JPH05120336A - Machine translation communication system - Google Patents

Abstract

PURPOSE:To realize the machine translation communication system which can improve operation convenience of users by reducing communication quantity until obtaining the final translated result completely composed of translated words rather than the conventional quantity even when a source sentence contains unknown words. CONSTITUTION:When the source sentence contains the unknown words unregistered on dictionary files 13 and 14, a server computer 3 transmits the unknown words and a source sentence document ID to a client computer 2 while storing the source sentence in a source sentence file 11 as it is. The client computer 2 fetches unknown word registration information, which is required for registering the received unknown words on the user dictionary file 14, by an unknown word registration part 7 and transmits the fetched information to the server computer 3 together with the source sentence document ID. After the received unknown word registration information is registered on the user dictionary file 14, the server computer retrieves the source sentence of the received ID from the source sentence file 11, translates the source sentence by a machine translation part 12 and transmits the translated result to the client computer 2.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a machine translation communication system in which a computer requests another computer connected via a network for translation and receives a translation result.

[0002]

2. Description of the Related Art Although a machine translation device has been developed, many computers lack the ability to mount the system and a dictionary, and the machine translation device is still expensive.
Therefore, a computer equipped with a machine translation device (hereinafter referred to as a server computer) is set as a node of a computer network, and another computer not equipped with a machine translation device (hereinafter referred to as a client computer) becomes this server computer. A machine translation communication system has already been proposed which requests a translation and receives a translation result. For example, "Fumito Nishino and Naoto Nakamura," Machine Translation E-mail System ",
IPSJ Technical Report, Natural Language Processing 75-5, 19
A machine translation communication system has already been disclosed in "January 18, 1990".

The document described in this document holds a document to be translated by the client computer in advance as a file, and sends the file to the server computer by e-mail.
The server computer translates the sent document and sends the translation result back to the client computer as the sender by e-mail.

[0004]

However, in the conventional machine translation communication system, when a sentence sent by the client computer includes a word (unknown word) that is not included in the dictionary of the machine translation device of the server computer. In addition, the machine translation device of the server computer returns the translation result to the client computer without being able to translate the unknown word or the sentence including the unknown word. At this time, the user operating the client computer recognizes that there is an unknown word not registered in the dictionary of the machine translation device of the server computer for the first time after seeing the returned translation result, and the unknown word is stored in the server computer. The operation of registering in the machine translation device was performed, and then the document desired to be translated was sent again.

Therefore, the following problems have occurred. (1) The client computer has to send the same document to the server computer twice, resulting in poor communication efficiency and high communication cost. (2) The server computer also outputs the translation result before and after registering the dictionary.
Each time it is sent to the client computer, the communication efficiency becomes poor and the communication cost becomes high. (3) On the client computer side, the user must look at the full text of the translation result returned and find an unknown word that is not registered in the dictionary of the machine translation device of the server computer.
In addition, it is necessary to recognize that an unknown word is included, which imposes a heavy burden on the user.

In practice, the number of sentences the client computer requests the server computer to translate is not limited to one sentence, but is often many sentences, and the above problem is a serious problem.

The present invention has been made in consideration of the above points, and when the original sentence contains unknown words, the communication amount until the final translation result including all translated words is obtained, compared with the conventional method. It is an object of the present invention to provide a machine translation communication system that can reduce the number and improve the usability of a user for processing unknown words.

[0008]

In order to solve such a problem, according to the present invention, a server computer equipped with a machine translation unit for performing machine translation using a system dictionary file and a user dictionary file, and a network for this server computer. In a machine translation communication system that has a client computer that gives a source text via a computer, requests translation, and receives a translation result obtained by the server computer from a network, the server computer is registered in the source text in the system dictionary file and the user dictionary file. When it is recognized that there is an unknown word that has not been processed, the following processing is performed.

That is, when an unknown word which is not registered in the system dictionary file and the user dictionary file is included in the original sentence sent from the client computer to the server computer, the server computer keeps the received original sentence. , The unknown word is extracted and transmitted to the client computer together with the identifier of the original document.

At this time, the client computer takes in the unknown word registration information necessary for registering the transmitted unknown word in the user dictionary file by the unknown word registration unit and transmits it to the server computer together with the identifier of the original document. ..

Upon receiving this transmission information, the server computer registers the transmitted unknown word registration information in the user dictionary file, and thereafter, the machine translation unit translates the original sentence associated with the transmitted original document identifier. Then, the obtained translation result is transmitted to the client computer.

[0012]

According to the present invention, when the server computer recognizes that the original sentence contains an unknown word which is not registered in the system dictionary file and the user dictionary file, the processing is performed as follows, and the translated word for the unknown word is determined. In addition, we tried to reduce the amount of communication to obtain the final translation result, including that the user could easily recognize the existence of unknown words and their terms.

That is, when there is an unknown word in the original sentence that is not registered in the system dictionary file and the user dictionary file, the server computer stores the original sentence instead of transmitting the translation result in which the unknown word is left as it is. As it is, unknown words in the original text are extracted and transmitted to the client computer together with the identifier of the original text document.

At this time, the client computer takes in the unknown word registration information necessary for registering the transmitted unknown word in the user dictionary file by the unknown word registration unit and transmits it to the server computer together with the identifier of the original document. ..
Here, since the original text remains stored in the server computer, the client computer does not need to send the original text again.

When the unknown word registration information is received, the server computer registers the transmitted unknown word registration information in the user dictionary file, and then stores the original text in the storage state related to the transmitted original text document identifier as a machine. The translation unit translates and the obtained translation result is transmitted to the client computer.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the drawings. Here, FIG. 1 shows the entire configuration of the machine translation communication system according to this embodiment.

In FIG. 1, this machine translation communication system requests a plurality of machine translation requests to the network 1 (see FIG. 1).
1 is taken out and shown) and at least one server computer 3 in which the machine translation configuration is installed are connected. The client computer 2 and the server computer 3 are actually composed of workstations, personal computers, etc., and functionally have the structure shown in FIG.

The client computer 2 includes a document input unit 4 for inputting an original sentence to be translated by a user, a document file 5 for storing the input original sentence, a stored translation request original sentence, etc. from the client computer 2 to the server computer. 3 to the client computer 2 from the server computer 3
A document communication unit 6 for receiving the translation result and the like transmitted to the server, an unknown word registration unit 7 for registering an unknown word, a translation result file 8 for storing the translation result given from the server computer 3, And a translation result output unit 9 for presenting the translation result to the user.

On the other hand, the server computer 3 receives the translation request original text or the like transmitted from the client computer 2 to the server computer 3 or transmits the translation result or the like from the server computer 3 to the client computer 2. Document communication unit 10, document identification information (hereinafter, abbreviated as document ID)
A source text file 11 for storing the source text with
A machine translation unit 12 for translating the original text stored in the original text file 11, a system dictionary file 13 for storing a system dictionary used by the machine translation unit 12 for translation, and a machine translation unit 12 for a user to arbitrarily register words. User dictionary file 14 used by the machine translation unit 12, an unknown word file 15 for storing an unknown word found when the machine translation unit 12 is translating the original sentence of the original sentence file 11, and the machine translation unit 12 is an original sentence file. The translation result file 16 stores a translation result (translation sentence) obtained by translating the 11 original sentences.

The document communication unit 6 of the client computer 2 and the document communication unit 10 of the server computer 3 perform processing according to the transmitted / received message. Regarding the unknown word processing which is the feature of this embodiment, the document communication unit 10 of the server computer 3 includes an unknown word information transmission unit 10a and an unknown word registration information reception unit 10b, and the document communication unit 6 of the client computer 2 is unknown. A word information receiving unit 6a and an unknown word registration information transmitting unit 6b are provided. The unknown word information transmission unit 10a outputs the information of the unknown word stored in the unknown word file 15 to the document I when the machine translation unit 12 instructs to transmit the unknown word information.
D is added and transmitted to the client computer 2. The unknown word information receiving unit 6a gives the unknown word information given from the server computer 3 to the unknown word registering unit 7. The unknown word registration information transmitting unit 6b uses the unknown word registration unit 7 to identify the unknown word information (hereinafter, translated word information and the like) (hereinafter,
The document ID is added to the unknown word registration information) and transmitted to the server computer 3. Unknown word registration information receiving unit 1
0b stores the unknown word registration information given from the client computer 2 in the user dictionary file 14. Therefore, the unknown word registration information is stored in the user dictionary file 1
It consists of the information necessary to register in 4.

FIG. 2 is a flow chart showing the processing flow of the machine translation communication system of this embodiment. The process until the client computer 2 finally outputs the translation result will be described below with reference to FIG.

The user inputs the original sentence desired to be translated through the document input unit 4 and stores it in the document file 5. When the input is completed, the document input unit 4 notifies the document communication unit 10 of that fact, and the document communication unit 10 assigns the document ID to the original sentence stored in the document file 5, and the message is a translation request message. It is made clear in a header or the like to transmit the information to the server computer 3 (step 100).

The original text and the document ID received by the document communication unit 10 of the server computer 3 are stored in the original text file 11 (step 101). When such a storing operation ends, the document communication unit 10 gives the machine translation unit 12 reception end information of the original sentence, and at this time, the machine translation unit 12 translates the original sentence stored in the original sentence file 11 into the system dictionary file 13 and The user dictionary file 14 is used for translation (step 102). And the machine translation unit 12
Determines whether or not there is an unknown word in the original sentence that is not registered in the system dictionary file 13 and the user dictionary file 14 (step 103).

As a result, when it is determined that the original sentence does not include an unknown word, the machine translation unit 12 stores the obtained translation result (target language sentence) in the translation result file 16 and then notifies the document communication unit to that effect. 10, the document communication unit 10 assigns the document ID given from the client computer 2 to the translation result, clarifies that the message is a translation result message with a header or the like, and sends it to the client computer 2 ( Steps 104, 105).

The document communication unit 6 of the client computer 2 is
When the message is received, the translation result and the document ID are stored in the translation result file 8, and when the storage is completed, the translation result output unit 9 is activated, and the translation result output unit 9 causes the translation result file 8 to translate. The result is presented to the user in association with the original text stored in the document file 5 (steps 106 and 107).

On the other hand, when the machine translation unit 12 determines that the original sentence contains an unknown word, it stores the unknown word in the original sentence in the unknown word file 15 and informs the document communication unit 10 that the processing has been performed. (Step 108). At this time, the unknown word information transmission unit 10a of the document communication unit 10 assigns the document ID of the current processing target to the list of unknown words in the original sentence stored in the unknown word file 15, and the electronic message is an unknown word information electronic message. It is clarified by a header or the like and transmitted to the client computer 2 (step 109).

On the client computer 2 side, the unknown word information receiving unit 6a of the document communication unit 6 gives the unknown word information sent from the server computer 2 to the unknown word registration unit 7, and the unknown word registration unit 7 receives the unknown word. Information is presented to the user and the information necessary for registering the translated word, the part of speech, etc. in the user dictionary file 14 is input, and the unknown word registration information obtained in this way is given to the document communication section 6 (step 110). .. Document communication unit 6
The unknown word registration information transmission unit 6b assigns a document ID to the unknown word registration information, clarifies that the electronic message is an unknown word registration information electronic message with a header or the like, and transmits it to the server computer 3 (step 111). ).

On the server computer 3 side, the document communication unit 1
The unknown word registration information receiving unit 10b of 0 takes out the unknown word registration information included in the received electronic message and registers it in the user dictionary file 14, and the machine translation unit 12 receives the document ID included in the electronic message. It is given to cause the machine translation unit 12 to retrieve the original sentence having the document ID from the original sentence file 11 and restart the translation process (steps 112, 11).
3). The machine translation unit 12 translates the searched original sentence using the system dictionary file 13 and the user dictionary file 14 (the unknown word is registered at this time) (step 114).

Since it is not necessary to confirm whether or not an unknown word is included in the translation result obtained in this way, the machine translation unit 12 proceeds to step 104 described above and immediately obtains the translation result obtained. The translation result file 16 is stored. Subsequent processing is the above-mentioned steps 105 to 105.
It is as 107.

Next, assuming that the original sentence shown in FIG. 3A is input, the above-mentioned series of processing will be described. When the original text shown in FIG. 3A is input to the client computer 2, a document ID (“0001” in FIG. 3) is added to the original text and the information section of the electronic text is configured as shown in FIG. 3B. Then, the translation request message is transmitted to the server computer 3.

Upon receipt of this translation request message, the server computer 3 stores the document ID and the original sentence contained in this message in the original sentence file 11, and then the machine translation unit 12 stores the original sentence in the system dictionary file 13 and Try to translate using the user dictionary file 14. Here, it is assumed that the word “kansei” in the original sentence shown in FIG. 3A is not registered in the system dictionary file 13 or the user dictionary file 14. In this case, as shown in FIG. 3C, the server computer 3 registers the word “sensitivity” as an unknown word in the unknown word file 16 together with the ID “0001” of the document to be translated, and then the server. The computer 3 transmits to the client computer 2 an unknown word information electronic message including the contents registered in the unknown word file 16 in the information section.

In the original sentence shown in FIG. 3A, the word "sensitivity" is included twice, but when the word is stored in the unknown word file 16, duplication is confirmed and one unknown word information is stored. Only stored in the unknown word file 16.

The client computer 2, which has received the unknown word information message, is insufficient to present the unknown word "sensitivity" included in the message to the user and register it in the user dictionary file 14 (for example, translated word or (Sentence of speech) is input by the user, whereby the client computer 2 uses the document ID “0001”, the unknown word “sensitivity”, its translation “impression”, and the part of speech “noun” in the information section as shown in FIG. The unknown word registration information message including the information is transmitted to the server computer 3.

When the server computer 3 receives the unknown word registration information telegram, the unknown word "sensitivity" and its translation "impression" and the part of speech "noun" inserted in the information part are received.
Etc. to the user dictionary file 16 to be registered, and then the document ID “0001” inserted in the information section.
To the machine translation unit 12 to search and translate the original sentence shown in FIG. 3 (A) stored in the original sentence file 11. At this time, since the information about the word “kansei” is registered in the user dictionary file 14, FIG.
There are no unknown words in the original sentence shown in (A), and a translation result with no unknown words is obtained.
Information in which the document ID “0001” is added to the translation result shown in is stored in the translation result file 16. Then, the document communication unit 10 transmits the translation result electronic message in which the stored content is inserted in the information unit to the client computer 2.

Thus, the translation result inserted in the translation result message along with the original sentence shown in FIG. 3A is presented to the user.

Therefore, according to the machine translation communication system of the above embodiment, when the server computer 3 determines that an unknown word exists in the original text, the server computer 3 transmits the unknown word information to the client computer 2 and the client computer 2 sends the unknown word information to the client computer 2. Since the unknown word registration information is transmitted from the computer 2 to the server computer 3 and registered in the user dictionary file 14 so as to correspond to the unknown word registration information, the client computer 2 obtains a final translation result as compared with the conventional communication system. It is possible to reduce the communication amount up to, and it is possible to reduce the communication cost.

Here, the unknown word information and the unknown word registration information are exchanged between the client computer 2 and the server computer 3. During this period, the original text is kept stored in the original text file 11 and the electronic text always includes the document ID. Therefore, it is not necessary to know which document the unknown word information or unknown word registration information pertains to, and it is also necessary to send and receive translation results including unknown words and send and receive the original text again. Absent. This also means that the server computer 3 can process other documents after transmitting the unknown word information message and before receiving the corresponding unknown word registration information message.

Further, according to the above-described embodiment, the unknown word registration unit 7 displays the unknown word and prompts the user to enter registration information. Unlike the conventional system that recognizes, the user can easily recognize the existence of the unknown word and the content of the unknown word, which is convenient.

In the machine translation communication system according to the present invention, not only the number of client computers is arbitrary, but the number of server computers is also arbitrary. For example,
It may be a communication system in which a separate computer has a machine translation configuration for each specific field. In such a case, a computer may function as a client computer as well as a server computer. Of course, the network may be wireless or wired.

In the above embodiment, the server computer determines whether or not an unknown word is included after the completion of a series of translation processes. However, at the end of morphological analysis or at the end of syntactic analysis. The presence of the unknown word may be determined at any time such as when the semantic analysis ends.

[0041]

As described above, according to the present invention, when the server computer determines that an unknown word exists in the original text, the server computer transmits the unknown word information to the client computer, and the client computer transmits the unknown word information. Since the unknown word registration information is sent to and registered in the user dictionary file so that it can be handled, it is possible to reduce the communication volume as a whole until the translation result is obtained, and improve the usability for unknown word recognition. It is possible to realize a machine translation communication system capable of performing.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a machine translation communication system according to an embodiment.

FIG. 2 is a flowchart showing the flow of the processing.

FIG. 3 is an explanatory diagram showing file contents and the like when a certain original sentence is input.

[Explanation of symbols]

1 ... Network, 2 ... Client computer, 3 ... Server computer, 6 ... Document communication unit, 6a ... Unknown word information receiving unit,
6b ... Unknown word registration information transmitting unit, 7 ... Unknown word registration unit, 10
Document communication unit, 10a ... Unknown word information transmitting unit, 10b ... Unknown word registration information receiving unit, 11 ... Original file, 12 ... Machine translation unit, 13 ... System dictionary file, 14 ... User dictionary file, 15 ... Unknown word file , 16 ... Translation result file.

Claims (1)

[Claims] 1. A server computer equipped with a machine translation unit for performing machine translation using a system dictionary file and a user dictionary file, and an original sentence is given to this server computer via a network to request translation, and the server computer obtains the translation. In a machine translation communication system having a client computer that receives a translation result from a network, an unknown text that is not registered in the system dictionary file and the user dictionary file in the original text sent from the client computer to the server computer. When there is a word, the server computer stores the original text,
The unknown word is extracted and transmitted to the client computer together with the identifier of the original document, and the client computer registers unknown word registration information necessary for registering the transmitted unknown word in the user dictionary file. And sends it to the server computer together with the identifier of the original document, the server computer registers the transmitted unknown word registration information in the user dictionary file, and then the identifier of the transmitted original document. A machine translation communication system, characterized in that the original sentence according to (1) is translated by the machine translation unit, and the obtained translation result is transmitted to the client computer.