JP2786211B2 - Post-editing device in machine translation equipment - Google Patents

Description

Description: TECHNICAL FIELD The present invention relates to a post-editing device for a machine translation device that uses a translation processing method.

2. Description of the Related Art In the conventional alternative word processing, a translated word is displayed by utilizing a translated word, or is displayed together with an attached word part. Alternatively, only the translation is displayed. Also, when replacing,
There is a difference in that processing such as making use of the word and replacing the attached word part is performed. However, as alternative translations, all can be selected as translations in the bilingual dictionary, and if you want to put the original in the translation as it is, look at the spelling of the original and It is necessary to insert it inside and delete the original translation. Therefore, there are drawbacks in that the spelling may be misspelled, and the unnecessary place may be erased by mistake.

With a so-called machine translation device that automatically translates from one language A to another language B, it is difficult to obtain a perfect translation. Therefore, in current machine translation apparatuses, it is common practice to post-edit the result of machine translation to obtain an accurate translation.

After that, as a part of the editing function, a translated word function is often provided. This means that from the dictionary (when there are multiple translations for a certain expression), the translations in the second and lower ranks are taken out and replaced with the first translation in the translation (some of them are also used at the same time). is there. However, this approach has problems when trying to translate meta-expressions. That is, when a sentence such as a description of a word in language A is translated, CPU means Central Processing Unit. Congress is on
When a sentence such as e type of meeting. is translated, a strange translation such as "the central processing unit is a central processing unit" or "a meeting is a type of meeting" is often made. Of course, for example CPU
In the case of, it can be solved by inserting the translation of CPU as the second translation, but if there is no suitable translation or if there is something like a synonym completely, the current Technology cannot solve it. Also, as in the case of the description of a grammatical event, there is a similar difficulty when it is desired to output a spelling of an original language as a translated word.

Object of the Invention The present invention has been made to solve the above-mentioned drawbacks, and a part (CPU, Congres
The purpose of the present invention is to provide a post-editing device for a machine translation device that automatically recognizes expressions such as CPU and Congress as spelled words as spelled words.

To achieve the above object, the invention of claim 1 divides at least a bilingual dictionary of language A and language B and a translation result of language A into language B into words (idioms and expressions). Means for inputting, means for displaying a translation of language B inputted for each unit, and language B for translation
Means for searching the corresponding language A in the bilingual dictionary for each word (each expression), and plural words from the bilingual dictionary searched for each word (each expression) in the designated language B. Means for extracting a translation and displaying it as another translation, means for specifying an appropriate translation from among the translations, exchanging the designated translation for the translation in the translation, and redisplaying the translation in language B In the post-editing device of a machine translation device having a means for performing the translation, the means for displaying as another translation word displays a headword in the searched dictionary as one of the other translation words. A bilingual dictionary of language A and language B, a means for dividing and inputting a translation result of language A into language B in units of words (idioms and expressions), and displaying a translation of language B input for each unit Means and translation language B
Means for searching another corresponding word in the bilingual dictionary for each word (each expression), and for each word (each expression) in the language B of the translated sentence, which expression in the original language A sentence A means for knowing whether the word is a translation, a means for taking out another corresponding word from the bilingual dictionary searched for each word (each expression) in the designated language B, and displaying the same as another word, Means for indicating an appropriate translation from among
In the post-editing device of the machine translation apparatus having a means for exchanging the designated translation with the translation in the translation and redisplaying the translation of language B, the means for displaying the translation as the translation is one of the translations. Displaying the expression in the sentence of the language A corresponding to the instructed translation as it is spelled. In the invention of claim 3, in the invention of claim 1, the searching means corresponds to the translation of language B This is a means of searching for the word of language A that appeared in the original sentence, and what kind of conjugation form the word is. According to a fourth aspect of the present invention, in the third aspect of the present invention, the means for displaying as a different translation word includes, as another translation word, a headword of a corresponding word in the language A, and an inflected form in the original text. To show both In the invention according to claim 5, in the invention according to claim 2, the means for displaying as a separate translation word, if the position of the word in the original sentence corresponding to the designated translation word is the beginning of a sentence, changes the first character to lowercase. The converted word is displayed as one of the translated words. According to a sixth aspect of the present invention, in the invention of the second aspect, the means for displaying the translated word is a position of a word in the original text corresponding to the designated translated word. Is the beginning of a sentence, the first character is converted to lowercase and the one that remains uppercase is displayed as another translated word. The invention according to claim 7, wherein In the present invention, the means for displaying as a separate translation includes, as one of the translations, an expression of the language A corresponding to the expression of the designated language B to be displayed as a translation having the lowest priority (as a final translation. ) Display, billing According to an eighth aspect of the present invention, in the invention according to any one of the first to sixth aspects, the means for displaying as another translation word includes, as one of the other translation words, an expression of the language A to be displayed as another translation word. The invention according to claim 9, wherein at least the bilingual dictionary of language A and language B and the translation result of language A into language B are divided and input in units of words (idioms or expressions). Means for displaying the translated sentence of the language B inputted for each unit together with the original sentence of the language A, and for each word (each expression) of the language A of the original sentence, Information means of correspondence, means for designating each expression in language A, and, for each word (each expression) in designated language A, a plurality of translation words are extracted from the retrieved bilingual dictionary and separated. Means to display as translations, and appropriate translations from among the translations And a post-editing machine translation device having means for exchanging the designated translation with the translation in the translation corresponding to the language A instructed and redisplaying the translation in language B. The means for displaying as another translation word displays a headword in a searched dictionary as one of the another translation words. The invention according to claim 10 is the invention as set forth in the invention of claim 9, wherein said means for displaying as another translation word is However, the expression of the designated language A is used as it is as one of the translated words. The invention of claim 11 is the invention of claim 9, wherein the means for displaying the translated word is one of the translated words. Displaying the inflected form found in the searched dictionary; The invention of claim 12 is the invention according to claim 9, wherein the means for displaying as another translated word includes, as one of the other translated words, Both words and inflected forms searched In the invention according to claim 10, the means for displaying as a separate translation word, when the position of the designated word in the original sentence is the beginning of the sentence, In the invention according to Claim 10, the means for displaying as a separate translation word is such that the position of the designated word in the original sentence is the beginning of the sentence. In the case of, both the first letter converted to lowercase and the one left in uppercase are displayed as separate translations. The invention according to claim 15 is the invention according to any one of claims 9 to 14 In the above, the means for displaying as an alternative translation may include, as one of the alternative translations, an expression of the language A corresponding to the expression of the indicated language B to be displayed as a translation having the lowest priority (as a final alternative translation). The invention of claim 16 is: In the invention according to any one of Claims 9 to 14, the means for displaying as another translation word displays the expression of the language A to be displayed as one of the other translation words separately from another translation word existing in the bilingual dictionary. It is characterized by the following. Hereinafter, an example in which the present invention is applied to an English-Japanese machine translation apparatus will be described. The present invention is realized as a part of a translation support editor in an English-Japanese machine translation device.

Start the translation support editor of the English-Japanese machine translation device and enter the English sentence. The input English sentence is translated by the English-Japanese machine translation unit as a Japanese sentence using a translation dictionary (Table 1).

The dictionary is a search utilization dictionary (1-1-1) for efficiency.
Tables) and bilingual dictionaries (Table 1-2), but both may be combined into a single form. Also,
Although the arrangement of the dictionaries is in alphabetical order, it does not matter that the arrangement of the dictionaries is appropriately changed depending on the search means used. In Table 1-1, an address indicating the corresponding entry in Table 1-2 is stored. In addition, a search from English words is performed by looking at the search utilization bulletin.
At the same time, a table (Table 2) in which the correspondence between the English word and the word being translated in Japanese is obtained is also output.

The translation support editor utilizing the present invention has a function of displaying the English words corresponding to the Japanese translations of the English words using the correspondence table of Table 2. For each word / expression, it is easy to realize because it has information as to which number in the table of another language it is.

In addition, since the length of each word / translation / attachment part can be easily known, it is easy to know which expression the cursor (or other pointing device) indicates by using existing technology. It is feasible.

In the present embodiment, an English-Japanese bilingual display is performed so that a Japanese word as a translated sentence may be selected or an English word as an original sentence may be selected.

It is also possible to display only the translated Japanese, without selecting the English-Japanese bilingual display, and to select only from the translated words.

In this case, there is no need for an English word table. Since the address of the searchable dictionary for the corresponding word is stored in the translation word table, the word in the original sentence can be easily searched therefrom.

FIG. 1 is an overall configuration diagram of a post-editing apparatus in machine translation according to one embodiment of the present invention, and FIG. 2 is a flowchart for explaining the operation of the post-editing apparatus. Apparatus, 3 is a machine translation unit, 4 is an alternative word processing unit, 5 is an English word table, 6 is a translation word table, 7 is an input unit, and 8 is a translation dictionary. Hereinafter, description will be made based on the flowchart of FIG.

When the alternative word selection device of the present invention is activated and the user selects an English word or a Japanese word (Step 1), 1. If an English word is selected, a predetermined word in the translation word table is selected from the correspondence table. get information.

When Japanese is selected, the information of the translation table corresponding to the Japanese can be obtained (Step2, Step
3). (The correspondence between the display / selection on the screen and each table makes it possible to easily know the position of each Japanese translation in the translation by the length of each translation / attachment. Of course. Even if there is no length information, it is easy to know the relative position from the character length of the translated word. In this embodiment, the length is set in a table in advance to improve efficiency.) The necessary dictionary information is obtained from the dictionary address in the translation table. (If the translated word table already contains information on all other translated words, the same effect can be obtained by extracting the information.) A predetermined position on the screen according to the dictionary information Then, all translations are displayed (Step 4). (This is achieved, for example, by creating a window for another word. This can also be easily achieved with existing technology.) 3. Insert a blank line. (This is not a blank line that is used to distinguish between the translations registered in the dictionary and the translations that come from the spelling. Of course, it is also possible to distinguish them by displaying another window, etc.) 4. As the last translation, the English spelling is displayed as it is (Step 5).

When the user selects one of the translations, the translation in the table is rewritten to the selected translation. If the translated word is an inflected word (the inflection information is not “none”), the subsequent word is used in accordance with the information in the inflection table (Table 3). At this time, if the utilization is in the form of a sound, the processing is performed according to the table of sound shapes (Table 4), the first character of the attached word part is determined, and the word is rewritten if necessary.

At this time, the information on the length of the translated word is also rewritten. The translated text in the translation editor is displayed again according to the rewritten table (Step 6 to Step 8).

4. The spelling of English in the above uses the spelling after the heading in the dictionary at present, but it is very easy to change it to display the spelling of the inflected form instead of the spelling of the headword. When the spelling of the headword and the spelling of the inflected form are different, it is needless to say that displaying both can be easily realized.

Further, it is also possible to make a change so that the spelling in the original sentence is displayed as it is by using the correspondence table without performing the dictionary search. In this case, in the case of the first word of the sentence, it is easy to add a process of converting the first character to lowercase, and it is also easy to display both the uppercase and the lowercase. It is possible.

In this embodiment, the dictionary search is performed using the address of the dictionary as the information. However, if the result of the dictionary search is changed to be input to the editing device as the information, the dictionary search is performed. Needless to say, the same effect as in the present embodiment can be obtained only by the input information.

 Hereinafter, an outline of the processing will be described based on an actual example.

Originally, "A Congress is one type of meetin
g. "is input and translated by the machine translation unit using a translation dictionary as shown in Table 1, and is output as a bilingual table as shown in Table 2.

Table 2-1 is a table of English words, in the order of the original sentence (words are used in the meaning of search units in a dictionary, and may be idioms or collocations).
It is a table showing the sequence. Translation support editor
By referring to this table, the original sentence can be easily restored, and the position from the beginning of the displayed original sentence can be easily searched for which word is the pointed position. Has become.

Table 2-2 is a table of translated words, and is a table of phrases (separated into words and attached word parts) in the order of created translations. The translation support editor can easily refer to this table to display the translated sentence, and easily determine which phrase the pointed position is a part of from the beginning of the displayed translated sentence. Can be searched.

This table is for keeping correspondence between the translated word and the original sentence, and the configuration method is not limited to this.

 From the translation table, it can be seen that the translation is "a meeting is a type of meeting."

The translation support editor is configured to display an original sentence and a translated sentence in association with each other. Specifically, after displaying the original sentence, a line is opened and the translated sentence is displayed.

At this point, the bilingual word processor is activated and the congress
Think about when you are pointed. (It is also possible to configure the translation support device such that the cursor or the position pointed to by the pointing device is automatically sent to the translated word processing device when the translated word processing device is activated.) Therefore, from the information on the position of the translation word table, the same process (the process described below) as that of "meeting" at the beginning of the translated sentence may be performed.

When the head of the translated sentence "meeting is" is pointed out, the translation dictionary (Table 1) is obtained from the dictionary address information in the translation word table.
To find out that the corresponding translated words are “meeting” and “meeting”. At this time, information on the part of speech and utilization of the translated word is also stored in association with each translated word.

After displaying the two translations of "meeting" and "meeting", it is a headline of the searched dictionary with one line open. Display congress. The part-of-speech for this translated word taken from the original sentence is considered to be noun and its use is “none”. (here,
Since the English language is limited to use only the 1-byte code system and the Japanese language only uses the 2-byte code system, code conversion is performed. Of course, it is also possible to configure so that the 1-byte code system can be included in Japanese sentences. In that case, the length information of the translation table is
Needless to say, it is necessary to consider the byte length of the code and the code length for code switching, if any. By the way, if the user selects congress, the utilization is "none", and there is no need to consider utilization processing.

In the translation table, change "conference" of the first translation to "cong".
ress ”, and at the same time,
Rewrite to 8.

After that, the displayed translation is displayed again according to the translated translation table, and the correct translation "congress is one type of meeting."

Effects As is clear from the above description, the present invention has the following effects.

In the post-editing device of claims 1 and 6, as one of the translated words,
Since the form of the headword of the word in the original sentence of the translated word can be obtained, even if it is a metalinguistic expression such as explanation of the word, it can be easily reflected in the translated sentence by selecting another translated word become.

In the post-editing device of claims 2 and 7, as one of the translated words,
You can obtain the exact form of the translated word in the original text, for example, The word meeting is the present participleof the
As in the case of meeting in the word "meet". (The word "meeting" is the present participle of the word "meet."), meta-expressions dealing with linguistic events can be easily reflected in the translation. Also, since no dictionary search is required, very high-speed processing is possible.

In the post-editing device according to claims 3 and 8, the same effect as in the post-editing device according to claim 2 can be obtained. Further, it is possible to solve the drawback that the head of the sentence in the post-editing device of the second aspect starts with an uppercase letter. Furthermore, since no dictionary search is performed, processing can be performed at higher speed.

The post-editing device according to claims 4 and 9 displays the word or expression in the original sentence as the last candidate of another translated word. Therefore, in the case of a general expression, it is not disturbed so much, and is used only when necessary. Enable.

In the post-editing device according to claims 5 and 10, since the general alternately translated words and the words or expressions in the original sentence are distinguished and displayed, the general alternately translated words can be used almost without being conscious when unnecessary.

Further, in the post-editing device according to claims 7 to 10, since the original sentence and the translated sentence can be viewed at the same time, it is possible to more easily select another translated word.

[Brief description of the drawings]

FIG. 1 is an overall configuration diagram of a post-editing apparatus in machine translation according to one embodiment of the present invention, and FIG. 2 is a flowchart for explaining the operation. 1 Display device 2 Post-translation processing device 3 Machine translation unit 4 Alternative word processing unit 5 English word table 6
... Translation word table, 7... Input unit, 8.

Claims (10)

(57) [Claims] At least a bilingual dictionary of language A and language B and a correspondence table between each word in the original text and each translated word in the translated sentence are input as a translation result from the original sentence of language A to the translated sentence of language B. A means for displaying a translated sentence for each word, a means for designating a word in the translated sentence, a means for searching a corresponding word in the original sentence from the correspondence table using the designated word in the translated sentence as a key, Means for searching the bilingual dictionary for a headword in language A and a translation or translation group in language B using the words in the original text retrieved from the correspondence table as keys, and a headword and its translation or translation in the bilingual dictionary Means for displaying the group as a separate translation, means for specifying a translation from the displayed translations, and means for redisplaying the translation by exchanging the designated translation for the original translation in the translation. Characteristic post-editing equipment in machine translation equipment . At least a bilingual dictionary of language A and language B and a correspondence table between each word in the original sentence and each translated word in the translated sentence are input as a translation result from the original sentence of language A to the translated sentence of language B. A means for displaying a translated sentence for each word, a means for designating a word in the translated sentence, a means for searching a corresponding word in the original sentence from the correspondence table using the designated word in the translated sentence as a key, Means for searching the translation dictionary for a translation or group of translations of language B using the words in the original sentence retrieved from the correspondence table as keys, and the words in the original sentence retrieved from the correspondence table and the words retrieved from the bilingual dictionary. Means for displaying a translation or a group of translations as separate translations, means for specifying a translation from among the displayed translations, and means for redisplaying the translation by exchanging the designated translation for the original translation in the translation In a machine translation device characterized by having Editing apparatus. 3. The means for displaying as another translation word, when the position of the word in the original sentence corresponding to the designated translation word is the beginning of the sentence, converts the first character of the word into a lowercase letter as one of the other translation words. 3. The post-editing apparatus in the machine translation apparatus according to claim 2, wherein the post-editing is performed. 4. The means for displaying as an alternative translation, as one of the alternative translations, an expression of language A corresponding to the expression of the indicated language B to be displayed as a translation with the lowest priority (final translation). 2. A display as a translation
A post-editing device in the machine translation device according to any one of claims 1 to 3. 5. The method according to claim 1, wherein said means for displaying as another translation word displays, as one of the other translation words, the expression of the language A to be displayed in distinction from the other translation words present in the bilingual dictionary. Item 4. A post-editing device in the machine translation device according to any one of Items 1 to 3. 6. At least a bilingual dictionary of language A and language B and a correspondence table between each word in the original sentence and each translated word in the translated sentence are input as a translation result from the original sentence of language A to the translated sentence of language B. Means, means for displaying both the original sentence and the translated sentence for each word, means for indicating a word in the original sentence, and means for searching a corresponding translated word in the translated sentence from the correspondence table using the designated word in the original sentence as a key Means for retrieving a headword of language A and a translation or group of translations of language B from the bilingual dictionary using the designated word in the original text as a key, and a headword retrieved from the bilingual dictionary and its translation or group of translations Has a means for displaying a translation as another translation, a means for indicating a translation from among the displayed translations, and a means for replacing the designated translation with a translation in a translation retrieved from a correspondence table and redisplaying the translation. Machine translation device Editing device after that. 7. At least a bilingual dictionary of language A and language B and a correspondence table between each word in the original sentence and each translated word in the translated sentence as a translation result from the original sentence of language A to the translated sentence of language B are inputted. Means, means for displaying both the original sentence and the translated sentence for each word, means for indicating a word in the original sentence, and means for searching a corresponding word in the translated sentence from the correspondence table using the designated word in the original sentence as a key Means for searching the bilingual dictionary for a headword of language A and a translation or group of translations of language B using the designated word of the source text as a key, and separating the word of the specified source text and its translation or translation group. Means for displaying as a translation,
A machine translation device having means for designating a translation from among the displayed translations, and means for redisplaying the translation by exchanging the designated translation with a translation in a translation retrieved from a correspondence table. Post-editing device. 8. The means for displaying as another translation word, wherein, when the position of the designated word in the original sentence is the beginning of a sentence, the first letter converted to lower case is displayed as one of the other translation words. The post-editing device in the machine translation device according to claim 7, wherein 9. The means for displaying as an alternative translation, as one of the alternative translations, an expression of the language A corresponding to the expression of the indicated language B to be displayed as the translation having the lowest priority (the final alternative). 7. Display as a translated word).
A post-editing device in the machine translation device according to any one of claims 1 to 8. 10. The method according to claim 1, wherein the means for displaying as another translation word displays the expression of the language A to be displayed as one of the other translation words, separately from the other translation word existing in the bilingual dictionary. Item 9. A post-editing device in the machine translation device according to any one of Items 6 to 8.