JPH056396A - Machine translation device - Google Patents

Abstract

(57) [Abstract] [Purpose] To provide a machine translation device capable of effectively obtaining a translation result without damaging information indicated by various non-character data included in document information. In a machine translation device that translates an input original sentence by a translation unit 5 using information stored in a translation dictionary unit 6 and outputs the obtained translated sentence, it is included in a character data string forming the original sentence. It has means for detecting non-character data relating to the format of the original sentence, and means for adding the non-character data detected by this means to the character data string forming the translated sentence.

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 capable of improving the efficiency of translation editing work.

[0002]

2. Description of the Related Art Recently, a machine translation apparatus that automatically machine-translates an input original sentence by using a computer and obtains the translated sentence has attracted attention. For example, various attempts have been made to develop a machine translation device for a natural original language that inputs a Japanese sentence to obtain its English translation, or inputs an English sentence to obtain its Japanese translation.

This type of device basically divides an input original sentence into predetermined processing units such as words (phrases) by performing morphological analysis and syntactic analysis. Then, the translation dictionary is searched for each processing unit to obtain translation words (translation phrases) corresponding to each processing unit, and these translation words (translation phrases) are combined according to a predetermined translation rule to obtain the translation sentence. .

[0004]

By the way, the document information is composed of character data representing the content of the document and non-character data indicating various attributes relating to the format and characters of the entire document. Machine translation processing translates original text information represented by character data in such document information,
Its main purpose is to obtain the translated text.

Therefore, if only the character data in the input document information is extracted and the machine translation process is executed sequentially,
Various kinds of non-character information included in the document information are lost, and the translation result is very difficult to understand. Specifically, if the format information is missing, for example, a great deal of labor is required to format the translated sentence that is the translation result. Moreover, if the paragraph information is lost, it causes a problem that the semantic interpretation of the translated text becomes very difficult.

The present invention has been made in consideration of such circumstances, and an object thereof is to maintain the information indicated by various non-character data contained in document information without damaging the information.
It is to provide a machine translation device that can effectively obtain the translation result.

[0007]

According to the present invention, an input original sentence is translated using information stored in a translation dictionary unit,
In a machine translation device that outputs a translated sentence obtained by this translation process, means for detecting non-character data regarding the format of the original sentence included in the character data string forming the original sentence, and non-character data detected by this means Is added to the character data string forming the translated sentence.

Specifically, when the non-character data specifies a format, a translation result according to the format is obtained, and when the non-character data indicates an underline, the original word underlined is added. Translation (phrase) for (phrase)
Is also output with an underline.

[0009]

In the present invention, when the translation process is completed and the translated sentence as a result of the translation is displayed / output, the character data string forming the translated sentence is formed according to the information of the non-character data such as the format information added to the original sentence. The form of display and output is controlled. That is, non-character data indicating the format of the original sentence can be reflected in the translated sentence output.

Therefore, it becomes possible to clearly show the correspondence between the original sentence and the translated sentence, and the burden on the post-processing and editing of the translation result can be greatly reduced. Therefore, it becomes possible to simplify the translation editing work, and to easily and efficiently obtain a translated sentence of an appropriate language expression.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. In this embodiment, an English sentence is input, and this is machine translated into Japanese. FIG. 1 is a schematic configuration diagram of the device of the embodiment.

In FIG. 1, reference numeral 1 is an input unit 1 including a keyboard and the like. The keyboard constituting the input unit 1 includes, for example, as shown in FIG. 2, a key group 1a for inputting character data, a key 1b for translation instruction, and a key group 1 for editing.
c, a function control key group 1d, a cursor control key group 1e in the display section 8 and the like.

Therefore, the English sentence input from the input unit 1 or the like is stored in the original sentence storage unit 2 as an original sentence to be subjected to translation processing. At this time, the non-character data such as the format information and the character attribute information input from the input unit 1 together with the character data forming the original sentence is recognized by the non-character data recognition unit 3. Then, according to the recognition result, the information of the non-character data is stored in the original text storage unit 2 in association with the character data in the vicinity of the detected position or in association with the sentence number or the like for managing the input original sentence.

The editing control section 4 sequentially reads the original texts stored in the original text storage section 2 under the control of the sentence numbers and gives them to the translation section 5. The translation unit 5 uses the knowledge information for translation processing stored in advance in the translation dictionary unit 6 to sequentially machine-translate the original text read from the original text storage unit 2 in predetermined processing units.

The knowledge information stored in the translation dictionary unit 6 includes, for example, a rule / irregular change dictionary 6a, a word (translated word) dictionary 6b, an unconnectable part-of-speech sequence rule dictionary 6c, a translated sentence dependency dictionary 6d, and the like. . A Japanese sentence, which is a translated sentence obtained by machine-translating the above original sentence using such knowledge information,
The translated sentence storage unit 7 is sequentially managed by associating with the obtained original sentence.
Stored in.

When storing the translated sentence in the translated sentence storage unit 7, the edit control unit 3 develops the information of the non-character data in the translated sentence according to the information of the non-character data recognized from the input original sentence. Inserting. The expansion / insertion of the non-character data into the translated text is performed by the non-character data expansion unit 8. The non-character data expansion unit 8 adds format information to the translated text in association with the sentence number, or associates character attribute information such as underline with the translated word for the original word, for example, according to the detection position of the non-character data in the original sentence. Processing such as adding. By expanding and inserting the information of the non-character data into the translated text, the information about the non-character data is added to the translated text stored in the translated text storage unit 7.

Under the control of the display control unit 10, the display unit 9 mutually translates the English sentence (original sentence) stored in the original sentence storage unit 2 and the Japanese sentence (translated sentence) stored in the translated sentence storage unit 6 into each other. It is displayed in association with each other. The display unit 9 also displays translation word candidate selection information necessary for machine translation processing. For example, as shown in FIG. 3, the display unit 9 has a display screen having a translation editing area 9a at the top of the screen, an original text display area 9b at the left side of the screen, and a translated text display area 9c at the right side of the screen.
It is divided into three parts. The input original texts stored in the original text storage unit 2 are sequentially displayed in the original text display area 9b,
In the translated text display area 9c, the translated texts stored in the translated text storage section 7 are horizontally displayed in correspondence with, for example, the original texts from which the translated texts are obtained. In the translation editing area 9a, information necessary for the translation processing such as the translation word candidates retrieved from the translation dictionary unit 6 and provided for the translation processing as described above is displayed.

In this way, the original sentence and its translated sentence are displayed on the display unit 9 and are used for post-editing processing of the translated sentence. The post-editing process is executed according to the control information input from the input unit 1, for example, by referring to the knowledge information stored in the translation dictionary unit 6 as described later.

A translated sentence (Japanese sentence) to the original sentence (English sentence) completed by such post-editing processing.
Is subjected to expansion processing based on the non-character data information by the non-character data expansion section 8 and then outputted in hard copy by the printing section 11.

FIG. 4 shows a basic operation sequence of the embodiment apparatus configured as described above. In accordance with such an operation sequence, the editing control unit 4 determines the translation end information given from the translation unit 5 and various key information inputted from the input unit 1, and interactively controls the translation / editing process. .

That is, the edit control unit 4 monitors the translation processing state in the translation unit 5 (step A), and when the completion of the translation process of one original sentence in the translation unit 5 is detected, it is obtained by the translation process. The translated sentence is expanded according to the information of the non-character data, and the information of the non-character data is inserted and stored in the translated sentence storage unit 7, and the translated sentence is displayed on the display unit 9 in association with the original sentence (step B). .

When the translation unit 5 does not give a signal indicating the completion of the translation process, the edit control unit 4 determines the key information input from the input unit 1 (steps C, D, E). , F). When the input key information is the "translation instruction key" (step C), the edit control unit 4 gives the input original sentence stored in the original sentence storage unit 2 to the translation unit 5 to start the translation process (step G).

In this translation processing, for example, as shown in the processing sequence in FIG. 5, the language form of the original sentence to be translated is first analyzed using the rule / irregular change dictionary 6a (step P). By this morphological analysis, for example, an original word that has undergone inflectional transformation or inflection change is converted into its original form (basic form). Specifically, it is a process of converting a word expressed in the past tense or progressive tense into the present tense, and converting a word expressed in the comparative or superlative into its original tense.

Next, for each original word of the original sentence subjected to the morphological analysis as described above, the translation word dictionary 6b is used to obtain information such as part-of-speech information and translation word candidates (step Q). This processing is performed by searching the translation word dictionary 6b using the original word as a headword.

Thereafter, the connectability of the translated word candidate is verified according to the information retrieved from the dictionary (step R). This verification is performed with reference to the unconnectable part-of-speech sequence rule 6c, and is repeated until a consistent syntactic analysis result (arrangement of part-of-speech sequences of translation word candidates) is obtained. By this syntactic analysis, the arrangement structure of the part of speech of the original language that constitutes the original sentence, its dependency relationship, the mode of tense, and the like are obtained.

After that, the structure of the parsed original sentence is converted into the syntactic structure of the translated sentence by using the translated sentence dependency dictionary 6d to generate a translated sentence composed of a sequence of translated word candidates of each original language (step S). . At this time, in accordance with the result of the syntactic analysis of the original sentence, each translated word candidate is subjected to inflectional transformation and ending transformation processing to make the translated sentence an appropriate language expression. By such a translation process, a translated sentence for the original sentence is obtained.

On the other hand, returning to FIG. 4, when the input key information is the "character key" (step D), the character code indicated by the character key is stored in the input buffer (step H). Then, the character code is stored in the original text storage unit 2, and the character pattern is displayed on the display unit 8 (step I). By displaying each character pattern of the character code stored in the input buffer, the original text input from the input unit 1 is displayed.

When the input key information is the "edit key" (step E), the edit process corresponding to the edit key is executed on the translated sentence (step J). Similarly, when the input key information is the "function key" (step F), the process is executed corresponding to the function key (step K).

When no key information is input, or when the input key information is other than the above-mentioned "key", other processing, for example, hard copy output of the translated text obtained in the translated text storage section 6 is performed. And so on.

When the operator operates the character input key group 1a of the keyboard to input a character by the operation sequence of the edit control section 3 as described above, the character information is sequentially set in the input buffer and is used for translation processing. The original texts are sequentially stored in the original text storage unit 2 (steps D and H). Then, the input original text is displayed in the original text display area 8b of the display unit 8 (step I).

However, at any time of character input, for example, 1
When the translation instruction key 1b is operated at the end of sentence input, the translation process for the input original sentence stored in the input buffer is started according to the key input information (step C,
G). When the translation process is completed, the translated sentence obtained by this is displayed in the translated sentence display area 8c of the display unit 8 (steps A and B).

If editing such as correction of the input original sentence is necessary, the cursor control key group 1e is operated while the original sentence is being input by operating the character input key group 1a to move the cursor to the corrected portion. And the edit key group 1c for correction / insertion / deletion etc. is operated, the edit processing is executed (steps E and J).

In this way, the basic operation of the machine translation process is controlled, and the translated sentence for the original sentence obtained by this translation process is stored in the translated sentence storage unit 7 and displayed corresponding to the original sentence. It When a plurality of translated texts for one original text are obtained, for example, only one translated text may be displayed, and at the same time, the fact that another translated text exists may be identified and displayed. Here, a brief description will be given of the translation editing process for the translated sentence by the instruction input of the editing key.

The translation editing process for this translated sentence is performed for the word (original language, original language phrase, translated word, translated phrase) designated by the cursor on the screen of the display unit 9 in accordance with the edited editing key. It is realized by Specifically, (1) If you insert a character before the cursor position by operating the insert key,

(2) Delete the character string in the range designated by the cursor by operating the delete key, (3) Move the character string in the range designated by the cursor by operating the move key, (4) Operate the cancel key to invalidate each editing function specified by each of the above keys, (5) Translate editing such as displaying other dependency candidates for the phrase pointed by the cursor by operating the dependency key Consists of processing. Further, when the function control key described above is operated, the following functions are presented, and are used for translation editing of the translated sentence described above. That is, (1) When the translated word display key is operated, the translated word is displayed for the word in the translated text pointed by the cursor. (2) When the dictionary display key is operated, the cursor points as described below. Display the contents of the translation dictionary with the words in the original text as headwords. (3) When the dictionary registration key is operated, the character string indicated by the cursor is registered as a new word or idiom in the dictionary. (4) Dictionary deletion When the key is operated, the new words or phrases that have been registered in the dictionary are unregistered. (5) When the partial translation key is operated, the partial translation for the original sentence that failed to be translated is indicated by the cursor. It is realized by the function.

A character string (word) by the above-mentioned cursor
These instructions are given by moving the cursor on the display screen by operating the cursor moving keys and changing the cursor size with the cursor control keys. By using such various functions, post-processing such as word translation correction for the translated text is interactively performed. The above is the basic translation processing function of this device. Here, the processing for non-character data, which is a feature of this apparatus, will be described in more detail.

The original text information (document information) input via the input unit 1 is character data forming the text itself,
It is divided into non-character data indicating the format and character attributes. FIG. 7 shows an example of the classification. That is, the data input from the input unit 1 is (1) character data that constitutes a document, (2) character position control data that controls the position of character data such as tabs, line feeds, page breaks, blanks, and indents, (3) Character type, size, presence of underline,
Character attribute data related to the attributes of characters such as shading,

(4) Layout data for specifying tab setting position, chart area, paper size, etc. (5) Added at the time of preprocessing and editing of the original sentence, and used as an aid for translation processing,
It can be classified as translation auxiliary data including insertion phrase designation information and scope information of conjunctions.

Here, the character position control data is usually assigned to one code unit like the character data. Character attribute data is expressed by using several bits of one code unit to represent the character attribute, or by preparing a shift code or code sequence indicating the attribute and embedding this code in the input data sequence. It The layout data is expressed by setting a code sequence indicating the layout and embedding the code sequence in the input data series. The translation assistance data can be given as information expressed as a special code or a character attribute.

The expression method for such non-character data can be set arbitrarily according to the specifications of the apparatus. Here, as shown in FIG. 7, character position control data and translation auxiliary data are coded respectively. Further, the character attribute data is given as a code indicating shift-in / shift-out of various character attributes, and further the layout data is given by setting code sequences of various layout information.

The non-character data recognizing section 3 has a code table as shown in FIG. 7, and recognizes the non-character data in the input data series by the sequence shown in FIG.

That is, first, the data sequence input from the input unit 1 is read into the recognition unit 3 (step a), and the non-character data conversion rule 3a prepared in the recognition unit 3 is used to read the non-character data in the input data sequence. We are aware of the data. This non-character data conversion rule 3a includes, for example, as shown in FIG. 8, a condition part (I) describing a pattern for an input data series,
A conversion unit that describes the conversion process for that pattern (II)
It consists of and.

The pattern of the condition part (I) is sequentially subjected to pattern matching processing with the input data series (step b), and at the time of pattern matching, the non-character data is expanded into the input data series by detecting non-character data. (Step c), the information of the conversion unit (II) is output (step d). Specifically, X = [abc] ... Substitutes abc into X {Z} ⁺ ... Matching with repetition of 1 or more Z {Z} ^* ... Matching with repetition of 0 or more Z Word ... Matching with non-blank characters sandwiched by 0 or more blanks before and after

The data series is pattern-matched in accordance with the rules described above. The item information "immediately preceding character" is information about the matching start point in the pattern matching process, and the conversion pattern specifies how to replace the matched pattern.
By such rules, for example, C / R 2.1 Software Model C /
RC / R

When the following data sequence is given (step a), the blank point next to the first C / R is used as a start, and the data sequence up to the position of the next C / R is pattern-matched (step a). b). In this case,
Since the pattern matching is performed with the first condition of the conversion rule of, the input data sequence is, for example, C / R TS 2.1 Software Model.
It will be converted as C / R C / R (step c). The pattern matching illustrated here indicates the detection of the title portion in the original sentence.

In this way, the non-character data recognizing unit 3 detects the non-character data in the input data series, deletes the non-character data as necessary, converts the non-character data series into an arbitrary symbol string, Alternatively, the edit control unit 4 is moved.
Have been converted to the data series system expected by. Then, the information of the input original text including the information of such non-character data is stored in the original text storage unit 2 (step d). The conversion processing of the non-character data can be directly described by using a program instead of being described in the form of rule as described above. By the way, the editing control section 4 handles the non-character data recognized as described above as follows.

The editing control unit 4 controls whether or not non-character data is displayed on the display screen according to the content thereof, and also adds information on the non-character data to the character data sent to the translation unit 5. It controls whether or not to do. Furthermore,
When outputting a translated sentence as a translation result, it is controlled whether or not the information of the non-character data is added to the character data string.

FIG. 10 shows an example of processing for such non-character data. That is, here, for example, when the information of the non-character data indicates "sentence start", the sentence number is generated and displayed for the original sentence display and the translated sentence display. Further, when the information of the non-character data indicates "line feed", a line feed mark is added to the end of the sentence so that it is displayed. Further, when the scope information of the conjunction of the original sentence is given, this is added to the corresponding part of the original sentence and given to the translation unit 5. In this way, the edit control unit 4 controls the display of the information of the non-character data according to the type of the non-character data.

Although the processing contents shown in FIG. 10 do not show processing for character attributes such as character type, size, and underline, for example, a character pattern reflecting the character attributes as it is on an actual display screen is displayed. Let the display. Alternatively, it is sufficient to display it as the attribute symbol like italic shift.

The processing mode shown here is set according to the specifications of the apparatus. Therefore, for other non-character data information not illustrated here, for example, the original sentence display, the translated sentence display, the transfer to the translation unit, and the like may be performed by a similar method. Further, in this example, the layout data is not displayed on the display screen, but it is of course possible to display the information together with other data.

Further, in this embodiment, the above-described processing is set assuming interactive processing, but when batch translation processing is performed in batch, another non-character data processing content is set. Of course, it is also possible.

In addition, in addition to the processing control relating to such non-character data, the edit control unit 4 also performs the correction processing of the non-character data. This correction processing moves the cursor to an arbitrary position on the display screen in accordance with the input signal from the input unit 1, inserts non-character data at the cursor position or positions before and after the cursor, or at that position. The process consists of deleting non-character data that exists in. The editing process for the non-character data is performed in the same manner as the editing process for the character data. By this processing, for example, work such as adding translation auxiliary data such as a conjunction scope to the original sentence and facilitating the translation processing is performed.

Under the control of the editing control unit 4 as described above, the expansion / insertion of non-character data in the non-character data expansion unit 8 is performed as follows. The expansion / insertion processing of this non-character data is performed using the information of the code recognition conversion table as shown in FIG. 9, for example.

This conversion table describes what kind of output data is obtained for non-character data according to the output device of the translation result. The output format of the character data forming the original sentence and the translated sentence is controlled according to the type (property) of the non-character data.

For example, when the paragraph end code PG exists at the end of the translated sentence, the code "0A0D0A0D" is output so that two blank lines are provided after the end of the sentence. That is, the line feed of the serial printer by "0A" and the carriage return by "0D" are repeatedly instructed, and the print output of the character data series forming the translated sentence is controlled. By controlling the output of the character data series based on such non-character data information, the output format of the translated text is controlled by reflecting the original text. Needless to say, such code conversion is determined according to the information of the non-character data and the control code of the output device.

Thus, according to the present apparatus having such a processing function for non-character data, when an English sentence as shown in FIG. 11 (a) is input as an original sentence for translation processing, the By recognizing the non-character code included in the data series and converting the code, the original text data including the information of the non-character data in the format shown in FIG. 11B is obtained in the original text storage unit 2.

The translation process of the original sentence is controlled according to the information of the non-character data in the original sentence. Then, the translated text obtained by this translation processing is added with the information of the non-character data, and stored in the translated text storage unit 7 as a data series as shown in FIG. 11 (c), for example.

Then, the information of the non-character data added to the original sentence and the translated sentence, respectively, is developed by the non-character data developing unit 8, and the input original sentence and its translated sentence are, for example, the first sentence.
It will be displayed as shown in FIG. That is, the translation result is displayed by adding the paragraph information of the original sentence to the display sentence and controlling the display form of the character according to the character attribute information such as underlining.

Therefore, according to this apparatus, the information of the non-character data given as the document information of the original sentence is directly reflected in the display / output of the translated sentence, and the correspondence relationship between the original sentence and the translated sentence is clearly grasped. It becomes possible to do. Therefore, it becomes possible to easily perform the translation editing process, and it is not necessary to format the translated text, so that it is possible to greatly reduce the burden on the operator, which is a great practical effect. To be

Moreover, since it is possible to effectively support the translation processing by the translation assist data which is non-character data, it is possible to expect an effect that the translation processing efficiency itself can be improved.

The present invention is not limited to the above embodiment. Although the machine translation from an English sentence to a Japanese sentence has been described here, it can be similarly applied to machine translation between other natural languages. The non-character data handled here and the processing for the non-character data may be determined according to the specifications of the device. In short, the present invention can be variously modified and implemented without departing from the scope of the invention.

[0062]

According to the present invention, when a translated sentence is displayed / output, the display / output mode of the character data string forming the translated sentence is controlled according to the non-character data relating to the format added to the original sentence. The non-character data indicating the format of can be reflected in the translated text output.

Therefore, it becomes possible to clearly show the correspondence relationship between the original sentence and the translated sentence, and the burden on the post-process editing for the translation result can be greatly reduced. Therefore, it becomes possible to simplify the translation editing work, and to easily and efficiently obtain a translated sentence of an appropriate language expression.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of a machine translation device according to an embodiment of the present invention.

FIG. 2 is a diagram showing a configuration example of a keyboard in the same embodiment.

FIG. 3 is a diagram showing an example of a display screen in the embodiment.

FIG. 4 is a diagram showing a basic operation sequence in the embodiment.

FIG. 5 is a diagram showing a translation processing sequence in the embodiment.

FIG. 6 is a diagram showing an example of a recognition processing sequence of non-character data in the same embodiment.

FIG. 7 is a diagram showing an example of classification of input data in the same embodiment.

FIG. 8 is a diagram showing an example of a recognition conversion rule for non-character data in the same embodiment.

FIG. 9 is a diagram showing an output conversion example of non-character data in the same embodiment.

FIG. 10 is a diagram showing an example of output control for non-character data in the embodiment.

FIG. 11 is a diagram showing an example of handling non-character data in the same embodiment.

FIG. 12 is a diagram showing an output display example of a translation result in the example.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input part, 2 ... Original text storage part, 3 ... Non-character data recognition part, 4 ... Editing control part, 5 ... Translation part, 6 ... Translation dictionary part, 7
... translation storage unit, 8 ... non-character data expansion unit, 9 ... display unit,
10 ... Display control unit, 11 ... Printing unit.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor, Amano Makoto             1st Komukai Toshiba-cho, Sachi-ku, Kawasaki-shi, Kanagawa             Inside the Toshiba Research Institute (72) Inventor Tsutomu Kawada             1st Komukai Toshiba-cho, Sachi-ku, Kawasaki-shi, Kanagawa             Inside the Toshiba Research Institute

Claims (2)

[Claims] 1. A machine translation device for translating an input original sentence using information stored in a translation dictionary unit, and outputting a translated sentence obtained by this translation process into a character data string forming the original sentence. Machine translation, comprising: means for detecting non-character data relating to the format of the original sentence, and means for adding the non-character data detected by this means to a character data string forming the translated sentence. apparatus. 2. The non-character data is tab, line feed, space,
2. The machine translation device according to claim 1, wherein the machine translation device is at least a part of character position control data for controlling a character position such as indentation, and layout data for specifying a tab setting position, a chart area, a paper size, and the like.