JPH09114832A - Machine translating communication system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a machine translating communication system which correctly recognizes an unknown word for an original sentence in which the unknown word is included and has a small amount of communication traffic till the translation result composed of all the translated words is obtained. SOLUTION: When the unknown word which is not included in the dictionaries 13 and 14 of a server computer 3 is included in an original sentence, a morpheme analysis is performe for the original sentence by a machine translating part 12, unknown word information emphasizing the unknown word is formed, and the unknown word information and a document ID are transmitted to a client computer 2 by a server document communication part 10. In the client computer 2, the unknown word information is presented to a user by an unknown word registration part 7, the unknown word registration information necessary for a dictionary registration and the presence or absence of the correction of the unknown word is obtained, and the unknown word registration information and the document ID are transmitted to the server computer 3 by a client document communication part 6. After the server computer 3 registers the received unknown word registration information in the user dictionary 14, the original sentence is retrieved from an original file 11 by the document ID, a translation processing is performed by a machine translation part 12 and the translation result is transmitted 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 comprising a server computer equipped with a machine translation device and a client computer connected via a communication line and not issuing a machine translation device for issuing a translation request. Regarding communication systems,
In particular, it relates to unknown word processing from requesting a translation to receiving a translation result.

[0002]

2. Description of the Related Art In a conventional client-server type machine translation communication system, a word sent from a client computer to a server computer for a translation request is not included in a dictionary included in a machine translation device of the server computer. If there is an (unknown word), the machine translation device of the server computer extracts the unknown word and sends it to the client computer together with the original text including the unknown word and the identifier of the original document, and the client computer sends the unknown word. On the other hand, there is a machine translation communication system that fetches information of unknown words required for registration in the user dictionary and sends it to the server computer together with the identifier of the original text document.

[0003]

However, in the conventional machine translation communication system, the cause of being an unknown word as a result of the translation is that the word is not registered in the dictionary on the server computer side. Moreover, it is possible that the morphological analysis algorithm fails during translation. For example, when performing a morphological analysis, if "new device" is not registered in the dictionary, "new device" is changed to "new device" (known word) and "location".
There is a problem that the correct unknown word cannot be obtained because it is divided into (unknown word).

In addition, when there is an unknown word, the server computer must send back all the original sentences including the unknown word to the client computer, resulting in poor communication efficiency and high communication cost.

The present invention has been made in consideration of the above points, and when an original word requested for translation contains an unknown word, the unknown word is correctly recognized and presented to the user in an easy-to-understand manner. The usability can be improved. Further, the present invention proposes a machine translation communication system that can reduce the communication cost until the final translation result in which the original sentence is all translated words can be obtained.

[0006]

In order to solve the above problems, according to the present invention, a client computer that issues a translation request and a client computer are connected via a network, and translation is performed using a system dictionary and a user dictionary. In a server computer that executes processing and generates a translation result consisting only of translated texts and sends it to the client computer, the processing when the server computer has an unknown word in the original sentence that is not registered in the system dictionary and the user dictionary In order to do so, it has the following means.

The server computer stores an original text file that stores the original text sent from the client computer and the identifier assigned to the original text, and if an unknown word exists in the original text when the original text of the original text file is translated, it is unknown. A machine translation unit that forms unknown word information in which an unknown word is emphasized by morphologically analyzing an original sentence in which a word exists, and a server document communication unit that transmits the unknown word information and the identifier of the original sentence to a client computer.

The client computer sends to the server computer an unknown word registration unit that fetches unknown word registration information necessary for presenting the unknown word information to the user and registering it in the user dictionary, and the unknown word registration information and the original text identifier. When the unknown word registration information and the original text identifier are transmitted to the server computer, the server document communication section registers the unknown word registration information in the user dictionary, and the machine translation section stores the original text. It is characterized in that the original sentence related to the identifier is taken out from the original sentence file, and the translation process is executed by using the user dictionary in which the unknown word registration information is registered to generate the translation result.

[0009]

According to the present invention, when the server computer recognizes that there is an unknown word in the original sentence that is not registered in the system dictionary and the user dictionary, morphological analysis is performed on one sentence including the unknown word. At this time, an unknown word recognized by the server computer can be obtained by adding a mark emphasizing the unknown word and sending the original text including the information of the morphological analysis to the client computer. In addition, the user can input the correct unknown word when the unknown word is incorrect. Since the document ID and the sentence number are given to the original sentence, only necessary sentences can be transmitted when exchanging unknown words between the client computer and the server computer. Further, when an unknown word exists, it is necessary to re-translate again, but only a sentence in which an unknown word exists can be re-translated.

[0010]

Embodiments of the present invention will be described in detail with reference to the drawings. FIG. 1 is an overall configuration diagram showing an embodiment of a machine translation communication system of the present invention.

In FIG. 1, the machine translation communication system is
It is used when a user directly operates this machine translation communication system on the network 1 and issues a machine translation request 1
One or more client computers 2 and at least one server computer 3 equipped with a machine translation device are connected. In the present invention, a plurality of client computers 2 may be connected to the network 1, but in FIG. 1, one client computer 2 is extracted and shown. In the following description, the number of client computers connected to the network is 1. I will explain as one. The client computer 2 and the server computer 3 are actually composed of workstations, personal computers, and the like.

The configuration of the client computer 2 will be described.
The document input unit 4 is an input unit for a user to input an original sentence to be translated, and is composed of a keyboard, a mouse and the like. FIG. 3 shows an example of the original document. Original file 5 is
1 in the document in which the document ID (hereinafter referred to as the document ID) and the document ID are added to the original text input from the document input unit 4
Stores the original sentence with a sentence number for managing the sentence. FIG. 4 shows the data format of the original text file 5.

The client document communication unit 6 sends the translation request original text and the like stored in the original text file 5 to the client computer 2.
From the server computer 3 to the server computer 3 or from the server computer 3 to the client computer 2. The unknown word registration unit 7 presents the unknown word information sent from the server computer 3 to the user and causes the user to register information such as translated words, and the translation result file 8 stores the translation result and the document ID provided from the server computer 3. And the sentence number are stored. FIG. 8 shows the data format of the translation result file 8. The translation result output unit 9 displays the translation result and the original sentence on a display or the like.

Next, the configuration of the server computer 3 will be described.
The server document communication unit 10 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. The original text file 11 has the same format as the original text file 5, and stores the original text, the document ID, and the sentence number.

The machine translation unit 12 translates the original text stored in the original text file 11. The machine translation unit 12 performs morphological analysis as one step in translating the original sentence.
The result of the morphological analysis is not output, and the output of the machine translation unit 12 shows the unknown word as a character string (untranslated) of the original sentence in the translated sentence when the unknown word exists. Therefore, if an unknown word exists in the original sentence, morpheme analysis is performed again on the original sentence of the sentence number that includes the unknown word, and information that divides the character string of the original sentence into sentence composition units such as words and recognizes it as an unknown word. The result of the morphological analysis in which the mark that the word is unknown is added to the character string is output. The result of this morphological analysis as unknown word information,
The server document communication unit 10 is instructed to transmit it together with the document ID. The system dictionary 13 is a system dictionary used by the machine translation unit 12 at the time of translation, and the user dictionary 14 is a dictionary used by the machine translation unit 12 at the time of translation so that the user can arbitrarily register words. Retranslation file 15
Stores the sentence number of the original sentence including the unknown word found when the machine translation unit 12 is translating the original sentence of the original sentence file 11 and the document ID of the original sentence. FIG. 5 shows the data format of the retranslation file 15. Translation result file 16
Stores the translation result (target language sentence) obtained by the machine translation unit 12 translating the original sentence of the original sentence file 11, the document ID, and the sentence number.

The server document communication unit 10 of the server computer 3,
And the client document communication unit 6 of the client computer 2
Is for performing processing according to a transmission / reception message. The server document communication unit 10 of the server computer 3 includes an unknown word registration information receiving unit 10a and an unknown word information transmitting unit 10b, and the client document communication unit 6 of the client computer 2 includes an unknown word information receiving unit 6a and unknown word registration information. The transmitter 6b is provided.

Unknown word information transmission unit 10b of the server computer 3
When the machine translation unit 12 instructs the unknown word information to be transmitted, the client computer 2 obtains the unknown word information to which the morphological analysis result is added to the original sentence of the sentence number including the unknown word and the document ID of the original sentence. To the client document communication unit 6 of. FIG. 6 shows the format of the unknown word information and the document ID data transmitted from the server computer 3 to the client computer 2.

The unknown word information receiving unit 6a in the client document communication unit 6 of the client computer 2 gives the unknown word information given from the server computer 3 to the unknown word registration unit 7. The unknown word information transmission unit 6b is configured by the unknown word registration unit 7
User dictionary 1 in which information such as translated words was clarified by
The information of an unknown word in a format that can be registered in No. 4 (hereinafter referred to as unknown word registration information) and the document ID of the original sentence are stored in the server computer 3
To the server document communication unit 10. FIG. 7 shows the format of the unknown word registration information and the document ID data transmitted from the client computer 2 to the server computer 3. The unknown word registration information receiving unit 10a in the server computer 3 uses the unknown word registration information transmitted from the client computer 2 as the user dictionary 1
Register to 4.

Next, the operation of the machine translation communication system of the present invention will be described with reference to the flowchart of FIG.

The client document communication unit 6 of the client computer 2 inputs the original sentence desired to be translated by the user from the document input unit 4, and the original sentence with the document ID and sentence number stored in the original sentence file 5 is added to the server computer. Send to 3. Further, it is made clear in the header or the like that the message to be transmitted is a translation request message (S100). The original sentence, the document ID, and the sentence number received by the server document communication unit 10 of the server computer 3 are stored in the original sentence file 11 (S
101).

When the original text is stored in the original text file 11, the server document communication section 10 supplies the reception completion information of the original text to the machine translation section 12, and at this time, the machine translation section 12 uses the original text stored in the original text file 11 as a system. The translation is performed using the dictionary 13 and the user dictionary 14 (S102). As a result of the translation, the machine translation unit 12 determines whether or not the original sentence has an unknown word, that is, a word that is not registered in the system dictionary 13 and the user dictionary 14. If the original sentence includes an unknown word, the process proceeds to S108, and if not, the process proceeds to S1.
The process proceeds to 04 (S103).

First, the processing when it is determined in S103 that the original sentence does not include an unknown word will be described. The machine translation unit 12 stores the obtained translation result, the document ID and the sentence number in the translation result file 16, and then notifies the server document communication unit 10 that the storage is completed, and the server document communication unit 10 notifies the translation result. Is sent to the client computer 2 together with the document ID and the sentence number. At this time, it is made clear in the header or the like that the message to be transmitted is a translation result message (S104 to S105).

The client document communication unit 6 of the client computer 2 stores the translation result, the document ID and the sentence number in the translation result file 8 when the translation result message is received, and the translation result output unit 9 when the translation result is stored. It is activated and, based on the document ID stored in the translation result file 8, the corresponding original sentence is extracted from the original sentence stored in the original sentence file 5 and presented to the user together with the translation result (S106 to S
107).

Next, the processing when it is determined in S103 that the original sentence includes an unknown word will be described. The machine translation unit 12 stores the sentence number of the original sentence including the unknown word and the document ID of the original sentence in the retranslation file 15 (S108).
Morphological analysis is performed only on the original sentence containing the unknown word corresponding to the sentence number stored in the retranslation file 15 based on the original sentence stored in the original sentence file 11. When the morphological analysis is completed, the server document communication unit 10 is notified that the morpheme analysis is completed, and the unknown word information transmission unit 10b of the server document communication unit 10 is notified.
Is the unknown word information and document ID with the morphological analysis result
Is transmitted from the server computer 3 to the client computer 2. At this time, it is made clear in the header or the like that the electronic message to be transmitted is an unknown word information electronic message (S109).

In the client computer 2, the unknown word information receiving section 6 of the client document communication section 6 receives the document ID and the unknown word information of the unknown word information message sent from the server computer 3.
The unknown word information received by the a is the unknown word registration unit 7
Given to. The unknown word registration unit 7 presents the given unknown word information to the user. The user confirms whether or not there is a correction with respect to the morphological analysis result mark added to the original sentence, and if there is a correction, inputs a correct unknown word from the document input unit 4.
The user also inputs information necessary for registering the translated word, the part of speech, or the like in the user dictionary 14 with respect to the confirmed unknown word.
The unknown word registration unit 7 inputs the unknown word registration information thus input by the user to the client document communication unit 6 (S110).

The unknown word registration information communication unit 6b of the client document communication unit 6 assigns a document ID corresponding to the unknown word registration information and sends it to the server computer 3. At this time, it is made clear in the header or the like that the electronic message to be transmitted is an unknown word registration information electronic message (S111).

In the server computer 3, the unknown word registration information and the document ID of the unknown word registration information message transmitted from the client computer 2 are received by the unknown word registration information receiving unit 10a of the server document communication unit 10, and the unknown word is received. The registration information receiving unit 10a extracts the unknown word registration information and extracts it from the user dictionary 14
(S112). Also, retrieve the document ID,
The sentence number corresponding to this document ID is searched from the retranslation file 15, and the original sentence corresponding to this sentence number is taken out from the original sentence file 11 based on the obtained sentence number (S11).
3). The extracted original text is given to the machine translation unit 12,
The machine translation unit 12 translates only an original sentence including an unknown word using the system dictionary 13 and the user dictionary 14. At this time,
In the user dictionary 14, translated words and parts of speech have already been registered for the unknown words (S114).

In the translation result thus obtained, since all unknown words are registered in the user dictionary 14 and there are no unknown words, the machine translation unit 12 proceeds to the processing of S104 described above and is obtained. The translation result file 16
To be stored. Subsequent processing is from S105 to S10 described above.
Same as 7.

Next, the flow of the above-mentioned processing assuming that the original text shown in FIG. 3 is input will be described. Client computer 2
When the original sentence shown in FIG. 3 is input from the document input unit 4 of
As shown in FIG. 4, the document ID “001” and the sentence number “01”
To "04" are added and the original sentence corresponding to the sentence number is stored in the original sentence file 5. The client document communication unit 6 sends the translation request message stored in the original document file 5 to the server computer 3
Send to

When the server computer 3 receives the translation request message transmitted from the client computer 2, the server computer 3 stores the original sentence, the document ID, and the sentence number in the original sentence file 11.
After that, the machine translation unit 12 translates the original sentence stored in the original sentence file 11 using the system dictionary 13 and the user dictionary 14. Here, the words "TV" and "Iroha" in the original sentence in FIG.
And it is not registered in the user dictionary 14. For the unknown words “TV” and “Iroha”, as shown in the retranslation file 15 of FIG. 5, the original document ID “001” and the sentence number “01” including the unknown word “TV” are unknown words. The document ID “001” and the sentence number “03” of the original sentence including “Iroha” are stored in the retranslation file 15.

When the document ID and the sentence number are stored in the retranslation file 15, the original sentence corresponding to the sentence number is extracted from the original sentence stored in the original sentence file 11 and morphological analysis is performed. FIG. 6 shows an original sentence obtained by performing morphological analysis on the document ID “001” and the sentence numbers “01” and “03”. Here, the unknown words "TV" and "Iroha" are underlined to indicate that they are unknown words.
After that, the server document communication unit 10 transmits the document ID “001” and the original text subjected to the morphological analysis to the client computer 2 as an unknown word information electronic message.

When the client document communication section 6 of the client computer 2 receives the unknown word information message, it presents the unknown word information contained in the received message to the user. Here, "television", which was regarded as an unknown word, is an error in morphological analysis, and "high-definition television" is an unknown word. The user corrects the unknown word to “high-definition television” and inputs the information necessary to register in the user dictionary 14. As a result, the client computer 2, as shown in FIG.
Document ID and corrected unknown word "Hi-vision TV" and its translation "high-vision" and part-of-speech "Noun", and unknown word "Iroha", without correction, translated word "IROHA"
And the part of speech “noun” are transmitted to the server computer 3 as an unknown word registration information electronic message.

When the server computer 3 receives the unknown word registration information message, the unknown word "Hi-vision TV" and its translation "high-vision", the part of speech "noun" and the unknown word "included in the received message. "Iroha" and its translation "IROH
The “A” and the part of speech “noun” are registered in the user dictionary 14. In addition, from the document ID “001” included in the electronic message, the sentence numbers “01” and “0” stored in the retranslation file 15 are stored.
3 ”is obtained, the corresponding original sentence is extracted from the original sentence file 11, and the machine translation unit 12 uses the system dictionary 13 and the user dictionary 14 to perform translation. Here, since the user dictionary 14 has already registered the information necessary for translation of the unknown words “high-definition television” and “Iroha”, the unknown words do not exist, and the sentence numbers “01” and “0” are stored.
It is possible to obtain a translation result with no unknown words for the original sentence "3". As shown in FIG. 8, the machine translation unit 12 stores in the translation result file 16 the translation result and the information with the document ID “001” and the sentence numbers “01” to “04”. After that, the server document communication unit 10 transmits the translation result, the document ID, and the sentence number stored in the translation result file 16 to the client computer 2 as a translation result electronic message.

In this way, the translation result inserted in the translation result message is presented to the user together with the original sentence shown in FIG.

In the machine translation communication system according to the above embodiment, not only the number of client computers but also the number of server computers is arbitrary. For example, it may be a communication device in which a separate computer has a machine translation device for each specific field. In such a case, a certain computer may function not only as a server computer but also as a client computer. The network may be wired or wireless. When the machine translation unit 12 recognizes an unknown word, the original sentence of the sentence number including the unknown word is subjected to the morpheme analysis again, but the original sentence may be subjected to the morpheme analysis again. Further, the morphological analysis result may be always output at the first morphological analysis.
As a result of morphological analysis, a mark is added when it is determined to be an unknown word, but any mark other than an underline may be used as a mark.

[0036]

As described above, according to the present invention, when the server computer determines that an unknown word exists in the original sentence, the server computer side performs a morphological analysis on the original sentence including the unknown word to obtain an unknown word. By sending the morphological analysis information and the original text information marked against it to the client side, and presenting it to the user and correcting if there is an error in the unknown word, the unknown word is correctly recognized and registered in the dictionary. You can Further, since the mark for emphasizing the unknown word is added, it is easy for the user to understand, and it is not necessary to find the unknown word from the translation result. If an unknown word is included in the original text by assigning the document identification number and the sentence number that manages one sentence to the original document, the entire original sentence is not sent to the client computer, and the original sentence with the unknown word number is included. Since only the transmission is sent, the amount of communication until the translation result is obtained can be reduced. Furthermore, when an unknown word exists, it is necessary to perform the translation process again, but since the sentence number is given, only the original sentence in which the unknown word exists can be translated.

[Brief description of the drawings]

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

FIG. 2 is a flowchart showing the operation of the configuration diagram of the embodiment.

FIG. 3 is an explanatory diagram of an original sentence input to a document input unit according to the embodiment.

FIG. 4 is an explanatory diagram of a format of an original text file according to the embodiment.

FIG. 5 is an explanatory diagram of a format of a retranslation file according to the embodiment.

FIG. 6 is an explanatory diagram of formats of unknown word information and an identifier transmitted from the server computer to the client computer according to the embodiment.

FIG. 7 is an explanatory diagram of formats of unknown word registration information and identifiers transmitted from the client computer to the server computer according to the embodiment.

FIG. 8 is an explanatory diagram of a format of a translation result file according to the embodiment.

[Explanation of symbols]

1 ... Network, 2 ... Client computer, 3 ...
・ Server computer, 4 ・ ・ Document input unit, 5 ・ ・ Original file, 6 ・ ・ Client document communication unit, 6a ・ ・ Unknown word information receiving unit, 6b ・ ・ Unknown word registration information transmitting unit, 7 ・ ・ Unknown word registration Part, 8 ... Translation result file, 9 ... Translation result output part, 10 ... Server document communication part, 10a ... Unknown word registration information receiving part, 10b ... Unknown word information sending part, 11 ...
Original file, 12 ... Machine translation unit, 13 ... System dictionary, 14 ... User dictionary, 15 ... Retranslation file,
16 ... Translation result file.

Claims (3)

[Claims] 1. A client computer which issues a translation request, and a client computer which is connected to the client computer via a network, executes a translation process using a system dictionary and a user dictionary, and generates a translation result consisting only of translated sentences to generate the translation result. A server computer for transmitting to a client computer, wherein the server computer stores an original text transmitted from the client computer and an original text file storing an identifier assigned to the original text; and the translation processing of the original text of the original text file. When I did
When an unknown word that is not registered in the system dictionary and the user dictionary exists in the original sentence, a machine translation unit that forms unknown word information by morphologically analyzing the original sentence and emphasizes the unknown word; The server computer has a server document communication unit that sends word information and the original text identifier to the client computer, and the client computer presents the unknown word information to a user and registers the unknown word necessary for registration in the user dictionary. An unknown word registration unit that captures information, and a client document communication unit that transmits the unknown word registration information and the original text identifier to the server computer, and the unknown word registration information and the original text identifier in the server computer. When transmitted, the server document communication unit registers the unknown word registration information in the user dictionary, and the machine translation unit identifies the original sentence. A machine translation communication system, characterized in that the original sentence relating to is extracted from the original sentence file, the translation process is executed using the user dictionary in which the unknown word registration information is registered, and the translation result is generated. 2. The original sentence file stores the original sentence, an identifier given to the original sentence, and a sentence number given to one sentence of the original sentence, and the machine translation unit, when the unknown word exists, Forming the unknown word information by morphologically analyzing only the original sentence of the sentence number in which the unknown word exists, further including a retranslation file storing the sentence number in which the unknown word exists and the identifier of the original sentence, the server When the unknown word registration information and the identifier of the original sentence are transmitted to the computer, the machine translation unit extracts the corresponding sentence number from the retranslation file based on the identifier of the original sentence, and adds the sentence number to the sentence number. The machine translation system according to claim 1, wherein only the corresponding original sentence is translated. 3. The unknown word registration unit presents the unknown word information to a user, corrects the unknown word if the unknown word is corrected, and fetches unknown word registration information for the corrected unknown word. The machine translation communication system according to claim 1 or 2, characterized in that.