JPH07141374A - Machine translation system - Google Patents

Abstract

PURPOSE:To resolve the ambiguity of a part of speech, which cannot be discrimi nated with only syntax information, while having a conversation with a user. CONSTITUTION:An input means by which the user designates a part of speech, a rule dictionary 3 where a rule, which is applied in the case that the ambiguity of the part of speech cannot be correctly resolved by an already prepared grammatical rule, is stored, and a display means 6 on which a translation executing part 4 prompts the user to perform the part-of-speech designating operation by the display are provided besides a conventional machine translation system. Thus, erroneous operation of the user is avoided to improve the efficiency of translating work.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation device which automatically converts a language into another language when the language is input.

[0002]

2. Description of the Related Art In recent years, machine translation devices have been widely used as devices for translating from one language to another.
However, with the current machine translation technology, it is difficult to perform a practical automatic translation, and for a sentence in which a semantic interpretation or a context must be taken into consideration, translation is performed with the intervention of the user. In particular, it is difficult to uniquely determine the part of speech of the original sentence only from the syntax information, and when translating a sentence that requires semantic interpretation, the user performs a pre-editing process in which the part of speech is designated in advance ( For example, Japanese Patent Laid-Open No. 03-2220
Disclosed in Japanese Patent No. 66).

[0003]

However, in the conventional machine translation device as described above, the user designates a part of speech, and (1) first translates once, and machine translation is performed for the original sentence whose translation result is incorrect. The device interprets it as an incorrect part of speech, so it guesses that the translation result was wrong, and assigns the part of speech to an appropriate word.

(2) The machine translation device preliminarily designates a part of speech for a word that is likely to have a wrong part of speech. It went as above.

According to the method (1), it takes too much time to translate a long original sentence twice. In addition, the user must infer from the translated text that "the machine translation device has interpreted as a wrong part of speech and proceeded with the translation". Naturally, the user must judge whether the translation error is due to the part-of-speech error based on only the translation result.
There is a possibility of performing an unnecessary part-of-speech designation operation, and even if the part-of-speech designation is performed and retranslation is performed, the translation result may not be correct.

According to the method (2), the user predicts a word that is likely to be misinterpreted by the translation device and specifies the part of speech in advance. Therefore, the user specifies the part of speech for a word that is completely unambiguous. There is a possibility that an unnecessary operation will be carried out by specifying the part of speech even for a word that can resolve the ambiguity with existing grammar rules.

In any case, the conventional machine translation device may cause the user to perform useless operations.

The present invention solves the above problems, and provides a machine translation apparatus capable of designating a part of speech while interacting in the intermediate processing stage of machine translation and performing accurate translation without wasteful operations. The purpose is to do.

[0009]

SUMMARY OF THE INVENTION In order to solve the above problems, the present invention uses a conventional machine translator, an input means by which a user designates a part of speech, and an ambiguous part of speech in an existing grammar rule. A rule dictionary that stores rules to be applied when the problem cannot be resolved, and a display unit that displays to prompt the user to perform a part-of-speech designation operation are provided.

[0010]

According to the above configuration, the ambiguity of the part of speech that cannot be resolved by the existing grammar rule is performed in the transitional state of the translation, so that the user can specify a word for which the part of speech is not required to be specified,
Further, it is not necessary to presume how the translation device makes a part-of-speech error in advance and specify the part-of-speech.

That is, by eliminating the ambiguity of the part of speech while interacting with the user, useless operations by the user are omitted, and the efficiency of the translation processing work is improved.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A machine translation device according to an embodiment of the present invention will be described below with reference to the drawings by taking an example of translation from English to Japanese.

FIG. 1 shows the configuration of a machine translation apparatus according to an embodiment of the present invention. In FIG. 1, the input means 1 is for inputting an original sentence,
Specify translation and part of speech. The word dictionary 2 stores grammatical information about words necessary for analyzing the original sentence, translation words for the words, and words in the target language and their translation word information. The rule dictionary 3 stores the grammatical rules in the source language and the target language, and the rules to be applied when the grammatical rule cannot be applied to correct the disambiguation of the part of speech. The original sentence / translated sentence information management table 5 stores the original sentence input by the input means 1 and the translation result in association with each other. The translation executing unit 4 refers to the word dictionary 2 and the rule dictionary 3 for the original text input by the input means 1 and translates the original text input by the input means 1. When the translation execution unit 4 displays the original sentence and the translated sentence stored in the original sentence translated text information management table 5 in association with each other on the display unit 6, and the translation execution unit 4 cannot resolve the ambiguity of the part of speech by the existing grammar rule. A message prompting the user to specify a part of speech is displayed.

FIG. 2 shows a circuit block diagram of the machine translation apparatus of this embodiment. In the figure, 7 is a keyboard, 8 is a central processing unit (hereinafter abbreviated as CPU), 9 is a cathode ray tube display (hereinafter abbreviated as CRT), 10 is a read-only memory (hereinafter abbreviated as ROM). , 11
Is a random access memory (hereinafter abbreviated as RAM).

The input means 1 executes the program stored in the ROM 10 while the keyboard 7 operates the word dictionary 2, the rule dictionary 3 the ROM 10, and the translation execution unit 4 causes the CPU 8 to exchange data with the ROM 10 and the RAM 11. According to the original translation information management table 5
Is realized by the RAM 11 and the display means 6 is realized by the CRT 9.

The operation of the English-Japanese machine translation apparatus of this embodiment constructed as described above will be described with reference to the flow chart of FIG.

First, in step A, the user inputs means 1
Enter the English sentence from. The entered English text is translated into the translation execution unit 4
Is stored in the original translated text information management table 5.

Here, the English sentence "Is is living." Will be described as being input.

The state of the original text translated text management table at this time is shown in FIG. In step B, the translation execution unit 4 translates using the word dictionary 2 and the rule dictionary 3. Translation execution unit 4
Is a morphological analysis process that divides the input sentence into words and retrieves the part-of-speech and grammatical information of each word by referring to the word dictionary 2, the part-of-speech and grammatical information of each word divided by the morphological analysis process, And the rule dictionary 3 to refer to the multi-part-of-speech elimination process for uniquely determining the part-of-speech, and the sequence of the part-of-speech determined by the multi-part-of-speech elimination process based on the grammatical information of the word dictionary 2 using the rule dictionary 3. Authenticate the syntax of the language, execute the analysis / conversion / generation process to convert and generate the target language. Here, in the multi-part-of-speech elimination processing, the translation execution unit 4 searches the rule dictionary 3 and applies the grammatical rule. If the grammatical rule to be applied at this time is a rule that entrusts the user to determine the part-of-speech, The display unit 6 displays a message prompting the user to specify the part of speech, and prompts the user to specify the part of speech of the word.
Note that the morphological analysis processing and the analysis / conversion / generation processing are the same as those of the existing translation device, and thus a detailed description thereof will not be given.

In the example sentence, first, the translation execution unit 4 performs a morphological analysis process, and the input sentence is "It", "is", "liv".
ing ”and refer to the word dictionary 2 to extract part of speech, grammatical information, and translated word information for the word (FIG. 5). The information for each word is shown in FIG.

Next, the translation execution unit 4 performs a multi-part-of-speech elimination process. Since "It" and "is" have no ambiguity in the part of speech, the part of speech is uniquely determined. Further, the translation execution unit 4 determines the part of speech of “living” when determining the rule dictionary 3
Search for applicable rules for. That is, the translation execution unit 4
Means that "living" immediately after the be verb is an adjective,
The word dictionary 2 as the current participle and noun of the verb "live"
Is registered in the dictionary, and the applicable rule in FIG. 7, namely rule 1 “A word registered in the dictionary as a past participle, an adjective, or a noun cannot uniquely determine the part of speech with the syntax information”. Apply. Furthermore, the translation execution unit 4
Recognizes that the rule 1 is a rule which entrusts the user with determining the part of speech, displays the word for which the part of speech is designated and the part of speech of the word on the display means 6 (FIG. 6), and waits for the user's designation operation.
Then, the translation process proceeds with the part of speech designated by the user.

Further, the translation execution unit 4 searches the rule dictionary 3, sequentially applies the applicable rules, advances the analysis / conversion / generation processing, stores the translated text in the original translated text information management table 5, and proceeds to step C. move on.

FIG. 8 shows the state of the original text translated sentence pair management table when the user uses the part of speech of "living" as an adjective in the multi-part-of-speech elimination processing in step B.

In step C, the translation execution unit 4 displays the original sentence and the translated sentence stored in the original sentence translated sentence information management table 5 in association with each other. FIG. 9 shows a state of the screen after the translation process in the case where the user uses the part of speech of “living” as an adjective in the multi-part-of-speech elimination processing of step B.

Further, in the machine translation device according to the second aspect of the present invention, a display characterizing which word in the original sentence is the word for which the user specifies the part of speech is displayed.

That is, in the multi-part-of-speech elimination processing in step B, the position information of the word for which the user is prompted to specify the part-of-speech is obtained, and the word in the original sentence is characterized by highlighting based on the position information. Display. The state of the screen at this time is shown in FIG.

In the machine translation device according to the third aspect of the present invention, the reason why the machine translation device cannot determine the part of speech is displayed to the user.

That is, in the multi-part-of-speech elimination processing of step B, when the translation execution unit 4 searches the rule dictionary 3 and displays a message prompting the user to specify the part-of-speech, the translation execution unit 4 is registered in the rule dictionary 3 at the same time. The reason why the determined part of speech cannot be determined is also displayed.

The contents of the rule dictionary 3 at this time are shown in FIG. 11, and the appearance of the display screen is shown in FIG.

As described above, according to the present embodiment, it is possible to provide a machine translation apparatus capable of performing part-of-speech designation while interacting in the intermediate processing stage of machine translation and performing accurate translation without wasteful operation. it can.

[0031]

As is apparent from the above description of the embodiments, according to the present invention, it is possible for the user to eliminate the ambiguity of the part of speech in the transitional state of translation, not in the pre-editing stage. The number of erroneous operations (to specify a part of speech for a word that can be solved by existing grammar rules) is reduced. That is, since the number of user operation instructions is reduced, the efficiency of translation work is improved.

[Brief description of drawings]

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

FIG. 2 is a block diagram of the same circuit.

FIG. 3 is a flowchart showing an outline of the operation of the machine translation device.

FIG. 4 is a diagram showing an input original sentence in the original sentence translated sentence management table in step A.

FIG. 5 is a diagram showing an input original sentence and a translated language in step B.

FIG. 6 is a diagram showing information about words in the input original sentence in step B.

FIG. 7 is a diagram showing a configuration of a rule dictionary of the machine translation device.

FIG. 8 is a diagram showing a state of an original translated sentence pair management table at the time of multi-part-of-speech elimination processing in step B.

FIG. 9 is a diagram showing a display after translation processing in step B.

FIG. 10 is a diagram showing a display when the user specifies a part of speech in step B.

FIG. 11 is a diagram showing the contents of a rule dictionary at the time of multi-part-of-speech elimination processing in step B.

FIG. 12 is a diagram showing a display of contents of a rule dictionary at the time of multi-part-of-speech elimination processing in step B.

[Explanation of symbols]

 1 Input Means 2 Word Dictionary 3 Rule Dictionary 4 Translation Execution Unit 5 Translation Information Management Table 6 Display Means 7 Keyboard 8 CPU 9 CRT 10 ROM 11 RAM

Claims (3)

[Claims] 1. Input means for inputting an original sentence, a translation instruction, and a part-of-speech designation, grammatical information about a word necessary to analyze the original sentence, a translated word for the word, and a word in the target language and its grammatical A word dictionary that stores information, a rule dictionary that stores a grammatical rule in a source language and a target language, and a rule that is applied when the grammatical rule cannot be used to correctly disambiguate a part of speech The original sentence is referred to the word dictionary and the rule dictionary, and a translation execution unit that translates the original sentence input by the input unit and the original sentence input by the input unit and the translation result are stored in association with each other. An original text translated text information management table, wherein the translation execution unit displays the original text and the translated text stored in the original text translated text information management table in association with each other, and the translation execution unit If you can not resolve the ambiguity of the part of speech in the existing grammar rules, machine translation device provided with a display means for performing display to prompt the part of speech specified operation to the user. 2. The translation execution unit obtains position information in a source sentence of a word that prompts the user to specify a part of speech, and based on the position information, the display means displays a characterized word in the source sentence. Machine translation device. 3. A rule dictionary stores part-of-speech combination information for which ambiguity cannot be resolved by a grammar rule and the reason why the part-of-speech ambiguity cannot be resolved at the same time. 2. The machine translation device according to claim 1, wherein the reason why the part of speech cannot be correctly corrected is displayed at the same time as the display prompting ".