JPH0926962A - Translation support device - Google Patents

Abstract

(57) [Abstract] [Problem] To provide a device for supporting translation so that a user can create a natural and easy-to-read translated sentence by eliminating a conventional long and difficult-to-read translated sentence that is one output translated sentence corresponding to one original sentence. . SOLUTION: A natural language sentence input by an input unit 1 is stored in an original sentence storage unit 2 and then translated by a translation unit 5. The translation is a syntactic analysis rule in a grammatical rule 6a of a translation dictionary unit 6. That is, focusing on the predicate included in the input sentence, the predicate is divided into information units according to the term structure required by the predicate and the limit of the number of predicates included in each information unit. After being stored in the translated sentence storage unit 3, it is displayed on the display unit 7. The user can freely create a translated sentence by checking the correspondence between the information unit displayed on the display unit 7 and the translation result.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a translation device,
More specifically, the present invention relates to a translation support device that divides an input sentence into information units and translates the sentence in order to assist the operator in creating a natural translated sentence.

[0002]

2. Description of the Related Art A conventional machine translation device executes a predetermined translation process (morphological analysis, syntactic analysis, conversion / generation process) on an input sentence and outputs a corresponding translated sentence. In this case, even in the case of a long original sentence, since one output sentence (translated sentence) is generated for one input sentence in principle, the translated sentence may be difficult to read or may not be understood. . In particular, when the language structure of the input language and the target language are significantly different (for example, English and Japanese), the correspondence between the elements in the input sentence and the elements in the output sentence becomes complicated, and the operator outputs It also imposes a great burden on editing a sentence to complete a translated sentence.

The following proposals have been made so far regarding the method of dividing an input sentence and translating it.
For example, in Japanese Unexamined Patent Publication No. 60-215282, "Natural language analysis syntax method and device" divides an input sentence, executes analysis in phrase units for each division unit, and collects the results into one to finish the analysis. That is. However, it is intended to perform the parsing process in parallel in small units, and is not a division process aiming at easy-to-understand translation. In addition, depending on how the data is compiled at the final stage of analysis, it may be difficult to understand the translated text. Further, the "machine translation device" of Japanese Patent Laid-Open No. 63-109574 is capable of limiting the translation time of one sentence and stopping the translation or executing an automatic division translation process as needed. However, it is intended to improve the efficiency of the translation process by limiting the translation time per sentence, and the automatic division translation process itself represents one processing form when the time limit is exceeded, and is a process aimed at easy-to-understand translation. Is different from. Moreover, since the whole sentence is not divided after the analysis process is performed, the translation speed is improved, but the accuracy tends to be deteriorated. The "machine translation device" of Japanese Patent Laid-Open No. 1-1113870 is capable of designating divided translations when there is no grammatical rule corresponding to a specific syntax or when it is a long sentence.
However, according to the division translation mark inserted in the input sentence by the operator, the processing is performed for each division unit, and the machine does not have a function of automatically dividing the input sentence. Also, not only is it inefficient that the operator must enter the split translation mark,
It is quite difficult to judge where to enter.

Further, the "translation processing method" of Japanese Patent Application Laid-Open No. 1-152573 is to divide an input sentence into one or more phrases and translate them, and edit the result to create a translated sentence. However, the division unit is limited to phrases. In addition, the phrase translation is for the purpose of creating a translated sentence on a sentence-by-sentence basis, and the operator must create the final translated sentence while changing the order of phrase translation. On the other hand, the "machine translation device" of Japanese Patent Laid-Open No. 5-346937.
Then, in order to solve the above problems, we propose a method that automatically divides the input sentence into certain information units, and further, that enables the generation of a translated sentence that makes it easy to match the input sentence with the translated sentence. ing. However, in this case, the present invention relates to divided translation and display aimed at the effect that the contents of the input sentence can be understood quickly, and is not intended to assist the operator in finally producing the translated sentence. Also, since the unit of information is relatively small (that is, the number of words is small), when translated, it may be suitable for reading but not for producing translated text. high. As described above, none of the above-mentioned publications has proposed a translation support method that allows an operator to automatically divide and translate an original sentence into units useful for creating a translated sentence.

Further, the following proposal has been made as a method of preparing a parallel translation database, searching and displaying a translation example using a word or phrase designated by an operator as a search key, and supporting the creation of a translated text. Japanese Patent Laid-Open No. 6-195380, "Inspection result display method in translation support system"
Then, in the bilingual example sentence search, it is possible to simultaneously refer to example sentences having different translated words without changing the dictionary heading. This is to enable parallel translation example sentences with different translations for the same headword to be searched at the same time, but the focus is on the operator determining the correct translation word based on the parallel translation example sentence. , It is not possible to achieve the purpose of referring to the expression style necessary for creating a more natural translation, that is, the expression method around the translated word.

[0006]

SUMMARY OF THE INVENTION It is an object of the present invention to solve the above-mentioned problems and to provide a translation support device that allows an operator to easily create a more natural and readable translation.

[0007]

In order to solve the above-mentioned problems, the invention according to claim 1 of the present invention comprises an input unit for inputting a natural language sentence and a translation target input from this input unit. An original sentence storage unit for storing the following original sentence, a translation dictionary unit accommodating knowledge information used for translation processing, a translation unit for executing the translation process of the input original sentence using the contents of this translation dictionary unit, A translated text storage unit that stores a translated text obtained as a result of the translation process, a display unit that displays the original text and the translated text, and each operation is controlled, and edit processing of the original text or the translated text displayed on the display unit is performed. In the translation support device including an edit control unit, the translation unit includes a division processing unit that divides an original sentence into information units that comply with a syntax analysis rule included in the translation dictionary unit to perform translation, or
In the invention according to claim 2, in the invention according to claim 1, the parsing rule in the division processing unit focuses on a predicate included in an input sentence, and includes the term structure and each information unit required by the predicate. In the invention according to claims 1 and 2, the natural language sentence input by the input unit is stored in the source sentence storage unit, and the stored source sentence is It is translated by the translation unit, but since the translation is divided into information units that comply with the parsing rules of the translation dictionary unit, it is possible that long and hard-to-read translated sentences are created, and collation work takes time. Moreover, the user can freely create a translated sentence from the correspondence between the information unit and the translation result. Further, the syntactic analysis rule pays attention to the predicate included in the input sentence, and is based on the term structure required by the predicate and the number of predicates included in each information unit, thereby providing more effective translation support. become.

In the invention according to claim 3 of the present invention, an input unit for inputting a natural language sentence, and an original sentence storage unit for storing an original sentence to be translated input from the input unit, A translation dictionary unit containing knowledge information used for the translation process, a translation unit that executes the divided translation process of the input original sentence using the contents of the translation dictionary unit, a parallel translation database that collects translation examples, and A translated text storage unit that stores a translated text obtained as a result of the translation process, a display unit that displays the original text and the translated text, and a search result of the parallel translation database,
In a translation support device including an editing control unit that controls each operation and edits an original sentence or a translated sentence displayed on the display unit, a translation from the parallel translation database is performed based on a word specified by an operator. In the invention according to claim 3, the natural language sentence input by the input unit is stored and stored in the original sentence storage unit. Although the original sentence is translated by the translation unit, the translation example is searched by the parallel translation example sentence search unit based on the words and phrases specified by the operator in the original sentence and the translated sentence from the translation database in which the translation examples are collected. Then, the operator creates a translated sentence.

Further, in the invention according to claim 4 of the present invention, in the invention according to claim 3, the parallel translation example sentence search section uses the pair of the original word and the translated word of the phrase designated by the operator as a search key. A translation example is searched from a bilingual database, and in the invention according to claim 4, the translation example is searched using a pair of an original word and a translated word as a search key, and the operator refers to this example, Create a translation.

In the invention according to claim 5 of the present invention, in the invention according to claim 3, the parallel translation example sentence search unit determines that the original word is the original word when the operator corrects the translation of the desired word. A translation example is searched from the parallel translation database using the corrected translation word pair as a search key. In the invention of claim 5, when the operator corrects the translation word, the translation word A translation example is searched using the corrected translation word pair as a key, and a translation is created with reference to this example. In the inventions according to claims 3 to 5, a parallel translation database in which a large number of translation examples are collected is prepared so that a specific word in an input sentence or a translation result by a machine is used as a search key. Can be searched for, and by referring to the search results, it is possible to support more natural translations that are not in direct translation.

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the translation support device of the present invention will be shown and described below in the order of the claims. FIG.
1 is a block diagram showing an embodiment of the invention described in claims 1 and 2, wherein 1 is an input unit, 2 is an original sentence storage unit, 3 is a translated sentence storage unit, 4 is an edit control unit, 5 is a translation unit, and 6 Is a translation dictionary unit, 6a is a grammar rule, 6b is a word dictionary, and 7 is a display unit. FIG. 2 shows a processing flow of the translation unit 5. The steps will be described in order below. First, when an original sentence is input (step 21), the translation unit performs morphological analysis of the original sentence (step 22). In the morpheme analysis, morphemes are extracted from the original sentence using the word dictionary 6b, and how the extracted morphemes are combined to form a word is analyzed. The result of this analysis is passed to the next parsing process (step 23). In the syntactic analysis, the tree structure is generated using the syntactic analysis rule in the grammatical rule 6a based on the part-of-speech information obtained by the morphological analysis. At this time, using the predicate of the input sentence as a key, a tree structure is generated in which the optimal structure is given the priority solution from all the syntactic structures that the predicate can take, and the result is used in the next syntax division process (step 24). Passed. The tree structure obtained by the syntax analysis process is divided into information units according to the syntax division rule in the grammar rule 6a. The information unit mentioned here refers to a group of constituent elements that form a unit of information. As shown in the invention of claim 2, this information unit is defined as a unit including the predicate as a center and the terms (also referred to as essential cases) that the predicate necessarily requires. FIG. 3 shows an example of a syntax division rule that defines such an information unit. Note that the syntax division rule example of FIG. 3 is written as a rule in which the number of predicates included in each information unit is limited to a maximum of one, but it can be rewritten to a rule that can include two or more predicates. Things can be done easily. The result of this division is the next division translation process (step
25) and the conversion in grammar rule 6a for each information unit
Translation is performed using the production rule, and the result is output (step 26).

Now, "This command specifies additiona
The translation process in the embodiment of the present invention will be described by taking an example of translating an English sentence such as “l devices which execute paging and swapping.” into Japanese. This input sentence is processed through steps 22 and 23 above 4 (A)
The tree structure is analyzed as shown in. In this case, the optimal structure is selected by using the predicate "specifies" of the main clause as a key, and the predicate "specifies" is "a" that functions as an object.
dditional devices which execute paging and swappin
g. "as a complement. In the next step 24, the syntax division rule as shown in FIG. 3 is applied to divide the tree structure received from the previous stage as shown in FIG. 4 (B). Then, in step 25, translation processing is performed for each information unit, and the translation as shown in the screen display example of Fig. 5 is output.
The operator creates a final translated sentence by referring to this translation result. The created translated text example is also shown as <translated text> in FIG.

Next, an embodiment of the translation support apparatus according to claim 3 will be described. FIG. 6 is a block diagram showing the configuration of the embodiment. In the figure, 1 is an input unit, 2 is an original sentence storage unit, 3 is a translated sentence storage unit, 4 is an edit control unit, 5 is a translation unit, and 6 is a translation unit. Dictionary part, 6a grammar rules, 6b word dictionary, 7 display part,
Reference numeral 8 is a parallel translation database, and 9 is a parallel translation example sentence search unit. FIG. 7 shows a processing flow of the translation unit 5. Hereinafter, the steps will be sequentially described. The processing from step 71 to step 76 is the same as the processing from step 21 to step 26 in FIG. Therefore, when the processing is performed in the same manner as described above and the translation is output for each information unit, the operator looks at the translation result and moves to the processing of searching a translation example for a certain specific phrase (step 77). This is done by clicking with the mouse on either the word in the source text you want to retrieve or the translation for that word (step 78). When the word to be searched is clicked with the mouse, the parallel translation example sentence search unit accesses the parallel translation database using the clicked word as a search key and executes the search process. Then, when the search result of the parallel translation database is displayed on the screen (step 79), the operator creates a translated text with reference to those parallel translation examples (step 80) and ends the processing.

Now, "This provides reliable transmiss
ion of data which is of crucialimportance to the c
The translation process in the embodiment of the present invention will be described by taking an example of translating an English sentence "lients." into Japanese. When step 76 is completed, the result as shown in FIG. 8A is displayed. Here, when the operator clicks the word "provides" with the mouse, the parallel translation example sentence search unit refers to the word dictionary 6b, accesses the parallel translation database using the headword "provide" of this word as a search key, and searches. As a result, the search result as shown in Fig. 8B is displayed on the screen, where the operator finds the search result, especially when "provide" is used as the main verb. Referring to the translation example, the overall structure of the Japanese translation is determined and translated, and an example of the translation created by the operator is shown in FIG.

Next, the invention according to claim 4 is characterized in that, when translating the above-mentioned English sentence into Japanese, when the parallel translation example sentence search unit accesses the parallel translation database, the search key is a pair of the original language and the translated language. Is to have. In the embodiment of the present invention, when the operator clicks "provides" or its translation "provide" with the mouse at the time when the translation result as shown in FIG. Access the bilingual database using the pair of as a search key and execute the search. In this case, a technique for associating the original sentence side word with the translated sentence side word is a known technique and can be easily realized, and therefore the description thereof is omitted here. In the bilingual database, the bilingual example sentence is preliminarily analyzed and the correspondence between the words is held, so that the search as described above can be easily performed. It is assumed that the search result is displayed as shown in FIG. Since the search example is narrowed down as compared with the case of FIG. 6B, the search time is shorter and the operator can easily refer to it. Here, as shown in FIG. 9C, the operator refers to a translation example including the expression “provide” and refers to, for example, “this provides reliable data transmission. This is extremely important to the client. "

Next, an embodiment of the invention described in claim 5 will be described. The English sentence “This provides reliable t
ransmission of data which is of crucial importance
The result of the divided translation process of “to the clients.” is shown in FIG.
It is assumed that it is as shown in (A). Here, the operator determines that “supplied” output as the translated word of “provides” is inappropriate and corrects it to the translated word of “provide” using the editing editor (FIG. 10).
(B)). After that, the example sentence search is executed as described above using the pair of “provide” and “provide” as the search key, and the operator creates a translated sentence by looking at the result (FIG. 10 (C), (D)).

[0017]

As is apparent from the above description, according to the present invention, the machine automatically divides the input sentence into information units and displays the translation corresponding to the division units on the screen, so that the operator can Conventionally, when creating a translated sentence, the collation work required between one original sentence and one machine output translated sentence is significantly reduced. In addition, it is possible to freely create a translated sentence according to an information unit and its translation result without being bound by the conventional limitation that one translated sentence is always generated for one original sentence. Therefore, a translation that is long and difficult to read is not created. Furthermore, by preparing a parallel translation database that collects a large number of translation examples, it is possible to search for parallel translation examples using a certain word in the input sentence or in the translation result by the machine as a search key. By doing so, it is possible to obtain the effect that it is possible to support more natural translation production, which is not a direct translation.

[Brief description of drawings]

FIG. 1 is a block diagram of an embodiment of the invention described in claims 1 and 2. FIG.

FIG. 2 is a diagram showing a processing flow of a translation unit in FIG.

FIG. 3 shows an example of syntax division rules in the translation dictionary section of FIG.

FIG. 4 is a diagram showing a specific example of a tree structure of an original sentence before (A) and after (B) division processing.

FIG. 5 is a diagram showing a specific example of a created translated sentence.

FIG. 6 is a block diagram of an embodiment of the invention described in claim 3.

7 is a diagram showing a processing flow of a translation unit in FIG.

8 is a diagram showing steps (A) to (C) of the translation support process of FIG.

FIG. 9 is a diagram showing paths (A) to (C) of translation support processing according to the embodiment of the invention as set forth in claim 4;

FIG. 10 is a diagram showing paths (A) to (D) of the translation support processing according to the embodiment of the invention described in claim 5;

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Input part, 2 ... Original text storage part, 3 ... Translation storage part, 4 ... Editing control part, 5 ... Translation part, 6 ... Translation dictionary part, 6a ... Grammar rule, 6b ... Word dictionary, 7 ... Display part, 8 ... parallel translation database, 9 ... parallel translation example sentence search unit, steps 21 to 26 ... processing steps by the translation unit in FIG. 1, steps 71 to 80 ... processing steps by the translation unit in FIG.

Claims (5)

[Claims] 1. An input unit for inputting a natural language sentence, an original sentence storage unit for storing an original sentence to be translated input from this input unit, and a translation dictionary unit for accommodating knowledge information used for translation processing. A translation unit that executes a translation process of the input original sentence using the contents of the translation dictionary unit, a translation storage unit that stores a translated sentence obtained as a result of this translation process, and a display unit that displays the original sentence and the translated sentence. And a translation support device including an edit control unit that controls each operation and edits an original sentence or a translated sentence displayed on the display unit, the translation unit having the original sentence in the translation dictionary unit. A translation support device comprising a division processing unit that performs translation by dividing into information units that comply with syntax analysis rules. 2. The parsing rule in the division processing unit focuses on a predicate included in an input sentence, and is based on a term structure required by the predicate and a limit on the number of predicates included in each information unit. The translation support device according to item 1. 3. An input unit for inputting a natural language sentence,
An original sentence storage unit that stores an original sentence to be translated input from this input unit, a translation dictionary unit that stores knowledge information used for translation processing, and a content of the translation dictionary unit that is used to store the input original sentence. A translation unit that executes divided translation processing, a parallel translation database that collects translation examples, a translation storage unit that stores translated texts obtained as a result of the translation processing, the original text and translated texts, and the search results of the parallel translation database are displayed. In the translation support device including a display unit that controls, and each operation, and an editing control unit that edits an original sentence or a translated sentence displayed on the display unit, based on a phrase specified by an operator, A translation support apparatus comprising a parallel translation example sentence search unit capable of searching a translation example from the parallel translation database. 4. The translation support apparatus according to claim 3, wherein the parallel translation example sentence search unit searches the translation database for translation examples using a pair of an original word and a translated word of a phrase designated by an operator as a search key. 5. The parallel translation example sentence search unit searches a translation example from the parallel translation database by using a pair of an original word and a corrected translation word as a search key when an operator corrects a translation word of a desired word. The translation support device according to claim 3.