JPH1115825A - Machine translation system and computer-readable recording medium recording machine translation processing program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To obtain a translation result better than the translation of sentences simply delimited in accordance with the numbers of characters by delimiting an original language document on a suitable position as a delimiter. SOLUTION: When an original language document to be translated is delimited by limitation to the number of characters in one sentence of the original language document which is set up on the system side at the time of reading out the original language document, morpheme analysis for the undelimited document is executed, the information of a table 47 describing plural positions to be delimited is referred to and a position suited as a delimiter is retrieved from the morpheme analysis result. When a position to be newly delimited exists, a character string up to the position is delimited as one sentence and stored in an input buffer area 31. Then a character string after the new delimited position is added to the head of a character string after a position delimited by a Japanese reading part 45 and the processing of a delimited position changing part 46 is allowed to end.

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 and a computer-readable recording medium on which a machine translation processing program is recorded.

[0002]

2. Description of the Related Art Conventionally, another language for which a source language is desired, that is,
In a machine translation device that translates into a target language, when a document in a source language to be translated already exists as document data, a function of reading the document data for translation is provided. With this function to read document data,
You can read each sentence separately. For example, when the source language is Japanese, a Japanese document is read by separating each period. In this case, the subsequent translation processing is performed one sentence at a time, so that the translation processing is not complicated.
It is more efficient than translating multiple sentences at once.

Further, if the length of one sentence is very long, the translation process becomes complicated and the translation result tends to be deteriorated. Therefore, the number of characters in one source language sentence may be limited.
For example, if one sentence of Japanese as a source language has 100 characters, but if the number of characters in one sentence of the source language of the system is limited to 50 characters, when reading a Japanese sentence, 2
It will be divided into sentences. As described above, in general, a translation process is performed by dividing a document into units suitable for the translation process.

[0004]

However, if the text is divided according to the number of characters as described above, a portion that is not actually suitable as a delimiter may be delimited. For example, when the source language is Japanese, the document to read is, "I heard that her condition was not so good, but when I went to the hospital yesterday, my complexion looked very good and I seemed fine, I was relieved. "
This sentence consists of 64 characters. If the number of characters in one sentence is limited to 50 in the system, it is read in two sentences. As a result, the first sentence was 2
The sentence reads, "I was relieved because I was out."

[0005] This break position is not appropriate for Japanese, and the result is reflected in the translation result. That is, when a Japanese sentence is decomposed for each part of speech registered in the dictionary for translation, the last part of the first sentence is divided into the nouns "Genki" and "So". “So” is an unregistered word that is not registered in the dictionary. Thus, even if a sentence segmented at an incorrect position is morphologically analyzed, a correct morphological analysis result cannot be obtained, and even if the result is translated, a correct translation result cannot be obtained.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-described problems. When reading source language document data, when one sentence is long and the sentences are separated due to the limitation of the number of characters, the separated portion is used. Machine translation device and machine translation process that can obtain better translation results than simply translating a sentence delimited by the number of characters by morphologically analyzing It is an object of the present invention to provide a computer-readable recording medium on which a program is recorded.

[0007]

In order to achieve this object, a machine translation apparatus according to claim 1 of the present invention converts a source language sentence given as a source language into another language sentence as a destination language. A translation dictionary that stores information necessary for translation,
First morphological analysis means for morphologically analyzing the given source language sentence with reference to the information stored in the translation dictionary; and converting the source language sentence morphologically analyzed by the first morphological analysis means into the translation A translation processing unit that stores grammatical rules for translating into a target language while referring to information stored in a dictionary, and reads source language document data prepared in advance as original text while dividing the data into units suitable for translation. A reading unit, wherein at least one of the conditions delimited by the reading unit is limited to the number of characters.

[0008] In particular, when the document is read by the reading means, if the document is divided due to the limitation of the number of characters, the second morphological analysis means for performing morphological analysis up to the divided part; Means for determining a position suitable as a break from the morphological analysis result by the means, a break position storing means describing a place to be regarded as a break, a morphological analysis result by the second morphological analysis means, And a break position changing means for changing a break position based on the information, wherein the document read by the reading means is re-separated by the break position changed by the break position changing means.

Therefore, according to the machine translation apparatus of the present invention, when the document is read by the reading means, if the document is separated by the character number restriction, the second morphological analysis means will use the morpheme up to the separated part. To analyze.
Then, based on the result of the morphological analysis by the second morphological analysis unit and the information of the delimiter position storage unit, the delimiter position changing unit changes the delimiter position. In this way, the document read by the reading unit is
The division is performed again at the division position changed by the division position changing means.

[0010] The machine translation device according to claim 2 is
The present invention is characterized in that the character string after the part newly separated by the delimiter position changing means is added to the head of the character string after the part delimited by the reading means. Therefore, by adding the character string after the delimiter determined from the result of the morphological analysis to the beginning of the character string after the delimiter determined by the number of characters, the Japanese character after the delimiter also has an appropriate shape, which is better. Translation results can be obtained.

The machine translation device according to claim 3 is
It is characterized by having a break position setting means which can set the information of the break position storage means by a user. Therefore, the user can set the delimiter position based on the morphological analysis result by himself / herself, and obtain a reading result that meets the user's desire.

Further, the machine translation device according to claim 4 is
It is characterized in that the result of the morphological analysis by the second morphological analysis unit is regarded as the first morphological analysis result by the first morphological analysis unit, and the processing is performed by the translation processing unit. Therefore, since the morphological analysis result in the delimiter position changing unit is stored and used as the morphological analysis result at the time of performing translation, it is possible to prevent performing the morphological analysis twice for the same sentence, and to efficiently perform the translation operation. Becomes possible.

[0013] The machine translation device according to claim 5 is
It is characterized in that the priorities described in the delimiter position storage means can be set. Therefore, by providing a priority order to the delimiter positions described in the delimiter position storage means, when a plurality of locations are searched, it is possible to delimit at a position more suitable as a delimiter.

A computer-readable recording medium storing the machine translation processing program according to claim 6 reads the source language document data given as the source language as an original while dividing the data into units suitable for translation, and A reading program in which at least one of the delimiting conditions is limited in the number of characters, a first morphological analysis program for morphologically analyzing a source language sentence with reference to information stored in a translation dictionary, and a first morphological analysis program A translation program for translating a morphologically analyzed source language sentence into another language as a target language while referring to information stored in the translation dictionary.

In particular, when the source language document data is divided by the character number limitation, a second morphological analysis program for morphologically analyzing up to the divided part,
A delimiter position storage program for describing a position regarded as a delimiter in order to determine a position suitable as a delimiter from a morphological analysis result by the second morphological analysis program;
A delimiter position changing program for changing a delimiter position based on a result of the morphological analysis by the second morphological analysis program and information of the delimiter position storage program; It is characterized in that a break is performed again at a break position changed by the position changing program.

Therefore, by executing the program using the recording medium, when the document is read by the reading program, if the document is divided by the number of characters, the location where the second morphological analysis program is divided Morphological analysis is performed for Then, based on the result of the morphological analysis by the second morphological analysis program and the information of the delimiter position storage program, the delimiter position change program changes the delimiter position. In this way, the document read by the reading program is re-separated at the delimiter position changed by the delimiter position changing program.

According to a seventh aspect of the present invention, there is provided a computer-readable recording medium storing the machine translation processing program according to the seventh aspect, wherein the delimiter position change program causes the read program to read a character string after a new delimiter. It is added to the beginning of the character string after the delimited part. Therefore, by executing the program using the recording medium, the character string after the delimiter determined from the result of the morphological analysis by the second morphological analysis program is replaced with the character string after the delimiter determined by the number of characters. By adding it to the beginning, the Japanese language after the delimiter is also in an appropriate form, and a better translation result can be obtained.

[0018] A computer-readable recording medium storing the machine translation processing program according to the present invention has a delimiter position setting program that allows a user to set information of the delimiter position storage program. . Therefore, by executing the program using the recording medium, the user can set the break position based on the result of the morphological analysis by the second morphological analysis program, and the reading result that meets the user's desire can be set. Can be obtained.

A computer-readable recording medium on which the machine translation processing program according to claim 9 is recorded, wherein the translation program performs a morphological analysis by the second morphological analysis program on the basis of the first morphological analysis result. The translation processing is performed by regarding the result of the morphological analysis by the analysis program. Therefore, by executing the program using the recording medium, the morphological analysis result by the second morphological analysis program in the break position change program is stored and used as the morphological analysis result at the time of performing translation. Performing a morphological analysis twice for the same sentence can be prevented, and efficient translation can be performed.

Further, a computer-readable recording medium recording the machine translation processing program according to claim 10 is characterized in that the information described by the delimiter position storage program can set a priority for delimiting. . Therefore, by executing a program using the recording medium, a priority order is set for the delimiter positions, so that when a plurality of locations are searched, it is possible to delimit at a position more suitable as a delimiter.

[0021]

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

Here, an example of a Japanese-English machine translator for translating a Japanese sentence as a source language into an English sentence as a target language will be described.

First, the configuration of the entire Japanese-English machine translation apparatus according to the present embodiment will be described with reference to FIG.

As shown in FIG. 1, the Japanese-to-English machine translation apparatus of the present embodiment comprises an input device 10, an output device 50, a CPU 2
0, a RAM 30, a ROM 40, an external storage device 60, and the like.

The CPU 20 is a central processing unit for controlling the entire apparatus, and includes an input device 10, an output device 50,
It is connected to the AM 30, the ROM 40, the external storage device 60, and the like.

The input device 10 comprises a keyboard or the like,
Enter a Japanese sentence or instruction to be translated.

The output device 50 comprises a CRT or the like, and displays an input Japanese sentence and a translation result.

The RAM 30 includes an input buffer area 31 for storing an input Japanese sentence, an output buffer area 32 for storing a translated English sentence, and a work area 33.

ROM 40 for storing programs, dictionaries, etc.
Is a control program 41 for performing various processes, a morphological analysis unit 42 as a first morphological analysis unit and a second morphological analysis unit for morphologically analyzing a given Japanese sentence, and a system for translating. A dictionary unit 43 as a translation dictionary, which stores a dictionary prepared in advance, a grammar unit 44 as a translation processing unit, which stores grammar rules for advancing translation while referring to dictionary information, and a delimiter position storage unit. Table 47.

The control program 41 includes a Japanese reading section 45 as reading means and a break position changing section 46 as break position changing means.

Further, in the external storage device 60, a technical term dictionary prepared by the user as desired, a user dictionary created by the user to reflect the translation result, and the like are stored.

FIG. 2 is a flowchart showing an outline of processing when reading Japanese document data to be translated in the Japanese-English translation apparatus of the present embodiment. At the time of reading Japanese document data in the present embodiment, the Japanese reading section 45 separates the document into punctuation marks, and when one sentence is composed of 50 characters or more, reads the sentence with 50 characters. It is assumed that it is configured to be

Here, the flow of processing in this embodiment when one sentence reads Japanese having 50 characters or more will be described with reference to the flowchart of FIG.

When an instruction to read document data is issued (Step 201; hereinafter, step is abbreviated as S), a counter for counting the number of characters is set to an initial value of 0 (S202). Thereafter, when the Japanese reading section 45 reads one character (S203), the value of the counter is incremented by one (S204).

The read character is not the 50th character (S2
05: NO), when it is not a period (S206: NO),
If there are no more characters behind (S207: NO), the reading process ends. If there is still a character behind (S20
7: YES), the next character is read (S203). If the character is a punctuation mark (S206: YES), up to the punctuation mark is divided as one sentence and stored in the input buffer area 31 (S208).

Further, if there is no character behind (S
(209: NO), the reading process is terminated, and if there is a character behind (S209: YES), the value of the counter is set to the initial value 0 (S202), and the character is read while counting the number of characters one by one again (S202). S203, S204).

When the read character is the 50th character (S20
5: YES), if the character is divided as one sentence (S
210), a process of checking the break position and re-separating by the break position changing unit 46 is performed (S211).

As shown in FIG. 5 (a), the Japanese sentence read, "I heard that her condition was not so good, but when I went to the hospital yesterday, she looked very well and looked fine. Therefore, I was relieved. "
Up to a sentence. This state is schematically shown in FIG.

FIG. 3 is a flowchart showing the flow of the processing in the break position changing unit 46. With reference to FIG. 3, the processing in the break position changing unit 46 will be described below.

In the delimiter position changing unit 46, first, the morphological analysis is performed by the morphological analysis unit 42 on the delimited 50-character Japanese sentence (S31). FIG. 5C schematically shows the result of the morphological analysis. "I'm fine" should be the stem of the adjective "Genki" and the final form of the auxiliary verb "Soda", but I can't get the correct morphological analysis because the sentence ends with "so". “So” is morphologically analyzed as unregistered and not registered in the dictionary.

Thereafter, based on the result of the morphological analysis, a delimiter position described in the table 47 is searched (S3).
2).

FIG. 4 is a diagram schematically showing information on the break positions stored in the table 47. As shown in FIG.

As shown in FIG. 4, the table 47 describes one or more positions to be delimited. The position stored in the table 47 is searched from the result of the morphological analysis of FIG. 5C, and if there is a part to be newly delimited (S33: Y
ES), the part up to that point is divided as one sentence and stored in the input buffer area 31 (S34).

As shown in FIG. 4, information indicating that the adjective conjunctive form is used as a delimiter is stored in the table 47.
Since "good" is a conjunctive form of an adjective in (c), this position is set as a new delimiter position to separate sentences.

The determination of the break position is performed by using the table 47
Is searched from the result of the morphological analysis, and if a plurality of positions are found, the position may be divided at the rearmost position, or the position of the delimiter stored in the table 47 may be used. May be set in the order of priority, and when a plurality of locations are searched, the sections may be separated at a position with a higher priority order.

Thereafter, the character string after the new delimiter is added to the beginning of the character string after the delimiter by the Japanese reading unit 45 (S35), and the processing in the delimiter position changing unit 46 is completed.

When the delimiter position described in the table 47 is not searched from the morphological analysis result (S3
3: NO), the processing of the break position changing unit 46 ends without setting a new break.

When the above processing is completed (S211), and there are more characters behind (S209: YES), the counter is returned to the initial value 0 again (S202), and the characters are read one by one while counting the number of characters (S203, S203). 204), the processing shown in the flowchart of FIG. 2 continues.

FIG. 5D schematically shows a state in which reading has been completed. The reading has been completed in a state where it is separated at a position different from the position separated by the character number limit.

When the reading process is completed, a translation process is executed. FIG. 6 is a flowchart showing the flow of the translation process. When the execution of the translation is instructed, the morphological analysis (S61) is performed for each sentence as shown in FIG.
Japanese analysis (S62), Japanese-to-English conversion (S63), and English generation (S64) are performed, and an English sentence is generated as a translation result. However, since this processing is a known technique, detailed description is omitted. .

Here, it is conceivable to use the result of the morphological analysis performed in S31 of FIG. 3 as the result of the morphological analysis in S61. That is, in S34 of FIG. 3, when the delimiter position is determined from the morphological analysis result and is stored in the input buffer area 31 as one sentence, the morphological analysis result is also stored, and the result is used as the morphological analysis result at the time of translation execution. This eliminates the need to perform morphological analysis on the same sentence twice.

The present invention is not limited to the above-described embodiment, and various changes can be made without departing from the gist of the present invention.

For example, the division is made by 50 characters, but may be made by another number such as 60 characters or 70 characters.

[0054]

As described above, according to the machine translation apparatus of the first aspect of the present invention, when a sentence is divided by the number of characters when reading a Japanese sentence, morphological analysis is performed. However, by re-segmenting the sentence at a place suitable as a sentence break, a correct morphological analysis result can be used for the subsequent translation processing, and a better translation result can be obtained.

According to the machine translation device of the second aspect, the character string after the delimiter determined from the result of the morphological analysis is added to the head of the character string after the delimiter determined by the number of characters. Also, the Japanese language behind the delimiter is also in an appropriate form, and a better translation result can be obtained.

Further, according to the machine translation device of the third aspect, the user himself / herself sets the delimiter position based on the result of the morphological analysis, so that a read result more desired by the user can be obtained.

According to the machine translation apparatus of the present invention, the result of the morphological analysis by the delimiter position changing means is stored and used as the result of the morphological analysis at the time of executing the translation. Performing morphological analysis can be prevented, and efficient translation work can be performed.

Further, according to the machine translation device of the fifth aspect, by providing a priority order to the delimiter positions described in the delimiter position storage means, when a plurality of locations are searched, a position more suitable as a delimiter is provided. Can be separated by

According to a computer-readable recording medium on which the machine translation processing program according to claim 6 is recorded, each program is suitable for a machine translation apparatus from various media such as a floppy disk and a CD-ROM. Stored in a recording medium. Then, by executing a program using this recording medium, when sentences are delimited by the number of characters when reading Japanese sentences, morphological analysis is performed, and the sentence is re-separated at a place suitable as a sentence delimiter. The correct morphological analysis result can be used for the translation processing of, and a better translation result can be obtained.

According to a computer-readable recording medium on which the machine translation processing program according to claim 7 is recorded, each program is suitable for a machine translation apparatus from various media such as a floppy disk and a CD-ROM. Stored in a recording medium. Then, by executing the program using this recording medium, the character string after the delimitation point determined from the result of the morphological analysis is added to the beginning of the character string after the delimitation point determined by the number of characters. The Japanese behind is also in an appropriate form, and better translation results can be obtained.

According to a computer-readable recording medium on which the machine translation processing program according to claim 8 is recorded, each program can be stored on a floppy disk or a CD.
-It can be provided by being stored in a recording medium suitable for a machine translation device from various media such as a ROM. Then, by executing a program using this recording medium,
The user can set the delimiter position based on the morphological analysis result by himself, and can obtain a read result more desired by the user.

According to a computer-readable recording medium having recorded thereon the machine translation processing program according to the ninth aspect, each program is suitable for a machine translation apparatus from various media such as a floppy disk and a CD-ROM. Stored in a recording medium. Then, by executing the program using this recording medium, the morphological analysis result in the delimiter position changing means is stored and used as the morphological analysis result at the time of performing the translation. Can be prevented, and efficient translation work can be performed.

According to a computer-readable recording medium on which the machine translation processing program according to claim 10 is recorded, each program can be stored on a floppy disk or
It can be provided by being stored in a recording medium suitable for a machine translation device from various media such as a D-ROM. And
By executing the program using this recording medium, priority is given to the delimiter positions described in the delimiter position storage program, and when a plurality of locations are searched, the delimiters can be partitioned at positions more suitable as delimiters.

[Brief description of the drawings]

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

FIG. 2 is a flowchart illustrating a flow of processing for reading a Japanese document according to the present embodiment.

FIG. 3 is a flowchart illustrating a flow of processing in a break position changing unit according to the present embodiment.

FIG. 4 is a diagram schematically illustrating information of a table indicating a break position according to the present embodiment.

FIG. 5 is a diagram schematically showing a state of Japanese at each stage of the present embodiment.

FIG. 6 is a flowchart illustrating a flow of a translation process according to the present embodiment.

[Explanation of symbols]

 Reference Signs List 30 RAM 31 Input buffer area 32 Output buffer area 40 ROM 42 Morphological analysis part 43 Dictionary part 44 Grammar part 45 Japanese reading part 46 Delimiter position change part 47 Table

Claims (10)

[Claims] 1. A translation dictionary storing information necessary for translating a source language sentence given as a source language into another language sentence as a target language, and a given source language sentence is stored in the translation dictionary. First morphological analysis means for performing a morphological analysis while referring to the stored information; and a source language sentence morphologically analyzed by the first morphological analysis means in a target language while referring to the information stored in the translation dictionary. A translation processing unit that stores grammar rules for translating the source language document data, and reading means for reading source language document data prepared in advance as original text while dividing the data into units suitable for translation. In a machine translation apparatus in which at least one of the conditions to be set is limited to the number of characters, when the reading unit reads the document data, the document is divided by the limited number of characters. In this case, a second morphological analysis unit that performs morphological analysis up to the divided portion, and a portion that is regarded as a break is described in order to determine a position suitable as a break from the morphological analysis result by the second morphological analyzer. A delimiter position storage unit; and a delimiter position changing unit that changes a delimiter position based on a result of the morphological analysis by the second morphological analysis unit and information of the delimiter position storage unit. A machine translation apparatus for re-separating a document at a break position changed by the break position changing means. 2. The apparatus according to claim 1, wherein a character string after a part newly separated by said separation position changing means is added to a head of a character string after a part separated by said reading means. Item 10. The machine translation device according to Item 1. 3. The machine translation apparatus according to claim 1, further comprising a break position setting unit that allows a user to set information in the break position storage unit. 4. A configuration in which a result of the morphological analysis by the second morphological analysis unit is regarded as a morphological analysis result by the first morphological analysis unit, and the processing by the translation processing unit is performed. The machine translation device according to claim 1, wherein 5. The machine translation apparatus according to claim 1, wherein the information described in the delimiter position storage means can be set so as to set a priority order for the delimiter. 6. A reading program in which source language document data given as a source language is read as an original while being divided into units suitable for translation, and at least one of the delimiting conditions is limited in the number of characters, and is stored in a translation dictionary. A first morphological analysis program for morphologically analyzing the source language sentence while referring to the information obtained from the source language sentence, and a source language sentence morphologically analyzed by the first morphological analysis program while referring to the information stored in the translation dictionary. In a computer-readable recording medium having a translation program for translating into another language as a target language, when the source language document data is divided by the character number limit, a morphological analysis is performed up to the divided portion. 2 and a morphological analysis result obtained by the second morphological analysis program. In order to determine a suitable position as a partition, a partition position storage program describing a portion to be regarded as a partition, a morphological analysis result by the second morphological analysis program, and information on the partition position storage program And a computer-readable recording medium having recorded thereon a machine translation processing program characterized by comprising: a delimiter position changing program for changing a position, wherein the original text delimited by the character number limit is re-separated at the delimiter position changed by the delimiter position changing program. Possible recording medium. 7. The program according to claim 6, wherein the delimiter position changing program adds a character string after the newly delimited part to the beginning of the character string after the part delimited by the reading program. A computer-readable recording medium on which the machine translation processing program described above is recorded. 8. A computer-readable storage medium storing a machine translation processing program according to claim 6, further comprising a break position setting program that allows a user to set information of the break position storage program. 9. The translation program according to claim 1, wherein a result of the morphological analysis performed by the second morphological analysis program is regarded as a result of the morphological analysis performed by the first morphological analysis program, and the translation program performs a translation process. A computer-readable recording medium recording the machine translation processing program according to claim 6. 10. The computer-readable recording medium according to claim 6, wherein the delimiter position storage program can set priorities for delimiting information to be described.