JP2003006455A - Translation service system - Google Patents

Abstract

(57) [Summary] [Problem] To provide translated content of WWW content at low cost. According to a content request from a user computer, a translation service server acquires content from a content server, creates and stores translated content in a machine translation unit, and transmits the translated content to the user computer.
A manual translation request notification of the content is transmitted to the translator computer, and when the translated content is manually received from the translator computer, the translated content is replaced by machine translation, and a request for the same content is issued from the user computer. In this case, the stored translated content is transmitted to the user.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a translation service system for translating content provided via the Internet and presenting it to a user.

[0002]

2. Description of the Related Art With the spread of the Internet, users can instantly obtain information on various parts of the world. However, the native language of the user who needs the information does not always match the description language of the content, and a user who understands only the native language cannot obtain information from the content described in a language other than the native language. Have difficulty.

Therefore, as one means, the user requests the translator to translate the content.
For example, Japanese Patent Laid-Open No. 2000-163423 discloses a translation support system in which a user who wants to translate content and a translator share a screen and perform translation.

However, such a request to a specific translator or collaborative work has a problem that the financial and time costs required for the translation are large.

To solve this problem, users can utilize machine translation. However, it is difficult for the current machine translation system to faithfully reflect the purpose of the original text and translate it accurately. This is because the machine translation system has no knowledge of the fields and backgrounds of the source text, but we have created machine translation environments such as machine translation dictionaries and grammar rule databases for each field to improve the content of the source text. It is possible to achieve a certain degree of accuracy improvement by switching accordingly.

For example, Japanese Unexamined Patent Publication No. 10-134052 discloses a machine translation system that automatically switches the machine translation environment according to the URL (location information) of the content, as a system for machine translation of information on the Internet. .

However, even if the machine translation environment is switched using the above conventional technique, it is difficult for the user to obtain information other than the native language with the current translation accuracy.

[0008]

On the other hand, the merit of obtaining information from the Internet is immediacy. In other words, the balance between immediacy and accuracy is important.

The contents of WWW information occupy an important position as information on the Internet.
The demand for so-called HP (homepage) described in L is very high. A content server that provides HP,
And the number of HP is also increasing in series. In addition, there are popular pages and non-popular pages on the HP, and there is not a small demand for popular pages to be viewed even if their native languages are different. On the contrary, it is rare that a user having a different native language wants to browse an unpopular page.

An object of the present invention is to provide more accurately translated content to as many people as possible in a translation service system that translates and provides Internet content.

[0011]

In order to achieve such an object, a translation service system of the present invention is a translation service server which provides a translation service, and a content management section for sharing a translation result among a plurality of users, and a user. In accordance with the request from the user, when the translation result of the content instructed by the user does not exist in the content management unit, machine translation is performed and registered in the content management unit, and at the same time, a service control unit that requests the translator to manually translate is provided. .

The service control unit divides the content into appropriate ranges such as paragraphs, and requests a plurality of translators to translate at the same time. In addition, the machine translation result is replaced each time the manual translation result is received from the translator.

As a result, the service control unit presents only the machine translation result to the user who originally requested the translation service for the certain content, and copes with the immediacy even if the translation quality is low. After that, when another user requests the same content, it can be expected that the manual translation result is partially obtained. Eventually, a manual translated result can be obtained throughout the content, which can be shared by many users. Further, when the initial user requests the same content next time, it can be expected to obtain more human-generated content having a high quality translated portion. In addition, by requesting translation at a convenient time for translators scattered all over the world, the translation service can be provided more efficiently.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the overall configuration of a translation service system of the present invention. The translation service system of the present invention includes a user computer 2 of a user who wants to browse content in his / her native language, a content server 3 for providing content, a translator computer 4 for a translator who manually translates the content, and an Internet 5 for these. It is composed of a translation service server 1 having a machine translation unit 15 as an intermediary.

For convenience of explanation, one user computer 2, one content server 3, and one translator computer 4 are described, but due to the nature of the Internet, a plurality of user computers actually connected to the Internet and contents. It goes without saying that the server and the translation service server 1 can be connected and can provide services to each.

FIG. 2 shows the detailed configuration of the translation service server 1.
Shown in. The translation service server 1 includes a service control unit 1
1, translator database 12, content management unit 13,
It is composed of a content database 14 and a machine translation unit 15. The outline of the operation of each component will be described below.

The service control unit 11 uses the Internet 5
The translated content request from the user computer is accepted via a, the content requested from the content server is acquired via the Internet 5b, and the content is delivered to the content management unit 13. The content management unit 13 determines whether the translated content of the content exists in the content database 14 and, if it exists, whether any part of the content has been changed.

According to the determination result of the content management section 13, the service control section 11 machine-translates the content by using the machine translation section 15, and registers the machine translation result in the content database via the content management section 13. A translator is selected from the translators registered in the translator database 12, and a manual translation of the content is requested via the Internet 5c. If the translation result of the content requested by the user is already registered in the content database 14, the content server 3
If there is no difference from the latest content, the service control unit 11 is notified that the translated content exists.

Based on the request from the service control unit 11, the content management unit 13 obtains the translated content from the content database 14, and then delivers it to the service control unit 11, and the service control unit 11 sends the translated content to the user computer 2. The content is transmitted via the Internet 5a. The result of manual translation by the translator is
It is sent to the service control unit 11 via the Internet 5c. The service control unit 11 registers the result in the content database 14 via the content management unit 13 in the form of overwriting the machine translation result.

In this translation service system, the user computer 2 and the translator computer 4 operate with the user and translator intervening, respectively, but the translation service server 1 and the content server 3 operate automatically without human intervention. It is configured to be. Further, the machine translation unit 15 includes machine translation means for a plurality of language pairs such as Japanese-English, English-Japanese, Korean-Japanese, etc., instead of only one language pair.

FIG. 3 shows a basic data flow among the translation service server 1, the user computer 2, the content server 3, and the translator computer 4. In FIG. 3, parenthesized numbers indicate sequence numbers.

First, when the user gives a request for translated content from the user computer 2, the location information of the content (generally called URL) and the native language type are transmitted to the translation service server 1 via the Internet 5. (Sequence 31).

The translation service server 1 requests the content server 3 to acquire the content via the Internet 5 based on the transmitted location information, and the location information is transmitted to the content server 3 (sequence 32). .
The content server 3 transmits the content to the translation service server 1 based on the location information (sequence 33).

The translation service server 1 having received the content machine-translates the content using the machine translation unit 15 corresponding to the language pair based on the language type of the content and the mother tongue received from the user, and obtains the translated content. It is sent to the user computer 2 (sequence 34). The user browses the translated content of machine translation on the user computer 2.

Subsequently, the translation service server 1 sends a translation request notification and the content to the translator computer 4 via the Internet 5 in order to request the translator to manually translate the previously acquired content (sequence). 35).

The translator, who receives the translation request notification through the translator computer 4, manually translates the content and uses the translated content by the translator computer 4.
It is transmitted to the translation service server 1 (sequence 36).

The translation service server 1, which has received the manually translated content from the translator computer 4, replaces the previously machine-translated translated content with the manually translated content.

[0028] Again, from the same user computer 2 or from another user's user computer 2 via the Internet 5, the location information of the content and the native language type are
It is transmitted to the translation service server 1 (sequence 3)
7).

If the location information sent is the same as before, the translation service server 1 sends to the user computer 2 the manually translated content previously sent by the translator (sequence 38). Here, if the location information is different, the sequence after the sequence 32 is newly performed.

It should be noted that, in the above or subsequent description, due to the nature of the Internet, users, user computers,
The content server, the translator, and the translator computer are not limited to a specific one or one.

The translated content request from the user computer 2 and the translated content sent from the translator computer 4 are asynchronous. For example, if the translation service server 2 receives a translation content request before receiving the translated content from the translator computer 4, if the same location information has been machine-translated before, it will be sent to the user. Will send the translated content by machine translation. If the amount of content is large, the time for translators to manually translate will increase proportionally,
This results in the presentation of translated content by machine translation with poor translation accuracy in response to translation content requests from many users.

This problem can be alleviated by dividing the content into predetermined units and requesting a plurality of translators to translate each smaller partial content. In addition, the content of the content server is not static, and the content is often updated with time. If the content is updated, the latest information must be translated and provided.

This problem is caused by the sequence No. Even if the location information is the same in 37, it can be solved by acquiring the content from the content server, checking whether or not there is an update, and retranslating the updated portion if the updated portion exists.

The translation service system of the present invention is designed to cope with this problem. In the following description, a translation service system incorporating these measures in the above-mentioned basic flow will be described.

In the following description, it is assumed that the user whose native language is Japanese requests the translated content for the English content shown in FIG. Further, the translation service system will be described as an interface with a user using a Web browser on the user computer 2. The current Web browser is capable of not only displaying the content but also executing the script included in the content, and the input / output of information can be performed by the interactive interaction with the user.

FIG. 4 shows the main system flow of the translation service system. When the user uses the user computer 2 to connect to the translation service server of the present system, the information input screen from the user, an example of which is shown in FIG. 6, is displayed on the user computer 2. In FIG. 6, the user's location information (generally called a URL)
After inputting, selecting the native language and pressing the execute button, a request for translated content is transmitted to the translation service server 1, and the service control unit 11 of the translation service server 1 causes the service control unit 11 of the translation service server 1 to locate the content that the user desires to translate. The information and the information in the mother tongue are acquired (S1001). Next, the service control unit 11 acquires the content of the specified location information from the content server 3 (S1002). Next, the service control unit 11 delivers the acquired content to the content management unit 13, and the content management unit 13 refers to the content database 14 and confirms whether the content has been translated as described below. . (S100
3).

FIG. 7 shows an example of the structure of the content database 14. In FIG. 7, the content database 14
Is composed of a content ID table shown in FIG. 7A and a sentence management table shown in FIG. 7B, which is uniquely identified by a composite ID of the content ID (CID) and a sentence ID described later. It is a relational database.

The content ID table (a) has a location information field for storing location information of content, a source language field indicating the language type of the source text, and a translated language field indicating the language type of the translated text, and a combination of these fields. Has a CID field for managing this as a key ID number. Further, it may have an attribute field that holds the number of times the combination of the location information, the source language, and the translated language is requested. The CID is used for the relation with the sentence management table (b). In the example of FIG. 7, the content is set assuming that the content requested by the user is as shown in FIG. The sentence management table (b) includes an original text field obtained by dividing the content into paragraphs and the like, and a translated text field for storing a corresponding translated text.
The translation is machine translation (indicated by the flag "machine")
It has a translated text flag field that identifies whether the record is manually or manually (indicated by a flag "manual"), and further identifies the ID (CID) of the content and the paragraph ID of the record in the content. have.

The content management unit 13 determines the location information requested by the user and the language type (source language) of the content,
And the content ID table (a) is searched for in the combination of the native language (translated language) specified by the user. If it does not exist as a result of the search, it is assumed that it has not been translated.
If it exists, the content is further compared with the text stored in the text management table (b). If all the texts match, it is determined that the text has been translated. This indicates that the information content has been updated, and it is determined that the information has not been translated. In this case, the record having the CID corresponding to the content in the updated existing sentence management table (b) is once deleted.

If the content has been translated, S10 is directly entered.
The processing moves to 08.

If the content has not been translated, S
In step 1005, the original text content is machine translated and registered in the content database 14. At this time, if there is no machine translation means corresponding to the requested language pair, nothing is stored in the translated text field.

Next, in S1005, it is determined whether manual translation is necessary. This is to determine whether or not to perform manual translation when there is a small number of requests from the user and to perform it when there is a certain number of reference times. When it is necessary to avoid manual translation as much as possible, it is possible to prevent the translation request from being made until the reference count reaches a certain value. For example, in FIG.
The attribute field of the content ID table in (a) may be used as a reference count, and the reference count may be incremented each time there is a reference, and if the trigger value is exceeded, manual translation may be requested.

When a manual translation is requested, it is easy to store (-1) so that it can be identified and control so as not to count up. When the trigger value is 1, when the machine translation is performed, the manual translation is always requested. If manual translation is not performed, the process proceeds to S1008.

If it is determined in S1006 that manual translation is necessary, the process advances to S1007 to request the translator to translate the original text content. Specifically, among the translators registered in the translator database 12, a translator request notification including the location information, the sentence ID, and the content of the corresponding original sentence is sent to the translators whose conditions match.

FIG. 8 shows an example of the translator database 12. The translator database 12 includes a language pair field that stores a translator's translatable language pair and an email address field that stores a translator's email address, and indicates a time convenient for the translator's translation work. It may have a time zone field or a field for storing the salary to be paid for translation and the transfer account number.

For these translators, a plurality of translators are selected for each unit (paragraph, etc.) managed by the sentence management table of the content database 14, location information, sentence ID,
The language type of the original sentence (source language), the language type of the translation destination (translated language), and the original sentence to be translated are shaped into a predetermined format and transmitted to the translator computer 4 as a translation request notification to request translation. Also, (-1) indicating that manual translation has started is set in the attribute field of the content ID table.

When the transmission of the translation request notification is completed, the result of machine translation is obtained from the content database 14 (S1008). To obtain the translated results from the content database 14, location information,
Content I is the CID specified in the source language and the target language
It is sufficient to obtain the records from the D table, obtain all the records having the CID from the sentence management table, and combine the contents of the translated sentence field in the order of the sentence ID.

Finally, the translated content is sent to the requesting user (S1009), and a series of operations is completed.

At the time of the first translation service for specific location information, only the result of machine translation is returned to the user. FIG. 10 shows a display example of translated content on the user computer 2.

In FIG. 3, unlike the above, the transmission of the translated content to the user computer 2 is processed before the translator computer 4 is notified of the translation request.
Since these two processes are independent, either process may be performed first. If it is necessary to improve the response to the user as much as possible, a process of acquiring translated content and a process of sending the translated content to the user computer 2 are added after S1005 in FIG. 4, and no manual translation is required in S1006. , Or the processing after S1007 is S
Instead of executing 1008 and S1009, the flow may be changed so as to end directly.

FIG. 5 is a processing flow of translated content sent from the translator through the translator computer 4 in response to a translation request notification. The translator translates the original text in the received formatted information into the language specified by the target language and overwrites it. In the given format,
Location information, sentence ID, which is given by the translation service server 1,
The source language and the target language are embedded, and these values are retained when the translator sends the translated content. After completing the translation, the translator uses the translator computer 4 to transmit the translated content to the translation service server 1 in a predetermined format.

When the translated content is received (S20
01), the translation service server 1 searches the content database 14 from the location information, source language, and translated language returned at the same time to acquire the CID, and further searches the translated content by machine translation using the sentence ID. Yes (S2002).

Next, the translation field of the content database 14 is replaced with the translated content received from the translator, and the translation flag field is "manual".
(S2003).

As described above, every time the translated content is received from each translator, the translated content by machine translation is replaced with the result of manual translation. If all the translators that have received the translation request notification send the translated content, all the translated texts in the content database are manually translated.

By the way, the user who browses the translated content can judge that it is unreliable if it is understood to be machine translation, and can judge that it is highly reliable if it is understood that the translation result is manually, and It can be used effectively.

The present system can provide such convenience to the user.

That is, when a request for translated content is sent from the user and the content has been translated and not updated, when the translation service server 1 reconstructs the translated content and sends it to the user. According to the value of the translated text flag field of the content database 14, information that allows the user to identify the result of machine translation or the result of manual translation for each sentence ID may be embedded in the translated content.

The information that can be identified by the user is, for example, a small pop-up window when characters are embedded like "[manual translation]" or "[machine translation]", the color is changed, or when the mouse cursor overlaps the translation. Various methods are conceivable, such as embedding a script for displaying and notifying.

For example, in the state of the sentence management table of FIG. 7B, "manual" is set in the translated sentence selection flag for CID = 0 and sentence ID = 1. In this state, CID =
When the user sends a request for translated content in Japanese to the location information content of 0 and the content has not been updated, when the translation service server 1 reconstructs the translated content, the content database According to the value of the translated text flag field, information for identifying whether it is manually or not is embedded and sent to the user computer 2.

In the user computer 2, for example, when a character is embedded, the identification information is displayed as shown by 111 and 112 in FIG. 11A, and when the script of the pop-up window is embedded, the user is in the corresponding place. When the mouse is positioned at, the identification information is displayed as indicated by 113 in FIG. 11 (b).

In the above description, if some of the contents stored in the contents server 3 have been updated,
I try to translate the whole content again, but DIF
Insertion, deletion, and change may be detected in detail by a known document comparison means such as an F tool, and only the inserted portion and the changed portion may be machine translated or translated by a translator. In this case, translation by the translator The effect of further reducing the cost can be obtained.

Further, in the above description of the embodiment, the notification to the translator computer 4 is made by mail, but it is also possible to make the notification directly to the translator computer 4 by using an IP address or the like instead of a mail address. . In this case, a unique protocol can be used between the translation service server 1 and the translator computer 4, and more advanced information other than text can be sent.

Further, in the above description of the embodiment, the translation is performed by converting the content of a certain language into the content of another language having the same meaning. However, even in the same language, for example, difficult-to-understand words are generally easy to understand. It may be expanded into a word, a redundant content may be summarized, or a combination thereof may be used. It is well known that the current natural language processing technology has made it possible to paraphrase words and summarize them by machine processing to some extent, but it has not yet reached the level of human labor, and the translation service system of the present invention is effective. Is.

[0064]

As described above, according to the translation service system of the present invention, at least a translated content by machine translation can be immediately provided to a user who first requests a certain content, and the same content can be provided after a certain amount of time. The requesting user can be provided with translated content that has been translated more accurately by the translator. Moreover, since one content is divided and a plurality of translators are requested to translate at the same time, many portions can be manually translated content in a shorter time. Further, since the translated content is shared by many users, the translated content can be provided at a lower cost.

[Brief description of drawings]

FIG. 1 is a block diagram showing a system configuration of an embodiment of the present invention.

FIG. 2 is a block diagram showing the configuration of a translation service server.

FIG. 3 is a sequence diagram showing a data flow according to the embodiment of the present invention.

FIG. 4 is a flowchart of a main process of the embodiment of the present invention.

FIG. 5 is a flowchart of translated content processing by a translator according to the embodiment of the present invention.

FIG. 6 is a diagram showing an example of a display screen of a translated content acquisition instruction on the user computer.

FIG. 7 is an explanatory diagram showing an example of a content database of a server computer.

FIG. 8 is an explanatory diagram showing an example of a translator database of a server computer.

FIG. 9 is a diagram showing an example of content to be translated.

FIG. 10 is a diagram showing a display example of translated content by machine translation.

FIG. 11 is a diagram showing a display example of translated content partially translated by a translator.

[Explanation of symbols]

1 ... Translation service server, 2 ... User computer, 3
... content server, 4 ... translator computer, 5 ... Internet, 11 ... service control section, 12 ... translator database, 13 ... content management section, 14 ... content database, 15 ... machine translation section.

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Atsuko Koizumi             1-280, Higashi Koikekubo, Kokubunji, Tokyo             Central Research Laboratory, Hitachi, Ltd. F-term (reference) 5B091 AA06 BA03 DA04 EA09

Claims (7)

[Claims] 1. A translation service in which a user computer of one or more users, one or more translator computers, a content server for delivering one or more contents, and a translation service server are connected via the Internet. In the system, the translation service server has a machine translation unit and a content holding unit, and when the translation service server receives the location information of the content desired to be translated from the user computer, the translation service server sets the designated content to the content. Obtained from the server,
The content is machine-translated using the machine translation unit, the machine-translated content and the location information of the content are held in the content holding unit, and the translated content held in the content holding unit is stored in the user computer. Then, the user computer presents the received translated content to the user, and at the same time, the translation service server sends the translator computer a translation request notification to the translator along with the content and the location information. A translation service system characterized by that. 2. When the translation service server receives translated content manually by a translator and location information from the translator computer, the translated content is held in the content holding means and identified by the location information. The translation service system according to claim 1, wherein 3. The translation service server, when the location information of the content desired to be translated is received from an arbitrary user computer, and the location information matches the location information held in the content holding means. 3. The translation service system according to claim 1, wherein the translated content held in the content holding means is transmitted to the user computer. 4. The translation service server divides the content acquired from the content server into predetermined units, machine-translates each divided partial content using the machine translation means, and machines each of the partial contents. The translated content, the location information of the content, and the division sequence number are held in the content holding means, and the content, the location information, and the division sequence number are translated to different translator computers for each divided partial content. When a translation request notification to a translator is transmitted and the translated content manually by the translator, the location information, and the division sequence number are received from the translator computer, the location information and the location information stored in the content holding unit are received. Replacing translated content identified by a split sequence Translation service system according to claim 1, wherein. 5. The translation service server receives the translated content manually translated by the translator, the location information, and the division sequence number from the translator computer, and the content holding means holds the location information and the division information. The translation service system according to claim 4, wherein when replacing the translated content identified by the sequence, an identifier indicating that the translation is manually performed by a translator is held in the content holding unit. 6. The translation service server, when receiving the location information of the content desired to be translated from an arbitrary user computer, if the location information matches the location information held in the content holding means. Characterized in that the translated content held by being divided by the content holding means is reconfigured into the whole translated content corresponding to the location information based on the division sequence number, and is transmitted to the user computer. The translation service system according to claim 5. 7. The translation service server, when receiving the location information of the content desired to be translated from an arbitrary user computer, if the location information matches the location information held in the content holding means. , The translated content held by the content holding means is manually translated by the translator when reconstructing the entire translated content corresponding to the location information based on the division sequence number. 7. The translation service system according to claim 6, wherein the translated content is provided with information by which the user can identify the translation result by machine translation or the translation result by the translator for each division unit by referring to the identifier indicating .