JPH0628397A - Machine translation apparatus - Google Patents

Abstract

PURPOSE:To easily use a sentence including the pre-edit symbol of the pre-edit processing prior to the machine translation as a sentence by deleting prescribed symbols in a batch from an input original sentence after the pre-processing. CONSTITUTION:The pre-edit symbol is roughly classified into two types according to its form. All the pre-symbols of the 1st type have engings ''''. The places just after the symbols are not blank. On the other hand, the places just before these symbols are always blank. The 2nd type of the pre-edit symbol is used to punctuate sentences before and after the symbol in consideration by some meanings. Namely, these symbols are inserted between words. In the case of the 2nd type of the pre-edit symbol, the places just before and after these symbols inserted into the text are blank as a rule. With the use of these features, the processing of deleting the pre-edit symbols in a batch is performed by the program to be executed by main CPU.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation apparatus capable of improving the efficiency and accuracy of translation by pre-editing the original text by the user during translation, and more particularly to a machine translation apparatus Regarding improvement.

[0002]

2. Description of the Related Art A conventional general machine translation apparatus includes a keyboard for inputting source language text, a translation module for translating the input source language text into a target language, and a translation module. A main dictionary (main processing unit), a memory for storing a dictionary composed of a main dictionary and a user dictionary, a grammar rule, a tree structure conversion rule, and the like, and a main CPU (central processing unit) for controlling the operation of each of these parts. Including.

The source language text input from the keyboard is input to the translation module under the control of the main CPU. The input source language text is translated into the target language by the translation module by using the dictionary including the main dictionary and the user dictionary, the grammar rules, and the tree structure conversion rules.

In such a machine translation device, when parsing a part of speech obtained as a result of morphological analysis using a dictionary and grammatical rules, all possible syntactic structures are assembled. Therefore, a plurality of translation results are output according to a plurality of syntactic structures. Therefore, it is necessary for the user to select which of the plurality of translation results is the correct translation result, and it takes a long time for parsing to assemble all possible syntactic structures. there were. These problems arise because the structure of the original text is not simple, but is complicated and vague in interpretation.

In order to reduce such problems, it has been proposed to subject the input source text to pre-editing processing prior to translation. The "pre-editing symbol" is used for the pre-editing process. The pre-edit symbol is a special character / symbol string that is not used in ordinary text. By inserting several types of pre-edit symbols into the original sentence, you can specify the part of speech of a specific word according to the inserted position and the type of pre-edit symbol, and perform some types of syntactic analysis. You can specify a specific syntactic structure as a specific structure.

The pre-editing process is performed as described above, and the machine translation device translates based on the information given by the pre-editing process. As a result, the syntactic structure candidates are limited and the part of speech of the word is clearly specified, so that the parsing speed can be increased. Also, the accuracy of the translation result is improved.

[0007]

The pre-editing symbol is a symbol used only for machine translation. Therefore, once the translation is complete, the preedit symbol inserted in the original text is not needed. However, this pre-edit symbol remains in the original text after translation. Therefore, for example, when the translated original sentence is desired to be used as a sentence, the pre-edit symbol must be found one by one using a document editing function (for example, an editor) and manually deleted.

Therefore, an object of the present invention is to provide a machine translation device which makes it possible to easily use a sentence including a preedit symbol of a preedit process performed prior to machine translation as a sentence. is there.

[0009]

A machine translation apparatus according to the present invention includes a preprocessed input source text for storing a preprocessed input source text in which a predetermined symbol for assisting automatic translation is inserted as a preprocessing. A storage means, an automatic translation means for translating the preprocessed input source text into another language while using a predetermined symbol as a translation aid, and a preprocessed input source text stored in the preprocessed input source text storage means. Preprocessing symbol removing means for generating a sentence from which a predetermined symbol is removed.

[0010]

In the machine translation device according to the present invention, the input original sentence containing a predetermined symbol for preprocessing is translated into another language by the automatic translation means. Then, from the preprocessed input original sentence after the translation process, the preprocessed symbol removing means generates a sentence from which a predetermined symbol for preprocessing is removed. It is not necessary to manually correct the input original text after preprocessing by using the document editing function to remove the symbols for preprocessing as in the conventional case.

[0011]

1 is a block diagram of an embodiment of a machine translation apparatus of the present invention. Referring to FIG. 1, the machine translation device includes a main CPU 1 and a bus 8 to which the main CPU 1 is connected. Furthermore, the following devices are also connected to the bus 8. That is, the machine translation device further requires a main memory 2, a display device 3 including a CRT (cathode ray tube), a keyboard 4, a translation module 5 for performing translation processing, a translation result, and the like. If so, a print 7 for outputting is included. The translation module 5 is connected to a memory 6 which stores a dictionary used for translation, grammar rules, tree structure conversion rules, and the like.

The process of collectively deleting the pre-edit symbols described below is performed by a program executed by the main CPU 1. A buffer required for executing this program is allocated to the main memory 2.

The source language text input from the keyboard 4 is sent to the translation module 5 under the control of the main CPU 1. The translation module 5 translates the input text in the source language into the target language using the dictionary, the grammar rules, the tree structure conversion rules, and the like stored in the memory 6. The result is temporarily stored in the main memory 2. The translation result is also displayed on the display device 3. If necessary, the translation result is printed on a piece of paper by the printer 7. The main memory 2 is also used as a register memory and a display buffer during execution of various programs.

In this case, the input source language text is temporarily stored in the main memory 2 in order to perform the pre-editing process. Then, a pre-edit symbol is inserted into the text by using an editing processing function such as an editor program controlled by the main CPU 1. This insertion is performed using the keyboard 4 and the display device 3. When the translation module 5 translates the pre-edited text as described above, the translation efficiency is improved and the accuracy thereof is improved as described above.

Hereinafter, for the purpose of explanation, assuming that the text original sentence is an English sentence and translating it into Japanese, the operation of batch deletion of preedit symbols from the input original sentence after preediting according to the present invention will be described. explain. Note that this processing is performed by the main CPU
1 is performed by the program executed.

2 to 4 are flowcharts of the pre-editing symbol batch deletion process. Before describing this flowchart in detail, an example of pre-edit symbols used in the machine translation device shown in FIG. 1 will be described. Pre-edit symbols are roughly classified into two types according to their forms. The first type of preedit symbol is shown in FIG. The second type of preedit symbol is shown in FIG.
Shown in.

The pre-editing symbol shown in FIG. 5 is provided at the beginning of a word to modify the word in some sense. As the types of modification, there are a part-of-speech designation symbol, a sentence pattern designation symbol, a dictionary designation symbol, etc., as shown in FIG. The part-of-speech designation symbol is for designating the part-of-speech of the word following this symbol. The sentence pattern designating symbol is for designating the sentence pattern of the sentence following this symbol. The dictionary designation symbol is for designating a dictionary when performing a dictionary lookup process for the word following this symbol.

The symbols shown in FIG. 5 all have the suffix "". The symbol is followed by a word, which is immediately followed by a word, that is, there is no blank immediately after these symbols, whereas there is always a blank immediately before these symbols.

Referring to FIG. 6, the second type of pre-edit symbol is used to separate the symbol before and after the symbol in some sense. That is, these symbols are inserted between words. In the case of the present embodiment, the second type of pre-edit symbol is the same symbol that represents parentheses.
It is a stack of pieces. Although "//" is also included in the example shown in FIG. 6, these are also collectively referred to as parentheses.

In the case of this second type of pre-editing symbols, there is in principle a blank just before and immediately after these symbols inserted in the text. However, in some cases, these blanks may not be provided. That is, there may be cases where there are characters that form a word immediately before and after the preedit symbol.

An outline of the process of translating the input original text which has been subjected to the pre-editing process using the above-mentioned pre-editing symbols will be described. First, the original text to be translated is read. The read English sentences are stored in the buffer S as shown in FIG.
It is stored in 1. A part of the main memory 2 shown in FIG. 1 is assigned to this buffer S1.

In the English sentence stored in the buffer S1,
Pre-editing processing is performed. The pre-editing process is performed by the user. Of course, this pre-editing process is performed only when necessary.

FIG. 7 shows an example of English text which has been preedited. With reference to FIG. 7, the pre-editing symbol “v3” used in the first sentence 5 corresponds to the third of the sentence pattern designating symbols in FIG. 5. That is, this symbol is the third sentence pattern in which the entire first sentence is governed by the verb “called” of the first sentence shown in FIG. 7. Has been specified.

The pre-editing symbol "<used in the second sentence
<”And“ >> ”are a phrase designation start symbol and a phrase designation end symbol, respectively, as shown in FIG.
The part sandwiched between these two types of symbols is handled as a group in the syntactic analysis during translation.

As described above, the efficiency of translation processing is improved and the accuracy is improved by inserting the pre-editing symbol in the input original sentence. The present invention is characterized in that the preedit symbol can be easily deleted from the input text that has been preedited after translation. The process of deleting the pre-edit symbol will be described below.

Referring to FIG. 2, first, step S001.
Then, the sentence in which the preedit symbol is inserted is stored in the buffer S1. The stored result is as shown in FIG. 7. The input text stored in the buffer S1 is assumed to be immediately after translation in the present description. However, it does not have to be immediately after the translation, and when the original sentence of the sentence translated in the past is separately stored, the stored sentence is stored in the buffer S1 and the subsequent pre-edit symbol deletion processing is performed. Is also possible. It is also possible to temporarily store the untranslated sentence in the buffer S1, generate a translated sentence after the pre-editing process, and delete the pre-editing symbol for the remaining original sentence immediately after the translation process.

Then, in step S002, the buffer S1
The pointer indicating the current character position to be processed is set to 0. As a result, the character pointed to by the pointer becomes the first character "H" of the sentence stored in the buffer S1.

Further, in step S003, the buffer S
The contents of 2 are cleared. The buffer S2 is the buffer S
It is used to temporarily store a sentence in which the preedit symbol is removed from the input original sentence stored in 1. The main memory 2 shown in FIG. 1 is also allocated to this buffer S2.

At step S004, in the buffer S2,
The pointer indicating the character writing position is set to 0 so as to point to the beginning of the buffer S2.

In step S005, the buffer W is cleared. The buffer W is a work buffer that is used to temporarily store a character string cut out from the original sentence in units of words. The main memory 2 shown in FIG. 1 is also allocated to the buffer W.

In step S006, the character position pointer of the buffer W is set to 0. As a result, the writing position of the character in the buffer W becomes the beginning thereof.

Thus, the preparation for inspecting the original character string is completed.

The actual character string check is started from step S007. First, in step S007, the character pointed to by the character position pointer in buffer S1 is examined. If the character is blank or brackets, the process proceeds to step S021. Otherwise, the process proceeds to step S008.

When processing the sentence shown in FIG.
The character at position 0 that the character position pointer initially points to is “H”. Therefore, in this case, the processing is step S
It will proceed to 008.

In this embodiment, as described above, the term "brackets" refers to the nine types of symbol characters shown in FIG. Note that the symbol "/" is also included in this case.

The processing of steps S021 to S024 will be described later.

In step S008, the character at the position indicated by the character position pointer in buffer S1 is copied to the position indicated by the character position pointer in buffer W.

In the case of the example shown in FIG. 7, the buffer S
The character "H" of 1 is copied to the beginning position of the buffer W.

Subsequently, in step S009, 1 is added to the character position pointer of the buffer W.

Referring to FIG. 3, in step S010, the character pointed by the character position pointer in buffer S1 is " It is determined whether or not That is, in this case,
The character copied from buffer S1 to buffer W in step S008 is " It is determined whether or not If the determination result is YES, the process proceeds to step S020; otherwise, the process proceeds to step S011. Step S0
The processing of 20 or less will be described later. letter" "Is a character always used at the end of the first type of pre-edit symbol shown in FIG. Therefore, when this character is detected in the character string of the buffer S1, it is highly likely that the pre-edit symbol exists at that position.

In the case of the example shown in FIG. 7, the buffer S1
Since the character pointed by the current character position pointer of is “H”, the process proceeds to step S011.

In step S011, the character position pointer in the buffer S1 is advanced by 1. As a result, the buffer S
The character pointed by the character position pointer 1 is “e”.

In step S012, it is determined whether or not a character exists after the character position pointer in the buffer S1. If it is determined that the character does not exist, the pre-edit symbol deletion function ends. That is, the process proceeds to step S027, the process of copying the character string in the buffer W to the current pointer 1 or later in the buffer S2 is performed, and the entire process ends. If there is the next character in the buffer S1, the process proceeds to step S013.

In the case of the English sentence shown in FIG. 7, the buffer S1
The character "e" is present at the position pointed to by the character position pointer. Therefore, the result of the determination in step S012 is YES, and the process proceeds to step S013.

In step S013, it is determined whether the character pointed by the character position pointer in the buffer S1 is blank or parenthesized. If it is blank or brackets, the process returns to step S014, otherwise the process returns to step S008 in FIG.

If the character pointed to by the character position pointer in buffer S1 is blank, then that part is considered to be a word break. If the characters are parentheses, the pre-edit symbol shown in FIG. 6 may be present at that location. If so, that part is also a word break.

In the case of the example sentence shown in FIG. 7, the character pointed by the character position pointer is "e". Therefore, the answer to the determination in step S013 is NO, and the process returns to step S008 in FIG.

Similarly, for the character "e" pointed to by the character position pointer in the buffer S1, step S in FIG.
008, S009, steps S010, S01 of FIG.
1 and S012 are sequentially executed. As a result, the character "e" is added to the buffer W.

Consider a case where the process proceeds to step S013. At this point, the character string "He" is stored in the buffer W as shown in FIG. Buffer S
The character pointed to by the character position pointer 1 is the blank position next to "e" with reference to FIG. Therefore, as a result of the determination in step S013, the processing is step S014.
Proceed to.

In step S014, a determination is made as to whether the character pointed to by the character position pointer in buffer S1 is blank. If it is blank, the process is shown in FIG.
Processing proceeds to step S015, and in other cases (that is, parentheses), the processing proceeds to step S025.

In the case of the example sentence shown in FIG. 7, the buffer S
The character pointed to by the 1 character position pointer is blank. Therefore, the process proceeds to step S015 (see FIG. 4).

When the process proceeds to step S015,
The character position pointer in buffer S1 has completely passed one word. During this time, each character of the word is buffer W
Have been sequentially copied to. Therefore, the character string stored in the buffer W is one complete word. In step S015, a process of copying the character string stored in this buffer W to a position after the character position pointer of buffer S2 is performed. As a result, the word "He" is copied to the top of the inside of the buffer S2 as shown in FIG.

In the next step S016, the buffer S2
A process of advancing the character position pointer by the number of characters in the buffer W is performed. This causes the character position pointer to point to the next position of the copied character.

In the case of the processing of the example sentence shown in FIG. 7, the word "He" currently being processed is a two-character word. Therefore, the character position pointer of the buffer S2 is advanced by 2 by the processing of S016. As a result, the character position pointer of the buffer S2 shown in FIG. 10 points to the next position of the character "e".

In the next step S017, the buffer S2
The process of setting a blank at the position pointed to by the character position pointer is performed. This is a process for generating a word break in the buffer S2.

In step S018, a process of advancing the pointer of buffer S2 by the amount of the inserted blank is performed. That is, the character position pointer of the buffer S2 is incremented by 1. Similarly, in step S019, the buffer S1
The character position pointer of is also incremented by 1. Step S019
The post-processing returns to step S005 in FIG.

The buffer W is cleared in step S005, and the character position pointer of the buffer W is set to 0 in step S006. The reason why the buffer W is cleared in this way is that the process of copying the word “He” being processed into the buffer S2 is completed, and the word stored in the buffer W is no longer needed.

In the above, the procedure of performing copying from the buffer S1 to the buffer S2 for a word without a preedit symbol has been described by taking the word "He" in the sentence stored in the buffer S1 as an example. As is clear from the above description,
Words without a preedit symbol are copied as is from buffer S1 to buffer S2 as shown in FIG.

In the first type of pre-edit symbol shown in FIG.
Some parts are processed as follows. Shown in Figure 7
Word "v3" next to "He" in the example sentence "called"
Is the first type of preedit symbol "v3 "
Is attached. Hereinafter, "v3 for "called"
As an example, the first type of pre-edit symbol shown in FIG.
The process for a word added with will be described.

When the copying of "He" is completed, the processing is in step S007. The character position pointer of the buffer S1 is "v3 "v" of "called". No character exists inside the buffer W. The character position pointer of the buffer W points to the beginning of the buffer W. The contents of the buffer S2 are as shown in FIG. There is one blank character after the word “He”. The character position pointer in the buffer S2 points to the position next to the blank.

Referring to FIG. 2, in step S007, the character pointed to by the current character position pointer in buffer S1 is not blank. Therefore, the process proceeds to step S008. Hereinafter, the character string “v3” is copied to the buffer W by repeating the loop of steps S008, S009, steps S010, S011, S012, S013 of FIG. 3, and step S008 of FIG. 2 twice. Buffer S1
The character position pointer of is "v3" Is pointing to.

Then, by the processing of steps S008 and S009, the character " Is added to the buffer W. As a result, the contents stored in the buffer W are as shown in FIG. Referring to FIG. 3, in step S010, the character pointed by the character position pointer in buffer S1 is " Since the answer is YES, the process proceeds to S020.

At S020, a determination is made as to whether the contents of buffer W match any of the pre-edit symbols shown in FIG. If none of them match, the process proceeds to step S011. In this case, the character " This is because "" is determined to be a part of a word that is unrelated to the preedit symbol. If the character string in the buffer W matches any of the pre-edit symbols, the process proceeds to step S.
Proceed to 028.

In the case of the example sentence shown in FIG. 7, the content of the buffer W is the character string "v3" as shown in FIG. It is. As can be seen from FIG. 5, this is a kind of preedit symbol. Therefore, the result of the determination in step S020 is YES, and the process proceeds to step S028.

In step S028, the character position pointer in the buffer S1 is advanced by 1. This makes the buffer S
The character indicated by the character position pointer of 1 is "c". The post-processing of step S028 returns to step S005.

The buffer W is cleared again in steps S005 and S006. Since the word stored in the buffer W is the pre-edit symbol, it is not necessary to copy it to the buffer S2, and it is necessary to further advance the process to the word next to this pre-edit symbol.

Hereinafter, referring again to FIG. 7, the buffer S1
The character pointed to by the current character position pointer of is "c". Therefore, referring to FIG. 2, the process proceeds to step S0.
It progresses from 07 to S008. Further steps S008-S
By repeating the loop of 013 → S008 six times, the character string “called” in the buffer S1 is copied to the buffer W. At this time, the character position pointer of the buffer S1 points to the space next to "called".

Further, the processing proceeds to steps S008 to S015. As a result, the word "called" stored in the buffer W is added to the buffer S2 as shown in FIG.

Further, a blank character is added to the buffer S2 by the processing of steps S016 to S018. As a result, the contents of the buffer S2 become as shown in FIG.

Further, by the processing of step S019 shown in FIG. 4, the character position pointer of the buffer S1 is "c".
It comes to point to the character "h" next to the blank after "alled". Further, steps S005 and S0 shown in FIG.
The processing returns to 06, and the buffer W is cleared.

By the above processing, the data is stored in the buffer S.
The word with a pre-edit symbol "v3 Called "is before
The word "called" with the editing symbol removed is displayed as a bar.
Is copied to the Saffa S2. The contents of the buffer S2 are shown in FIG.
3 is shown.

In the case of the example sentence shown in FIG. 7, "bough
There is no preedit symbol in the part up to t. Therefore, the processing of each word during this period is performed by the above-mentioned "He".
It is performed in the same manner as the processing of the word. The word "bough
The contents stored in the buffer S2 are as shown in FIG.

With reference to FIG. 7, there is a preedit symbol "<<" between the word "boot" and the next word "necklaces". Hereafter, the previous editing symbol "<
The process for the second type of pre-edit symbol shown in FIG. 6 will be described by taking the process of "<" as an example.

The word "bough" in the example sentence shown in FIG.
When the processing up to t ”is completed, the processing steps are as shown in FIG.
Is step S007. The character position pointer of the buffer S1 points to "<" on the left side of the preedit symbol "<<". The contents of buffer W have been cleared,
The character position pointer points to the beginning of the buffer W. The contents of the buffer S2 are as shown in FIG. In FIG. 14, there is one blank character after the word "boot". The character position pointer points to the position after the blank.

In the processing of the pre-edit symbol "<<", the result of the determination is YES in step S007 (see FIG. 2). Therefore, the process proceeds to step S021.

In step S021, it is determined whether or not the character pointed by the pointer of the buffer S1 is blank. If it is blank, the process proceeds to step S022. If it is not blank, that is, if it is in parentheses, the process proceeds to step S023.

In the current state, the character pointed to by the pointer is "<", and the process proceeds to step S023.

In step S023, it is determined whether or not the next character is the same character, that is, "<" in the present case. If not the same, the process proceeds to step S008, and if the same, the process proceeds to step S024. The process of step S023 is performed for the following reason. If the same parenthesis is used as a preedit symbol, the same two parentheses are used. If brackets are present alone, they are considered to be just brackets as symbols in the text. Therefore, in step S023, therefore, the process proceeds to step S024 only when the same pair of parentheses occurs consecutively, and otherwise proceeds to step S008 as in the normal case.

In step S024, the character position pointer in the buffer S1 is advanced by 2. This is because, in the case of the pre-edit symbol using parentheses, the two parentheses are used as one set. As a result of the processing in step S024, the character position pointer of the buffer S1 points to the blank space next to “<”. The post-processing of step S024 returns to step S007.

At step S007, 1 of the buffer S1
The character pointed to by the pointer is checked again. If it is blank or brackets, the process proceeds to step S021, and otherwise, the process proceeds to step S008.

In the case of the example sentence shown in FIG. 7, the character pointed to by the pointer is blank. Therefore, in this case, the process proceeds to step S021, and further proceeds to step S022. Then, in step S022, the pointer of the buffer S1 is 1
Can be advanced. As a result, the character position pointer of buffer S1 points to the first character "n" of the word "necklaces".

By the above processing, the character position pointer of the buffer S1 passes through the pre-edit symbol "<<" without copying the pre-edit symbol "<<" existing in the buffer S1 to the buffer S2. . Previous edit symbol "<
<”Is never copied to the buffer S2.

In the case of the example sentence shown in FIG. 7, the next word "n"
The processing of “ecklaces” is performed in the same manner as the processing of the word “He”. As a result, the contents stored in the buffer S2 are as shown in FIG. The word "b" in buffer S1
The pre-edit symbol "<<" existing between "out" and "necklaces" does not exist in the sentence of the buffer S2.

Hereinafter, the processing after the word "of" shown in FIG. 7 can be performed by sequentially applying the above method.

Consider the time when the processing of the last word "silver." Is completed. The character position pointer of the buffer S1 points to “.” At the end of “silver.”. The character string “silver.” Is stored in the buffer W. At this point, the process is step S009 in FIG. Steps S009 to S010 and S in FIG.
The processing proceeds from 011 and S012. In the determination in step S012, since the character next to "." Is no longer present, the process proceeds to step S027.

In step S027, the content "silver." Of the buffer W is added to the buffer S2 as described above.
Add to. This completes the contents of buffer S2,
The whole process ends.

When the above processing is completed, the buffer S2
The contents stored in are as shown in FIG. Of the example sentences shown in FIG. 7, the pre-edit symbols shown in FIGS. 5 and 6 have been completely removed.

In the flow chart shown in FIG. 3, the processes of steps S014, S025 and S026 are as follows.
Steps S021, S023, S024 shown in FIG.
It is exactly the same as the processing of. In this way, step S014
The following processing is provided for the following reason. As described above, in the case of the first type of pre-edit symbol shown in FIG. 6, there may be no blank before and after it. In such a case, steps S021, S023, S
The processing of 024 cannot handle this. Therefore, from the time of step S013, the second time shown in FIG.
The routines of steps S014, S025, and S026 are provided as a procedure for processing the pre-edit symbols of types.

As described above, in the machine translation device of the present invention,
Of the input text that has been preedited, the original text from which only the preedit symbols have been removed is automatically generated. There is no need for the user to manually search for the pre-edit symbols one by one using the document editing function and delete them manually as in the prior art. Therefore, the translated document can be efficiently used.

[0090]

As described above, in the machine translation apparatus of the present invention, the pre-editing symbol removing means generates the pre-edited input original sentence by removing a predetermined symbol for pre-editing. It is not necessary to manually search and remove the preedit symbols one by one.

As a result, the sentence of the input original sentence which has been preprocessed for translation can be easily used as a sentence.

[Brief description of drawings]

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

FIG. 2 is a first half flowchart of a pre-editing symbol removal process.

FIG. 3 is an intermediate flowchart of a pre-editing symbol removal process.

FIG. 4 is a second half flowchart of the pre-editing symbol removal processing.

FIG. 5 is a diagram showing a first type of pre-edit symbol in a table format.

FIG. 6 is a diagram showing a second type of pre-edit symbol in a table format.

FIG. 7 is a schematic diagram showing stored contents of a buffer S1.

FIG. 8 is a schematic diagram showing brackets.

FIG. 9 is a schematic diagram showing the contents of a buffer W.

FIG. 10 is a schematic diagram of a buffer S2.

FIG. 11 is a schematic diagram of a buffer W.

FIG. 12 is a schematic diagram of a buffer W.

FIG. 13 is a schematic diagram of a buffer S2.

FIG. 14 is a schematic diagram of a buffer S2.

FIG. 15 is a schematic diagram of a buffer S2.

FIG. 16 is a schematic diagram of a buffer S2.

[Explanation of symbols]

 1 main CPU 2 main memory 3 display device 4 keyboard 5 translation module 6 memory 7 printer

Claims (1)

[Claims] 1. A preprocessed input source sentence storage unit for storing a preprocessed input source sentence in which a predetermined symbol for assisting automatic translation is inserted as a preprocess, and the preprocessed input source sentence is stored in the storage unit. Automatic translation means for translating into a different language while using a predetermined symbol as an aid of translation, and a sentence obtained by removing the predetermined symbol from the preprocessed input source text stored in the preprocessed input source text storage means. And a preprocessing symbol removing means for generating the.