corresponding to the instruction input is performed in the translation unit 2,
It returns to the translation processing screen display of step B1.

In this embodiment, the translation document name to be stored and other accompanying information are stored in the learning dictionary use history area in the learning dictionary. However, the use history information of the learning dictionary is provided separately from the learning dictionary. It is also possible to adopt a method in which a mechanism for storing is stored in association with the learning dictionary.

The operation of displaying the information stored as the learning dictionary use history as described above will be described with reference to the flowchart of FIG.

This processing C is step A8 of processing A,
When the display of the contents of the learning dictionary is designated, the process starts.

In step C1, the user selects whether to display the information of the learning dictionary use history or the contents of learning. In step C2, the learning dictionary display control unit 5 determines whether or not the display content selected in step C1 is a display of the learning dictionary usage history. If it is not the display of the learning dictionary usage history, it is an instruction to display the learning content.
The process proceeds to step C4, the learning content is displayed, and the process returns to step A9.

If it is determined that the display of the learning dictionary usage history in step C2 is selected, the learning dictionary display control section 5 sends the learning dictionary usage history stored in the learning dictionary to the output display section 3 in step C3. Send, display the learning dictionary usage history as in the example of the display screen shown in FIG.
Then, the process returns to step A9.

For example, this display screen example shows BUSINESS. Gaku
This is an example of displaying the usage history of the learning dictionary called shu. This learning dictionary is used for businessweek, newsweek, and APbusi.
It is used when translating a document named ness, and the field information specified at the time of translation and the date of translation are displayed as accompanying information at the time of translation of each document.
In this display screen example, the incidental information other than the translated document name is only the field information set at the time of translation and the date of translation. Necessary information such as the user dictionary name can be stored as accompanying information.

As described above, in this embodiment, when a certain document is translated using a certain learning dictionary, the name of the translated document and other accompanying information are stored as the history of use of the learning dictionary. , By displaying, you can see not only the information about which translated word of which word is being learned in a certain learning dictionary, but also what document was used to translate that learning dictionary. The range of judgment criteria when deciding to use is expanded, and the learning dictionary can be reused more effectively.

(Embodiment 3) Next, a machine translation device according to a third embodiment of the present invention will be described.

The machine translation apparatus of this embodiment particularly corresponds to the invention of claim 3, and the learning dictionary history storage unit 8 in the machine translation apparatus of the second embodiment is used for learning. The original sentence and the translated sentence are stored as a learning history sentence in association with the learned translated sentence.

Next, the operation will be described with reference to the flowcharts of FIGS. 5, 6 and 8. Here, FIG. 8 shows the flow of the storage process of the learning history sentence.

First, steps A1 to A1 of the process A shown in FIG.
1, the learning dictionary to be used is set, translation execution is selected in steps B1 to B3 of process B shown in FIG. 6, and the first language sentence (original sentence) to be translated is input in step B14.
In step B15, translation processing is performed.

This translation process corresponds to the process D shown in FIG.

That is, when the process D shown in FIG. 8 is started, first, the process of translating into the second language is performed in step D1, and the sentence (translated sentence) of the second language which is the translation result is displayed in step D2.

Next, in step D3, the user decides whether to adopt the displayed translated text. If adopted, the process returns to the translation process screen display of step B1 of process B shown in FIG.

If the translated sentence that is the translation result is not adopted, the process proceeds to step D4, and the user specifies an inappropriately translated word in the original sentence. Next, when an instruction input for selecting another translation of the word is performed in step D5, translation word candidates selectable in step D6 are displayed, for example, in FIG.
It is displayed as in the display screen example of (b). Among the candidates that can be selected here, candidates for instructing to cancel the selection of the translated word are also prepared.

In this display screen example, with respect to the noun task, three candidates for which translations can be selected are shown: "NO.1: task", "NO.2: job", and "NO.3: job". Further, it is shown that there is an item (NO. 4) for canceling the translation word selection as a selection candidate in addition to the candidate for which the translation word can be selected.

Next, in step D7, when the user selects one of the displayed candidates, in step D8, the translation unit 2 determines whether or not the instruction to cancel the translation selection is selected. If it is determined that the instruction to cancel the translation selection is selected, the CPU processing returns to the translation result display in step D2.

If the instruction to cancel the translation selection is not selected, it means that the translation has been selected.
The user selects whether or not to learn the translated word selected in step 7 (step D9).

If learning is not to be performed, the process is step D13.
Then, the translation unit 2 replaces the translated word of the translation result with the selected translated word, and returns to the display of the translated sentence in step D2.

When learning is selected in step D9, the translation unit 2 learns the selected translation word in step D10. The learning of the translated word can be realized, for example, by adding the <learning> mark to the selected translated word and deleting the mark if the <learning> mark is added to the selected translated word.

Next, in step D11, the user selects whether or not to save the original sentence and the translated sentence selected as the translated words in association with the learned translated sentence as the learning history sentence. If not saved here, the process proceeds to step D13.
The translation unit 2 replaces the selected translation word and returns to the translated sentence display in step D2.

If the learning history sentence is saved in step D11, the process proceeds to step D12, and the learning dictionary history storage section 8 associates the original sentence with the learned word in the learning history area of the set learning dictionary. The translated sentence is stored in the translation unit 2, and the translation unit 2 replaces the selected translated word in Step D13. Then, the process returns to Step D2 where the translated sentence is displayed.

The learning history sentence stored in this way is displayed by the learning dictionary display control section 5 as in the example of the display screen of FIG. 13 (c).

For example, the translated word "work" of the noun task is "I am doing a
I learned when I translated the sentence "good task.", and as a result, the translated sentence shows that "I am doing a good job."

In this embodiment, as the learning history sentence,
Although both the original sentence and the translated sentence including the learned word are stored, either one of them may be stored as the learning history sentence.

As described above, in the present embodiment, at the time of learning, the original sentence and / or the translated sentence including the learned word can be stored as the learning history sentence. After learning, you can understand what kind of translation result is output as a result, and you can grasp the intention of learning. Therefore, the details of learning can be known in more detail, and the learning dictionary can be reused more effectively.

(Embodiment 4) FIG. 3 is a block diagram for explaining a machine translation device according to a fourth embodiment of the present invention .

In the figure, reference numeral 104 denotes a machine translation device of this embodiment, which connects the learning dictionary display control unit 5 in the third embodiment with a translation unit so that the control unit 5 allows a specific word to be translated during translation. On the other hand, the control for displaying the learning content is performed.

Next, the operation of this embodiment will be described with reference to the flowcharts of FIGS. 8 and 9.

FIG. 9 is a diagram showing the flow of the processing E for displaying the learning history sentence. This processing E is step D5 in FIG.
When the translation selection is instructed in, it starts.

First, in step E1, the translation unit 2 checks whether or not there is a learned translated word among the translated word candidates from which the translated word can be selected. If there are translations that have been learned,
The translation unit 2 clearly indicates that the word learned in step E2 is a learned word by adding a symbol indicating that the word has been learned, and outputs a selectable candidate word. In addition to the display on the display unit 3, the item for instructing the display of the learning history sentence is also displayed as a selection candidate. An example of the display screen at this time is the display screen example shown in FIG.

In this display screen example, with respect to the noun task, three candidates “NO.1: task”, “NO.2: job” and “NO.3: job” whose translations can be selected are shown. , "NO. 2: work" is learned by clearly indicating ← <learning>. Further, an item (NO. 4) for instructing the display of the learning history sentence as a selection candidate in addition to the candidate for which the translation word can be selected
And that there is an item (NO. 5) for canceling the translation selection.

Next, in step E3, the user selects one from the selection items shown in the display screen example of FIG. 14A. In step E4, the translation unit 2 determines whether or not the candidate selected in step E3 is an instruction to display the learning history sentence.

If it is determined in step E4 that the learning history sentence display instruction has been selected, the user selects a translated word candidate for which the learning history sentence is desired to be referred to in step E5. Next, in step E6, the information of the translated word selected in step E5 is sent from the translation unit 2 to the learning content display control unit 5 connected to the translation unit 2, and the learning content display control unit receives the information. 5 checks whether the learning history sentence corresponding to the selected translated word exists in the learning history storage area. If there is a learning history sentence, the learning content display control unit 5 sends the learning history sentence of the translated word and the learning content of the translated word from the learning dictionary to the output display unit 3 and displays them on the output display unit 3 (step E7). .

For example, in the selection item shown in the display screen example of FIG. Select << Learning history >> of No. 4 and select the translated word NO. When the instruction to display the learning history related to “work” in 2 is given, as shown in the display screen example of FIG. 14B, the original sentence “I am doing a good task.
"And its translation" I am doing a good job "is displayed along with the learning content of the translation.

Next, in step E8, an instruction for ending the display of the learning history sentence is waited for.
Returning to step 6, the learning history sentence continues to be displayed if there is no display end instruction input.

If there is no learning history sentence in step E6, a message indicating that the learning history sentence cannot be displayed because it does not exist is displayed, and the processing of the CPU returns to step E2.

If no translation word candidate is learned in step E1, the normal translation word candidate is displayed.
The process of the CPU proceeds to step D6 of process D, and only the normal selection candidates are displayed without the selection candidates for instructing the display of the learning history sentence and the fact that the learning has been performed.

Further, when the learning history sentence is not displayed in step E4, the CPU processing is step D of processing D.
Return to 8 and perform the subsequent processing.

As described above, in this embodiment, with respect to the specific word appearing in the sentence being translated, the learning history of the translated word can be confirmed by the learning history sentence. For this reason,
It is easy to discover that there is inappropriate learning, and it is possible to cancel or change inappropriate learning while confirming the intention of learning at any time.

(Fifth Embodiment) FIG. 4 is a block diagram for explaining a machine translation device according to a fifth embodiment of the present invention .

In the figure, reference numeral 105 denotes a machine translation apparatus of this embodiment, which is effective only for normal learning and a specific document at the time of learning in addition to the configuration of the machine translation apparatus of the fourth embodiment. A learning dictionary registration unit 9 that distinguishes between learning that is effective for a specific sentence and learning that is effective only for a specific sentence, and a learning type for distinguishing and translating translation words that have been learned by the learning dictionary registration unit 9 The learning type discriminating unit 10 is provided.

Next, the operation will be described with reference to the flowcharts of FIGS. 8 and 10.

FIG. 10 shows a flow of the process F for selecting a learning type and learning a translated word.

This process F starts when an instruction to learn the selected translated word is given in step D9 of the process D shown in FIG.

First, at step F1, learning type candidates are displayed. In the present embodiment, the display screen shows normal learning,
As shown in the display screen example of FIG. 14C, three types of document-limited learning in which learning is effective only for a specific document and one-sentence limited learning in which learning is effective only for one specific sentence are displayed.

Next, when the user selects one of these learning types in step F2, the processing is divided depending on the selected learning type (step F3).

First, when "normal learning" is selected, in step F4, the learning dictionary registration unit 9 selects the translation word to be learned.
Set the normal learning flag that indicates normal learning,
Learning is implemented in step D10 of process D.

When "document limited learning" is selected from the learning types, a document name for which learning is limited is input in step F5. Next, in step F6, the learning dictionary history storage unit 8 stores the document name input in step F5 in the learning history storage area. Next, in step F7, the learning dictionary registration unit 9 sets a document limited learning flag indicating that the translation is to be learned, and the learning is performed in step D10 of the process D.

When "single sentence limited learning" is selected from the learning types, in step F8 the learning dictionary registration unit 9 stores the original sentence currently being translated in the learning history area as a learning limited sentence for limiting learning. . Next, in step F9, the learning dictionary registration unit 9 sets a one-sentence limited learning flag for the translated word that is the learning target, and executes learning in step D10 of process D. .

Examples of the contents of the learning dictionary at this time are the learning dictionary contents example shown in FIG. 15A and the learning history contents example shown in FIG. 15B. A flag indicating the type of learning is set for each translated word stored in the learning dictionary. In the book (book) of 01, a flag "one sentence limit" indicating learning effective only for one sentence is set, so that the learning of this translation word is NO.
“He has bee” stored in the storage area of the history sentence of 01
Only one sentence of "n keeping the books." is valid, and in other sentences, this learning is invalid.

In addition, NO. 02 pa
Since a flag "document limit" indicating learning effective only for a specific document is given to the per (newspaper), learning of this translated word is NO. The document "businesswee" stored in the storage area of the learning-limited document name 02.
It is shown to be valid only in "k".

The translation process (step B15) when the learning dictionary that has been learned as described above is used is the process G.
Will be described in detail along with the flowchart of FIG.

First, when the translating unit 2 refers to an arbitrary learning dictionary in the learning dictionary 4 and translates it, the learning type discriminating unit 10 determines whether or not the normal learning flag is set for the translated word selected in the learning dictionary. Check whether or not (step G1).

If YES, the process of the CPU proceeds to step G4, where the translation process is performed using the selected translated word, the translation result is output at step G5, and the process ends.

If NO at step G1, step G
In 2, the learning dictionary discrimination unit 10 checks whether or not there is a document limited learning flag. If YES, the process advances to step G6 to check whether or not the learning-restricted document name stored in the learning history storage area matches the currently translated document name. If they match, the process proceeds to step G4, and translation is performed using the selected translated word.

If the learning-restricted document name and the name of the document being translated do not match in step G6, translation is performed without using the selected translated word in step G7, and the translation result is output in step G5, and the processing is executed. finish.

If the document limited learning flag is not set in step G2, it means that the one sentence limited learning flag is set, so the process proceeds to step G3, and the original sentence of the learning history sentence stored in the learning history storage area. And the original text being translated are checked. If YES, in step G4, the translation unit 2 translates using the translated word. If NO, in step G7, the translation is performed without using the translation word.

For example, "He has been keeping the book
The original sentence "s." is the learning dictionary shown in FIG.
BUSINESS. If you translate using gakushu, book
"Book" that is learned in this learning dictionary as a translation of
Is selected. Next, the learning type flag of the learning dictionary is checked. Since the learning type flag of this translated word is the one-sentence limited learning flag as shown in the learning dictionary content example, the original sentence of the history sentence stored in the learning history storage area of the learning dictionary matches the original sentence being translated. To find out.

Here, since they match, this learning is valid and the selected translation "book" is translated as the translation of book.

Further, a document called newsweek is shown in FIG.
When you translate using the learning dictionary BUSINESS. Since the document limitation flag is set in "," it is checked whether the learning limited document name stored in the learning history storage area of the learning dictionary matches the translated document name. As shown in the learning history content of FIG. 15 (b), the translation “paper” of paper is a valid learning only for business week, so this learning is invalid and the translation “newspaper” selected is not translated. become.

As described above, according to the present embodiment, learning is performed separately for each range in which learning is effective, and the learning is performed by discriminating and translating them at the time of translation. Therefore, learning in which learning contents having different application ranges are mixed. It can be effectively used in a dictionary.

When learning, the learning type is distinguished by the set of flags indicating the learning type. Therefore, the learning is performed by changing the flag and manipulating the information for limiting the learning of the learning history storage area. It is possible to easily change the type of learning for a particular translated word.

In the present embodiment, when the one-sentence limited learning is performed, only the original sentence when the learning is performed is stored.
When a sentence similar to the stored one sentence is translated, the one-sentence learning works effectively, but if it is not desired to reflect the one-sentence learning even in the same sentence depending on the context of the document to be translated, learning limitation is limited. By storing the document name and the learning-limited sentence as a pair, learning can be limited to a specific sentence of the specific document.

[0137]

[0138]

As described above, the machine translation apparatus according to the present invention
According to the document name of a document that has been translated using a certain learning dictionary. Thus in association with each memory and each learning dictionary, it is possible to provide the use history of the learning dictionary to the user, and therefore, Not only the person who created the learning dictionary but also other users can obtain information that serves as a criterion for determining whether or not to use the learning dictionary, and it is possible to select a learning dictionary suitable for the content of the original text to be translated. Has the effect of becoming.

Further, according to the present invention , since the original sentence at the time of learning and the translated sentence are stored in association with the learned translated word, in addition to the above effect, the intention of learning can be confirmed. The learning history sentence of can be provided to the user, and therefore, it is possible to maintain the quality of the contents of the learning dictionary without accidentally canceling the learning contents without knowing the learning intention. Further, since the learning intention can be confirmed even if the learning dictionary is not created by oneself, there is an effect that a plurality of users can confirm the learning intention and reuse the learning dictionary with peace of mind.

Further, according to the present invention , since the learning history sentence for a specific word can be displayed during translation, in addition to the above effect, information for confirming the intention and validity of learning for a specific word can be displayed. It can be provided to the user during translation, so that the user can cancel or change learning while confirming the intention of learning, and inappropriate learning that is difficult to find just by listing the entire contents of the learning dictionary. This has the effect of making it easier to find.

Further, according to the present invention , normal learning,
In addition to the above effect, learning is performed by distinguishing between learning that is effective only for a specific document and learning, and in addition to the above effects, learning dictionaries that have learning contents with different application ranges can be used for special learning. The learning dictionary can be used without worrying about the influence, and thus the range in which the learning dictionary can be used is expanded, and the range of reuse by a plurality of users is expanded. Also, this
According to the invention, it is effective only for normal learning and a specific sentence.
Learning and distinguishing them when translating
In addition to the above effects, the range of application is different.
The learning dictionary with learning content
The range in which learning dictionaries can be used without
And widespread reuse by multiple users.
The effect is

Further, since general-purpose learning is gradually built up, it becomes possible for a plurality of users to effectively use a high-quality learning dictionary, and the learning distinction is flagged to indicate the type of learning. Since this is done by the operation, there is also an effect that the range of application of learning can be easily changed by operating this flag.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining a configuration of a machine translation device according to a first embodiment of the present invention.

FIG. 2 is a block diagram for explaining a configuration of a machine translation device according to second and third embodiments of the present invention.

FIG. 3 is a block diagram for explaining the configuration of a machine translation device according to a fourth embodiment of the present invention.

FIG. 4 is a block diagram for explaining the configuration of a machine translation device according to a fifth embodiment of the present invention.

FIG. 5 is a diagram showing a flow of processing for setting a learning dictionary in each of the embodiments.

FIG. 6 is a diagram showing a flow of an operation of mainly storing a usage history of a learning dictionary in the second to fifth embodiments.

FIG. 7 is a diagram showing a processing flow of learning content display and learning dictionary usage history display in the second to fifth embodiments.

FIG. 8 is a diagram showing a flow of processing for storing learning history sentences in the third to fifth embodiments.

FIG. 9 is a diagram showing a flow of processing for displaying learning history sentences in the fourth and fifth embodiments.

FIG. 10 is a diagram showing a flow of a process F for learning a translated word by selecting a learning type in the fifth embodiment.

FIG. 11 is a diagram showing a flow of a translation process using a learning dictionary in which a learning type is selected and a translated word is learned in the fifth embodiment.

FIG. 12 is a diagram showing an example of a display screen in each of the above embodiments, FIG. 12 (a) is an initial menu screen, and FIG. 12 (b).
Shows a learning dictionary selection screen, and FIG. 12 (c) shows a learning dictionary content display screen.

13 is a diagram showing an example of a display screen in the above embodiment, FIG. 13 (a) is a learning dictionary use history display screen in the second to fifth embodiments, and FIG. 13 (b) is a third to fifth embodiment. 13 (c) shows a learning content display screen in the third to fifth embodiments.

FIG. 14 is a diagram showing an example of a display screen in the above embodiment, FIG. 14 (a) is a translation word selection screen in the fourth and fifth embodiments, and FIG. 14 (b) is a fourth and fifth embodiment. 14C shows a learning history sentence display screen in FIG. 14, and FIG. 14C shows a learning type selection screen in the fifth embodiment.

FIG. 15 is a diagram showing an example of a display screen in the fifth embodiment, FIG. 15 (a) is an example of contents of a learning dictionary, and FIG.
(B) has shown the example of a learning history content.

[Explanation of symbols]

1 Input section 2 Translation Department 3 Output display section 4 Translation dictionary section 5 Learning dictionary display controller 6 basic dictionary 7 learning dictionary section 7a, 7b, 7c Learning dictionary 8 Learning dictionary history storage 9 Learning dictionary registration section 10 Learning species discriminator 101, 102, 104, 105 Machine translation device

Claims (5)

(57) [Claims] 1. An input unit for inputting a document including a sentence in a first language, a basic word dictionary storing a plurality of words in the first language and dictionary information in a second language corresponding to each word, And a translation dictionary unit having a learning dictionary unit composed of various types of learning dictionaries for storing the dictionary information together with the words in the first language according to the usage history, and the input sentence in the first language is translated into the translation dictionary unit. The translation unit includes a translation unit that translates the sentence in the second language using the dictionary information of the dictionary unit, and an output display unit that displays processing information in the translation unit. The translation unit translates a sentence in a certain first language. When translated, the dictionary information of the basic word dictionary used for translation of this sentence is stored in the learning dictionary unit, and a machine learning function equipped with a dictionary learning function that preferentially uses these dictionary information for subsequent translations. A device that translates using a predefined learning dictionary Document name of the document was, machine translation apparatus having a learning dictionary history storage unit which associates and stores each learning dictionary. 2. An input unit for inputting a document including a sentence in a first language; a basic word dictionary storing a plurality of words in the first language and dictionary information in a second language corresponding to each word; And a translation dictionary unit having a learning dictionary unit for storing the dictionary information together with the words in the first language according to the usage history, and using the dictionary information of the translation dictionary unit for the input first language sentence. A translation unit for translating into a sentence in a second language and an output display unit for displaying processing information in the translation unit. When the translation unit translates a sentence in a first language, A machine translation device equipped with a dictionary learning function that stores dictionary information of a basic word dictionary used for translation in the learning dictionary unit and preferentially uses the dictionary information for subsequent translations. When the dictionary information is stored in the dictionary section, the translated At least one of the learning history statements first language sentence and the second language sentence of the machine translation apparatus having a learning dictionary history storage unit for storing the said learning dictionary section in association with the dictionary specification information. 3. The machine translation device according to claim 2, further comprising a learning dictionary display control unit that displays a learning history sentence for a specific translated word stored in the learning dictionary history storage unit during translation. 4. An input unit for inputting a document including a sentence in a first language; a basic word dictionary storing a plurality of words in the first language and dictionary information in a second language corresponding to each word; And a translation dictionary unit having a learning dictionary unit for storing the dictionary information together with the words in the first language according to the usage history, and using the dictionary information of the translation dictionary unit for the input first language sentence. A translation unit for translating into a sentence in a second language and an output display unit for displaying processing information in the translation unit. When the translation unit translates a sentence in a first language, A machine translation device equipped with a dictionary learning function that stores dictionary information of a basic word dictionary used for translation in the learning dictionary unit and preferentially uses the dictionary information for subsequent translations. When storing the dictionary information in the dictionary section, normal learning and A learning dictionary registration unit that stores the dictionary information by distinguishing it from learning that is valid only for a specific document, and dictionary information that is stored as learning that is valid only for the specific document when translating a specific document; A machine translation apparatus that translates by preferentially using the dictionary information stored as the normal learning, and preferentially uses only the dictionary information stored as the normal learning when translating other than the specific document. . 5. An input unit for inputting a document including a sentence in a first language, a basic word dictionary storing a plurality of words in the first language and dictionary information in a second language corresponding to each word, And a translation dictionary unit having a learning dictionary unit for storing the dictionary information together with the words in the first language according to the usage history, and using the dictionary information of the translation dictionary unit for the input first language sentence. A translation unit for translating into a sentence in a second language and an output display unit for displaying processing information in the translation unit. When the translation unit translates a sentence in a first language, A machine translation device equipped with a dictionary learning function that stores dictionary information of a basic word dictionary used for translation in the learning dictionary unit and preferentially uses the dictionary information for subsequent translations. The first translated when storing the dictionary information in the dictionary section There is a learning dictionary history storage unit that stores a sentence of a word as a learning history sentence in the learning storage unit in association with the dictionary information, and a normal learning when the dictionary information is stored in the learning dictionary unit. A learning dictionary registration unit that stores the dictionary information by distinguishing it from learning that is effective only for a specific sentence, and a learning type discriminating unit that discriminates these learning types during translation, When the learning is effective only for one specific sentence having a certain type, the translation unit performs the learning only when the sentence being translated matches the learning history sentence stored in the learning dictionary history storage unit. A machine translation device that translates by preferentially using dictionary information in the dictionary section.