Electronic Message Translation and Analysis System

Abstract

To provide an electronic message translation and analysis system which automatically translates an electronic message and analyses contents of the electronic message.SOLUTION: An electronic message translation and analysis system 1 includes a translation and analysis server 9 which translates an original sentence of a mail and analyses the original sentence. The translation and analysis server 9 includes: a translation unit 916 which translates the original sentence of the mail; and a spoof analysis unit 929 which estimates a probability of spoofing of the original sentence creator by analyzing the original sentence while using at least a Kanji ratio, a hiragana ratio, and a Term Frequency-Inverse Document Frequency (TF-IDF).SELECTED DRAWING: Figure 12

Description

[Technical Field]

[0001] The present invention relates to an electronic message translation and analysis system that translates the text of an electronic message and analyzes the text. [Background technology]

[0002] With the advancement of globalization, the exchange of information with other countries using electronic messages such as email and chat has increased. On the other hand, there has also been the problem of communication being difficult when the sender and receiver of information speak different languages. Various translation systems have been proposed as a solution to this problem, such as the email translation system described in Patent Document 1 below. This translation system associates a user's email address and language pair with the recipient's email address and language pair and registers them in advance. For emails sent by the user, addressed to the user, or both, the translation system finds the sender and recipient email address pair from the user registration, identifies the translation language, and performs translation. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] Japanese Patent Application Laid-Open No. 2002-24141 Summary of the Invention [Problem to be solved by the invention]

[0004] However, while the invention of Patent Document 1 facilitates communication through translation, there is a risk that the nuances of the original text may be lost during the process of translating an e-mail addressed to a user, making it impossible to accurately read the recipient's feelings from the translated text. There is also a risk that the original text of an e-mail sent by a user may contain nuances that the user did not intend, and if these are translated and sent as is, the recipient may be given a wrong impression.

[0005] Furthermore, even if an e-mail is created and sent by impersonating a sender who owns the e-mail address of the sender of the e-mail, there is a risk that the impersonation may be difficult to detect due to the translation process.

[0006] There has been a demand for an electronic message translation analysis system that can analyze electronic messages such as emails and chats in order to detect problems that could potentially occur as a negative side of translation.

[0007] The present invention has been made in view of the above circumstances, and an object of the present invention is to provide an electronic message translation and analysis system that automatically translates electronic messages and analyzes the contents of the electronic messages. [Means for solving the problem]

[0008] In order to solve this problem, the invention described in claim 1 is an electronic message translation and analysis system that translates and analyzes the text of an electronic message, an original message receiving unit that receives the electronic message to be translated and analyzed; a translation unit that translates the original text, which is the body of the electronic message and is written in a source language, into a translation language; and a spoofing analysis unit that analyzes the original text using at least a kanji ratio, a hiragana ratio, and TF-IDF to estimate the probability of spoofing the creator of the original text. a providing unit that provides the translation of the original text translated by the translation unit and the probability of spoofing estimated by the spoofing analysis unit to a terminal that is a destination of the electronic message or a terminal that created the electronic message; The present invention is characterized by comprising:

[0009] The invention of claim 2 is an electronic message translation and analysis system for translating and analyzing the text of an electronic message, comprising: an original message receiving unit for receiving the electronic message to be translated and analyzed; a translation unit for translating the original text, which is the text of the electronic message and is written in an original language, into a translation language; a text analysis unit including an impersonation analysis unit for analyzing the original text using at least a kanji ratio, a hiragana ratio, and TF-IDF to estimate the probability of impersonation of the original text creator; and a message creation screen providing unit for providing an electronic message creation screen to a client terminal via a communication line, wherein after a message to be sent is created as the electronic message on the electronic message creation screen displayed on the client terminal, a pre-sending confirmation request sent from the client terminal is and when a request to send the outgoing message is received by the message creation screen providing unit, the outgoing message created on the client terminal is received by the original message receiving unit, the translation unit translates the received outgoing message into a translation, the sentence analysis unit estimates the probability of spoofing based on the received outgoing message, and a pre-sending confirmation screen displaying the estimated probability of spoofing is provided to the client terminal, and after the pre-sending confirmation screen is provided, when a request to send the outgoing message from the client terminal is received by the message creation screen providing unit, the translation of the outgoing message is sent as a outgoing message document to a destination electronic address of the outgoing message document.

[0010] The invention of claim 3 is characterized in that, in addition to the configuration of claim 2, the pre-send confirmation screen provided to the client terminal includes the translation of the outgoing message translated by the translation unit along with the analysis result of the sentence analysis unit, and when the translation included in the pre-send confirmation screen displayed on the client terminal is corrected from the client terminal, the translation translated by the translation unit is rewritten with the corrected translation corrected on the pre-send confirmation screen, and when the transmission request for the outgoing message from the client terminal is accepted by the message creation screen providing unit, the rewritten corrected translation is sent as the outgoing message document to the destination electronic address of the outgoing message document.

[0011] The invention of claim 4 is characterized in that, in addition to the configuration described in claim 2, the outgoing message document sent to the destination electronic address is configured to display the analysis results of the sentence analysis unit in addition to the translated text.

[0012] The invention of claim 5 is characterized in that, in addition to the configuration described in claim 2, the outgoing message document sent to the destination electronic address does not include the analysis results of the sentence analysis unit but is composed of the translated text.

[0013] The invention of claim 6 is an electronic message translation and analysis system for translating and analyzing the text of an electronic message, comprising: an original message receiving unit for receiving an original message, which is the electronic message to be translated and analyzed; a translation unit for translating the original text, which is the text of the original message and is written in an original language, into a translation language; a text analysis unit including an impersonation analysis unit for analyzing the original text using at least a kanji ratio, a hiragana ratio, and TF-IDF to estimate the probability of impersonation of the creator of the original text; and a text analysis unit for analyzing the original text, the translation of the original text translated by the translation unit, and the translation estimated by the text analysis unit. The system comprises an aggregated information construction unit that constructs message aggregated information by aggregating messages to include the probability of spoofing, and a translation analysis control unit that controls the operation of translating and analyzing the original messages received by the original message receiving unit to construct the message aggregated information, wherein the translation analysis control unit inputs the original text to the translation unit to create the translation, inputs the original text to the sentence analysis unit to estimate the probability of spoofing, and controls the original text, the created translation, and the estimated probability of spoofing to be input to the aggregated information construction unit to construct the message aggregated information.

[0014] The invention according to claim 7 is characterized in that, in addition to the configuration according to any one of claims 1 to 6, the electronic message is an email.

[0015] The invention according to claim 8 is characterized in that, in addition to the configuration according to any one of claims 1 to 6, the electronic message is a chat. [Effects of the Invention]

[0016] According to the invention of claim 1, the original text of an electronic message can be translated, and the original text can be analyzed to estimate the probability of impersonation of the original creator. The translation of the original text of the electronic message and the estimated probability of spoofing are provided to the terminal to which the electronic message is addressed or the terminal that created the electronic message, so that the user can check the translation and the probability of spoofing for the electronic message he or she has received, and can also check the translation and the probability of spoofing for the electronic message he or she has created.

[0017] According to the invention of claim 2, the original text of an electronic message is translated and analyzed to estimate the probability of impersonation of the original creator. Therefore, the estimated probability of impersonation of the message to be sent can be confirmed before sending the message. The message creator can send the message after confirming the probability of impersonation.

[0018] According to the invention of claim 3, the translation of the message to be sent can be manually corrected from the client terminal.

[0019] According to the invention of claim 4, the outgoing message document is configured to display the translation and analysis results and is sent to the destination electronic address, so that the translation and analysis results can be confirmed on the terminal corresponding to the destination electronic address that receives the outgoing message document.

[0020] According to the invention of claim 5, a message document consisting of a translation and not including the analysis results is sent to the destination electronic address. Therefore, even if the terminal receiving the message document cannot display the analysis results, the translation can be checked without any problems.

[0021] According to the invention of claim 6, the original text of an electronic message is translated and analyzed to estimate the probability of impersonation of the original text creator. Furthermore, the translation of the original text and the estimated probability of impersonation are aggregated as message aggregation information. Therefore, the translation and the probability of impersonation can be extracted from this message aggregation information and displayed on a client terminal.

[0022] According to the invention of claim 7, an e-mail translation and analysis system is configured that translates and analyzes e-mails.

[0023] According to the invention of claim 8, a chat translation and analysis system is configured that translates and analyzes chats. [Brief explanation of the drawings]

[0024] [Figure 1]1 is a block diagram showing the configuration of an email system including an email translation and analysis system in accordance with Embodiment 1 of the present invention. [Figure 2] FIG. 2 is a schematic block diagram of an InBound SMTP server that transfers received emails to the email translation analysis system according to the first embodiment. [Figure 3] 2 is a schematic block diagram showing the configuration of a mail gateway server according to the first embodiment. FIG. [Figure 4] 3 is a diagram showing an example of the configuration of personal setting information according to the first embodiment. FIG. [Figure 5] FIG. 10 is a diagram showing an example of a personal setting information registration / display screen according to the first embodiment. [Figure 6] FIG. 2 is a diagram showing an example of the configuration of email aggregation information according to the first embodiment. [Figure 7] 2 is a schematic block diagram showing the configuration of a mail server according to the first embodiment. FIG. [Figure 8] 2 is a schematic block diagram showing the configuration of a WEB / APP server according to the first embodiment. FIG. [Figure 9] 1A is a diagram showing an example of the configuration of address information according to the first embodiment, and FIG. 1B is a diagram showing an example of the configuration of an address book storage unit. [Figure 10] FIG. 10 is a diagram showing an example of an address registration screen according to the first embodiment. [Figure 11] FIG. 2 is a schematic block diagram of an OutBound SMTP server that receives outgoing emails from the email translation analysis system according to the first embodiment and transmits them via the Internet. [Figure 12] 2 is a schematic block diagram showing the configuration of a translation analysis server according to the first embodiment. FIG. [Figure 13] (a) is a flowchart showing the process of generating a trained translation estimation model, and (b) is a flowchart showing the process of creating a translation from an original email. [Figure 14](a) is a flowchart showing the process of generating a trained emotion estimation model, and (b) is a flowchart showing the process of estimating emotion levels from original emails. [Figure 15] (a) is a flowchart showing the process of generating a trained misspelling and omission detection model, and (b) is a flowchart showing the process of detecting misspellings and omissions from original emails. [Figure 16] (a) is a flowchart showing the process of generating a trained politeness level estimation model, and (b) is a flowchart showing the process of estimating the politeness level from an original email. [Figure 17] (a) is a flowchart showing the process of generating a trained harassment level estimation model, and (b) is a flowchart showing the process of estimating a harassment level from an original email. [Figure 18] 10A is a flowchart showing the process of generating a trained spoofing probability estimation model, and FIG. 10B is a flowchart showing the process of estimating the spoofing probability from an original email. [Figure 19] FIG. 2 is a diagram showing an example of a received email display screen according to the first embodiment. [Figure 20] FIG. 2 is a diagram showing an example of an email creation screen according to the first embodiment. [Figure 21] FIG. 10 is a diagram showing an example of a translation preview screen of a composed email according to the first embodiment. [Figure 22] FIG. 10 is a diagram showing an example of a status screen display showing the analysis results of an email. [Figure 23] 10 is a diagram showing a schematic flowchart of the process up to the construction and storage of email aggregation information of received emails in the email translation and analysis system according to the first embodiment. FIG. [Figure 24] 10 is a diagram showing a schematic flowchart when the contents of email aggregation information stored in the email server are displayed on a client terminal in the email translation and analysis system according to the first embodiment. FIG. [Figure 25]10 is a diagram showing a schematic flowchart of the process from when a user creates a mail document to when the mail document is sent to a destination mail address in the email translation and analysis system according to the first embodiment. FIG. [Figure 26] FIG. 10 is a block diagram showing the configuration of a chat system including a chat translation and analysis system in accordance with a second embodiment of the present invention. [Figure 27] FIG. 10 is a schematic block diagram showing the configuration of a chat gateway server according to the second embodiment. [Figure 28] FIG. 10 is a diagram showing an example of the configuration of personal setting information according to the second embodiment. [Figure 29] FIG. 10 is a diagram showing an example of a personal setting information registration and display screen according to the second embodiment. [Figure 30] FIG. 11 is a diagram showing an example of the structure of message aggregation information according to the second embodiment. [Figure 31] 10 is a schematic block diagram showing the configuration of a WEB / APP server according to the second embodiment. FIG. [Figure 32] 10(a) is a diagram showing an example of the configuration of address information according to the second embodiment, and FIG. 10(b) is a diagram showing an example of the configuration of an address book storage unit. [Figure 33] FIG. 10 is a diagram showing an example of an address information registration screen according to the second embodiment. [Figure 34] 10 is a schematic block diagram showing the configuration of a translation analysis server according to the second embodiment. FIG. [Figure 35] FIG. 10 is a diagram showing an example of a chat display / creation screen according to the second embodiment. [Figure 36] FIG. 10 is a schematic flowchart showing a process from when a user creates a message text to when the message text is sent to a group chat address in the chat translation and analysis system according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0025] [First embodiment of the invention] A first embodiment of the present invention will be described with reference to FIGS. 1 to 25. FIG.

[0026] 1 is a block diagram showing the configuration of an electronic mail system including an electronic mail translation analysis system. The electronic mail translation analysis system 1, which serves as an "electronic message translation analysis system" according to the present invention, includes a mail gateway server 4, a mail server 5, a web / application (WEB / APP) server 6, and a translation analysis server 9. The mail gateway server 4, mail server 5, and web / app server 6 are connected via a local area network (LAN) 8, and the mail gateway server 4 and web / app server 6 are connected to the translation analysis server 9 via the Internet 12.

[0027] Furthermore, organization 2 that provides services using this email translation analysis system 1 is equipped with an InBound SMTP (InBound Simple Mail Transfer Protocol) server 3, an OutBound SMTP (OutBound Simple Mail Transfer Protocol) server 7, and the like, in addition to the components of email translation analysis system 1 excluding translation analysis server 9. This email system 100 is configured such that client terminals 101, 102, 103, ..., where each user views, creates, and distributes emails, which are "electronic messages," and an external mail server 11 that sends emails from outside to organization 2's InBound SMTP server 3, are connected to the components installed in organization 2 via the Internet 12 as a "communication line."

[0028] The email translation analysis system 1 according to the present invention may be configured to include an InBound SMTP server 3 and an OutBound SMTP server 7. Furthermore, a translation analysis server 9 may be connected to a LAN 8 and installed within the organization 2.

[0029] A user of this email translation and analysis system 1 accesses a mailer running on the WEB / APP server 6 through a browser installed on client terminals 101, 102, 103, ... (hereinafter referred to as "client terminal 10") that use information devices such as PCs (Personal Computers) and smartphones, and views sent and received emails, composes email documents, and sends email documents. A mailer is web application software that provides the client terminal with functions such as displaying received emails addressed to the user, composing email documents by the user, and sending the created email documents to the other party.

[0030] By using this translation analysis system 1, it is possible to automatically translate received emails written in a foreign language into the user's language, and to translate email documents created by the user into the language of the other party. It is also possible to analyze received emails to estimate the emotions of the email creator and the probability of impersonation, and to analyze the emotions, detect typos, analyze the level of politeness, analyze the level of harassment, and analyze the probability of impersonation for emails created by the user. [InBound SMTP Server] The InBound SMTP server 3 receives e-mail delivered via the Internet 12 and forwards it to the mail gateway server 4. As shown in the schematic block diagram of the InBound SMTP server 3 in Figure 2, this server 3 is configured to include an InBound SMTP server control unit 300, a mail receiving unit 302, and a mail forwarding unit 304. The InBound SMTP server control unit 300 is made up of a CPU (not shown) that executes programs and performs arithmetic processing, a ROM (not shown) that is a non-volatile memory that constitutes an auxiliary storage device that stores programs and data, and a RAM (not shown) that is a volatile memory that operates as a work area for the CPU's program execution and arithmetic processing, and receives e-mail at the mail receiving unit 302 and controls the operation of forwarding the mail from the mail forwarding unit 304 to the mail gateway server 4 via the LAN 8 within the organization 2. [Mail gateway server] The mail gateway server 4 accepts incoming mail transferred from the InBound SMTP server 3, inputs the text of this mail to the translation and analysis server 9 (described later) for translation and analysis, and generates mail aggregation information 460 as "message aggregation information" that aggregates the original text, translation, analysis results, etc., which are the text of this mail. As shown in the schematic block diagram of the mail gateway server 4 in FIG. 3, this server 4 is configured to include a mail gateway server control unit 400, an incoming mail reception unit 402, an address information acquisition unit 404, a language identification unit 406, an incoming mail transmission unit 408, a language notification unit 410, an author information notification unit 411, an analysis content selection unit 412, an incoming mail translation reception unit 414, an audio information reception unit 416, an analysis result reception unit 418, an incoming mail aggregation information composition unit 420, an incoming mail aggregation information transmission unit 422, a personal setting information registration / display screen provision unit 424, a personal setting information storage unit 426, a personal setting information request reception unit 428, and a personal setting information provision unit 430.

[0031] The mail gateway server control unit 400 is composed of a CPU, ROM, and RAM (not shown), and accepts incoming mail, identifies the language of the body of the received mail and the translation language, performs translation and analysis, and controls the operation of creating mail aggregation information 460 based on the results. Therefore, when accepting incoming mail, this mail gateway server control unit 400 operates as a "translation and analysis control unit."

[0032] The received mail reception unit 402 receives received mail transferred from the InBound SMTP server 3. For this received mail, which is an original mail as an "original message" to be translated and analyzed, the received mail reception unit 402 corresponds to the "original message reception unit."

[0033] The address information acquisition unit 404 acquires address information 660 for each sender and user from the address book storage unit 638 of the WEB / APP server 6. As will be described in detail later, the address book storage unit 638 stores an address book 670 for each user, and this address book 670 stores destinations registered by the user, the email addresses of the users, and the language of the destinations as address information 660. The address information acquisition unit 404 sends the email address of the sender of the received email and the email address of the user who is the destination, together with a request to provide address information, to the WEB / APP server 6, and acquires address information 660 from the server 6 that matches these email addresses.

[0034] The language identification unit 406 identifies the language of the received email from the sender's address information 660 acquired by the address information acquisition unit 404, and identifies the translation language from the user's address information 660. For received emails, this language identification unit 406 operates as an "original language identification unit" that identifies the original language of the body of the original email, and as a "translation language identification unit" that identifies the translation language.

[0035] The received email sending unit 408 inputs the received email to the translation and analysis server 9 via the Internet 12 for translation and analysis. In addition, the language notification unit 410 inputs the original language and translation language of the received email identified by the language identification unit 406 to the translation and analysis server 9. If there are multiple translation languages, the multiple translation languages ​​are input.

[0036] The creator information notification unit 411 inputs the sender email address of the received email as "creator information" that identifies the creator of the received email text to the translation analysis server 9. The translation analysis server 9 stores the original creator in association with the email address, so it is possible to identify the original creator from the input sender email address.

[0037] The analysis content selection unit 412 reads and selects analysis content for the received email in the translation analysis server 9 from personal setting information 438 registered in advance by the user, and causes the translation analysis server 9 to analyze it.

[0038] FIG. 4 shows an example of the configuration of personal setting information 438, which allows the user to register receiving settings 440 for accepting incoming emails and sending settings 445 for emails created by the user. Translation settings 441 and 446 have three options: translate while preserving the original text, translate without preserving the original text, and do not translate. Emotion level checks 442 and 447 can be enabled or disabled and set whether to analyze the emotions of the email creator based on the original email. Spoofing probability checks 443 and 451 set whether to analyze the probability that the creator of the original email is not the same as the pre-registered original creator. Typo / misspelling check 448 sets whether to detect the content of typos and the location of typos in the original email. Politeness level check 449 sets whether to analyze the politeness of the original email, and harassment level check 450 sets whether to analyze the harassment level of the original email.

[0039] In this example, the receiving settings 440 do not include a check for typos, politeness level, or harassment level, but it is possible to register these settings in the same way as the sending settings 445, and analyze received emails for typos, politeness, and harassment level.

[0040] The personal setting information 438 registered by the user is stored in the personal setting information storage unit 426 shown in Fig. 3. The personal setting information registration / display screen providing unit 424 provides a screen for displaying and registering the personal setting information 438 to the client terminal 10 via the Internet 12. Fig. 5 shows an example of a personal setting information registration / display screen 456, and when the user opens this screen 456, the registered personal setting information 438 is displayed by checking a check box. When the user wants to register, he or she checks a check box on this screen 456 and clicks a register button 457.

[0041] In addition, when the personal setting information request receiving unit 428 shown in Figure 3 receives a provision request from the WEB / APP server 6, the requested personal setting information 438 is read from the personal setting information storage unit 426 and provided to the WEB / APP server 6 from the personal setting information providing unit 430.

[0042] The received email translation reception unit 414, audio information reception unit 416, and analysis result reception unit 418 respectively receive the translation of the original text that is the body of the received email, audio information that enables the translation to be played aloud, and the analysis results of the original text of the received email from the translation and analysis server 9.

[0043] The received mail aggregation information composition unit 420 aggregates the original text, translation, and analysis results of received mail to compose the mail aggregation information 460. For received mail, this received mail aggregation information composition unit 420 operates as the "aggregated information composition unit."

[0044] 6, a sender email address area 461 stores the sender email address of the email, and a destination email address area 462, a CC (Carbon Copy) email address area 463, and a BCC (Blind Carbon Copy) email address area 464 store the destination email address, CC email address, and BCC email address of the email. A subject area 465 stores the translated subject and the original subject. Furthermore, an original text area 466, a translation text area 467, an audio information file area 468, an audio information URL (Uniform Resource Locator) area 469, an original text language area 470, and a translation text language area 471 store the original text of the email, its translation, an audio information file that enables the translation to be played aloud, a URL for playing the audio information file, the language of the original text, and the language of the translation, respectively. If different translation languages ​​are set for the destination email address, CC email address, and BCC email address, multiple types of translations and audio information files corresponding to those translation languages ​​are stored in the translation text area 467 and audio information file area 468, etc.

[0045] The emotion level area 472, typo information area 473, politeness level area 474, harassment level area 475, and spoofing probability area 476 store the emotion of the original email, typo information detected from the contents of typos and the positions of typos, the politeness level, the harassment level, and the analysis results of the probability that the creator of the original email is not the person in question.

[0046] If the translation settings 441, 446 (see FIG. 4) of the personal setting information 438 are set to not leave the original text, the original text of the email may not be stored in the original text area 466 of the email aggregation information 460. Also, if the translation settings 441, 446 are set to not translate, no translation is created, so the translation text area 467 of the email aggregation information 460 becomes empty and no data is stored.

[0047] The received mail aggregation information sending unit 422 shown in FIG. 3 sends the received mail aggregation information 460 of received mails configured by the received mail aggregation information configuration unit 420 to the mail server 5 for storage. [Mail Server] The mail server 5 stores mail aggregation information 460 and provides the mail aggregation information 460 upon request. This mail server 5 is an IMAP (Internet Message Access Protocol) system, and manages emails while storing them on this server 5. A user accesses the server 5 from a client terminal 10 to perform operations such as viewing emails on this server 5. As shown in the schematic block diagram of the mail server 5 in FIG. 7, this server 5 is configured to include a mail server control unit 500, a received mail aggregation information acceptance unit 502, a received mail aggregation information storage unit 504, a created mail aggregation information acceptance unit 506, a created mail aggregation information storage unit 508, a received mail information request acceptance unit 510, a received mail aggregation information provision unit 512, a created mail information request acceptance unit 514, and a created mail aggregation information provision unit 516.

[0048] The mail server control unit 500 is composed of a CPU, ROM, and RAM (not shown), and controls the operation of accepting mail aggregation information 460 of received mails and storing it in the received mail aggregation information storage unit 504, and accepting mail aggregation information 460 of mails created by the user and storing it in the created mail aggregation information storage unit 508. The control unit 500 also controls the operation of accepting a request and providing the stored mail aggregation information 460 to the request source.

[0049] The received mail aggregation information receiving unit 502 receives the mail aggregation information 460 of received mail sent from the mail gateway server 4, and the received mail aggregation information storage unit 504 stores this mail aggregation information 460. For received mail, the received mail aggregation information storage unit 504 functions as the "aggregated information storage unit."

[0050] The created email aggregation information receiving unit 506 is configured with the WEB / APP server 6 as described below, and receives email aggregation information 460 of emails created by users that is sent from the server 6. Then, the created email aggregation information storage unit 508 stores the email aggregation information 460 received by the created email aggregation information receiving unit 506. In this way, the created email aggregation information storage unit 508 functions as an "aggregated information storage unit" for emails created by users.

[0051] The received email information request receiving unit 510 receives a request to provide email aggregation information 460 about received emails sent from the WEB / APP server 6, and the received email aggregation information providing unit 512 provides the email aggregation information 460 corresponding to the provision request read from the received email aggregation information storage unit 504 to the WEB / APP server 6.

[0052] In a similar manner, the created email information request receiving unit 514 receives a request to provide email aggregation information 460 for user-created emails sent from the WEB / APP server 6, and the created email aggregation information providing unit 516 provides the email aggregation information 460 corresponding to the provision request read from the created email aggregation information storage unit 508 to the WEB / APP server 6. [WEB / APP server] When the WEB / APP server 6 receives a read request from a client terminal 10 so that the user can view sent and received emails, it provides the requesting terminal 10 with an email display screen or an email composition screen so that the user can compose an email document. The server 6 also inputs emails composed by the user into the translation and analysis server 9, aggregates the results, and creates email aggregation information 460. The server 6 then displays the contents of this email aggregation information 460 on the client terminal 10 so that the user can check the translations and analysis results. Furthermore, when a send request is received from the user's client terminal 10, it generates an outgoing email document as a "sent message document" based on the email aggregation information 460 and sends it to the Outbound SMTP server 7 for delivery to the destination email address.

[0053] As shown in the schematic block diagram of the WEB / APP server 6 in Figure 8, this server 6 is configured to include a WEB / APP server control unit 600, an email display / creation screen providing unit 602, a received email information request sending unit 604, a received email aggregation information acquisition unit 606, a personal setting information acquisition unit 608, a language identification unit 610, a created email sending unit 612, a language notification unit 614, a creator information notification unit 615, an analysis content selection unit 616, a created email translation receiving unit 618, a voice information receiving unit 620, an analysis result receiving unit 622, a created email aggregation information configuration unit 624, an outgoing email document generation unit 626, an outgoing email sending unit 628, a created email aggregation information sending unit 630, a created email information request sending unit 632, a created email aggregation information acquisition unit 634, an address registration / display screen providing unit 636, an address book storage unit 638, an address information request receiving unit 640, and an address information providing unit 642.

[0054] The WEB / APP server control unit 600 is composed of a CPU, ROM, and RAM (not shown), and controls operations such as providing a mail display screen configured based on the mail aggregation information 460 to the client terminal 10 and providing a mail creation screen. Therefore, the WEB / APP server control unit 600 corresponds to an "information provision control unit" that controls operations for providing the mail aggregation information 460 via the Internet 12.

[0055] The control unit 600 also accepts emails created by users, identifies the language of the created email and the translation language, translates and analyzes the email, and controls the operation of composing email aggregation information 460 based on the results. Therefore, when accepting emails created by users, the WEB / APP server control unit 600 operates as a "translation and analysis control unit."

[0056] Furthermore, the control unit 600 also controls the operation of generating an outgoing mail document based on the mail aggregation information 460 of user-created mail and sending it to a destination mail address.

[0057] The email display / creation screen providing unit 602 creates an email display screen to be displayed on the client terminal 10 based on the email aggregation information 460 in order to view received emails or emails that the user has created and sent, and provides this to the terminal 10 via the Internet 12. Therefore, the email display / creation screen providing unit 602 operates as an "aggregated information providing unit" that provides the email aggregation information 460 via the Internet 12.

[0058] When a user views a received email or an email that the user has created and sent, the user specifies the email to be displayed on the email display screen. This specification causes the client terminal 10 to request the WEB / APP server 6 via the Internet 12 to read the email aggregation information 460 for the specified email. When this read request is made, the email aggregation information 460 for the specified email is read from the mail server 5 and provided to the requesting client terminal 10. That is, the contents of the email, such as the translation, original text, and analysis results, are displayed on the client terminal 10 based on the read email aggregation information 460.

[0059] The email display / creation screen providing unit 602 also provides an email creation screen so that the user can create an email document using the client terminal 10. The WEB / APP server 6 accepts the email created by the user via this email creation screen. For this user-created email, which is the original email as the "original message" to be translated and analyzed, the email display / creation screen providing unit 602 corresponds to the "original message accepting unit."

[0060] As described above, the WEB / APP server 6 creates mail aggregation information 460 based on the received user-created mail, and the contents of this mail aggregation information 460 are provided and displayed on the client terminal 10 that created the created mail from the mail display / creation screen provider 602 via the Internet 12. When the user checks the contents on the mail creation screen and issues a transmission request, the transmission request is sent to the WEB / APP server 6 via the Internet 12. This server 6 creates a mail document to be sent based on the mail aggregation information 460, and sends this mail document as a mail to be sent to the destination address.

[0061] When there is a request from the client terminal 10 to read the mail aggregation information 460 of received mail, the received mail information request sending unit 604 sends a provision request to the mail server 5. In response to this provision request, the received mail aggregation information acquisition unit 606 acquires the mail aggregation information 460 of received mail sent from the mail server 5. The content of the acquired mail aggregation information 460 is displayed on the mail display screen of the client terminal 10.

[0062] To check the translation and analysis results of an email document created by a user, the user sends a request to create email aggregation information 460 from the email creation screen to this WEB / APP server 6. This server 6 specifies the translation language and selects the analysis content, and inputs the user-created email into the translation and analysis server 9 for translation and analysis.

[0063] The personal setting information acquisition unit 608 sends a request to the mail gateway server 4 to acquire the user's personal setting information 438 (see FIG. 4). From this personal setting information 438, the analysis content for the created email can be read.

[0064] The language identification unit 610 acquires address information 660 for each user and the recipient of the created email from the address book storage unit 638, identifies the original language of the body of the created email from the user's address information 660, and identifies the translation language from the recipient's address information 660. For emails created by a user, this language identification unit 610 operates as an "original language identification unit" that identifies the language of the original email and a "translation language identification unit" that identifies the translation language.

[0065] 9(a) shows an example of the configuration of address information 660, which is configured to include an email address 662 as an "electronic address" that identifies the user of the email, a name 664 corresponding to the email address 662, and an address-compatible language 666 that is the language corresponding to the address 662. As shown in FIG. 9(b), an address book 670 is configured from multiple destination address information 672 registered by the user and the user's own user address information 674. The address book 670 is configured for each user and is stored in the address book storage unit 638 (see FIG. 8).

[0066] To read address information 660 from address book storage unit 638, first, in order to identify the user's address book, the user's email address is compared with the email addresses 662 stored in user address information 674, and a matching address book 670 is extracted. User address information 674 is identified from this address book 670. Next, the email address of the destination to be read is compared with the email addresses 662 of multiple destination address information 672 registered in this address book 670, and matching destination address information 672 is extracted. Address information 660 of the destination to be read is identified from this extracted destination address information 672.

[0067] By specifying the user address information 674 and the destination address information 672 in this way, it is possible to read the address correspondence language 666 registered in each address information 660.

[0068] For user-created emails, the sender is the user, so the address-compatible language 666 in the user address information 674 becomes the language of the original email. The address-compatible language 666 in the destination address information 672 becomes the translation language. On the other hand, for received emails, the destination of the email is the user, so the address-compatible language 666 in the user address information 674 becomes the translation language. The address-compatible language 666 in the destination address information 672 corresponding to the sender of the received email becomes the language of the original email.

[0069] 8 inputs a user-created email to the translation and analysis server 9 via the Internet 12 for translation and analysis. In addition, the language notification unit 614 inputs the original language and translation language of the body of the created email identified by the language identification unit 610 to the translation and analysis server 9.

[0070] Creator information notification unit 615 inputs the user's email address as "creator information" that identifies the user who created the body of the user-created email to translation analysis server 9. Since translation analysis server 9 stores the original text creator in association with the email address, it is possible to identify the user who created the original text from the input user's email address.

[0071] The analysis content selection unit 616 reads and selects the analysis content of the user-created email registered as the sending settings 445 of the personal setting information 438 acquired by the personal setting information acquisition unit 608 , and causes the translation analysis server 9 to analyze it.

[0072] The translation of email received by the translation analysis server 9 is received by the translation analysis server 9. The translation of the email body is received by the translation analysis server 9. The voice information is received by the voice information receiving unit 620. The analysis result is received by the analysis server 9. The translation of the email body is received by the translation analysis server 9. The voice information is received by the analysis server 9. The analysis result is received by the analysis server 9. The analysis result is received by the translation ...

[0073] The created email aggregation information construction unit 624 consolidates the original text, translation, and analysis results of user-created emails to construct the email aggregation information 460 shown in Fig. 6. For user-created emails, this created email aggregation information construction unit 624 operates as the "aggregated information construction unit."

[0074] When a user requests to send a created email, an outgoing email document generation unit 626 serving as an "outgoing message generation unit" shown in Fig. 8 generates an outgoing email document as an "outgoing message document" consisting of a translation and original text to be sent to a destination email address, which is a "destination electronic address," based on email aggregation information 460 of the created email. Then, an outgoing email sending unit 628 serving as an "outgoing message sending unit" sends the generated outgoing email document to the OutBound SMTP server 7, and this outgoing email document is delivered as an outgoing email from the OutBound SMTP server 7 to the destination address via the Internet 12. Furthermore, an outgoing email aggregation information sending unit 630 sends the email aggregation information 460 of the created email to the mail server 5 for storage.

[0075] When there is a request from the client terminal 10 to read the mail aggregation information 460 of user-created mails, the created mail information request sending unit 632 sends a provision request to the mail server 5. In response to this provision request, the created mail aggregation information acquisition unit 634 acquires the mail aggregation information 460 of created mails sent from the mail server 5. The contents of this mail aggregation information 460 are displayed on the mail display screen of the client terminal 10.

[0076] The address registration / display screen providing unit 636 displays on the client terminal 10 a registration screen for address information 660 and a display screen for registered address information 660. In the example of the address registration screen 680 shown in Fig. 10, by entering the name, email address, and language of the recipient in a name field 682, an email address field 684, and a language field 686 and clicking a registration button 688, the recipient address information 672 (see Fig. 9(b)) can be registered in the address book 670 of the address book storage unit 638.

[0077] 8 stores an address book 670 for each user. The address information request receiving unit 640 receives a request to provide address information from the mail gateway server 4. In response to the received address information request, the address information providing unit 642 provides the requesting mail gateway server 4 with user address information 674 specified by the user's email address and destination address information 672 specified by the specified email address. [Outbound SMTP Server] The OutBound SMTP server 7 accepts outgoing mail containing outgoing mail documents sent from the WEB / APP server 6 and delivers the outgoing mail to the destination address via the Internet 12. As shown in the schematic block diagram of the OutBound SMTP server 7 in Figure 11, this server 7 is configured to include an OutBound SMTP server control unit 700, a mail sending unit 702, and a mail acceptance unit 704. The OutBound SMTP server control unit 700 is made up of a CPU, ROM, RAM, etc. (not shown), and controls the operation of accepting outgoing mail from the WEB / APP server 6 in the mail acceptance unit 704 and delivering the e-mail from the mail sending unit 702 to the destination address. [Translation analysis server] Translation and analysis server 9 receives via Internet 12 the original email text, its original language, translation language, analysis content, and the email address of the creator as "creator information" that identifies the creator of the original email, and translates and analyzes the body of the received original email text. In this email translation and analysis system 1, the original email text and other information is input from mail gateway server 4 for received email, and from WEB / APP server 6 for email created by the user. Translation and analysis server 9 then sends the translation, analysis results, and other information to the mail gateway server 4 or WEB / APP server 6 that are the input sources.

[0078] Input and output between the mail gateway server 4 and the WEB / APP server 6, which are the input sources, and the translation analysis server 9 is performed via an API (Application Programming Interface) provided by the translation analysis server 9.

[0079] As shown in the schematic block diagram of the translation analysis server 9 in FIG. 12, this server 9 is configured to include a translation analysis server control unit 900, an original email acquisition unit 902, a language acquisition unit 904, an author information acquisition unit 905, an analysis content acquisition unit 906, an email translation sending unit 908, an audio information sending unit 910, an analysis result sending unit 912, a new word / popular phrase acquisition unit 914, a translation unit 916, an audio information creation unit 918, a sentence analysis unit 920, and a learning unit 930.

[0080] The translation analysis server control unit 900 is composed of a CPU, ROM, and RAM (not shown), and controls the operation of accepting an original email, its language, translation language, analysis content, and creator email address, and translating and analyzing the original text, which is the body of the original email. It also controls the operation of generating, by machine learning, trained models used in the translation unit 916 and each analysis unit constituting the text analysis unit 920.

[0081] The original email acquisition unit 902 accepts the original email to be translated and analyzed. The language acquisition unit 904 acquires the original language of the original email body and the translation language. The creator information acquisition unit 905 acquires the creator email address used to identify the creator of the original email. Whether the original email is a received email or a user-created email, the sender email address of the email is entered as the creator email address. By entering the sender email address, the address of the sender who created the email is entered for received email, and the address of the user who created the email is entered for user-created email. In addition, the analysis content acquisition unit 906 acquires the analysis content of the original email. Analysis is performed on the original email based on the acquired analysis content.

[0082] The translated email sending unit 908, audio information sending unit 910, and analysis result sending unit 912 respectively send the translation of the original email text, the audio information of that translation, and the analysis results to the source of the original email. The source server receives this output information and uses it to create email aggregation information 460.

[0083] The translation unit 916 translates the original text, which is the body of the original email written in the original language acquired by the original email acquisition unit 902, into a translation language. The translation unit 916 performs the translation using a dictionary registered in advance and a trained translation estimation model.

[0084] The trained translation estimation model used for translation is generated in advance by machine learning in the training unit 930. FIG. 13(a) is a flowchart showing an outline of the process for generating a trained translation estimation model. Using emails stored in the mail server 5 or sentences obtained via the Internet 12, a large number of combinations of training original sentences to be translated and their translations are created and used as training data (step S800). Deep learning, which is machine learning, is performed using this training data (step S801), and a trained translation estimation model is constructed (step S802).

[0085] FIG. 13(b) is a flowchart showing a general flow of translating an original email text into a translation text in the translation unit 916. First, the original email, the original language of the original email text, and the translation language are input to the translation unit 916 (step S810). The input original email text is translated into the translation language through two routes: translation using a trained translation estimation model and translation using a dictionary. The original email text to be translated is processed assuming that it is written in the input original language. In the route using the translation estimation model, the text of the original email text is divided into predetermined units (step S811), which are input to the translation estimation model for translation (step S812). On the other hand, in the route using a dictionary, the original email text is divided into morphemes, which are the smallest linguistic units of a sentence, and morphological analysis is performed to determine the parts of speech of each morphemes (step S820), and each element is translated using a dictionary (step S821). Next, the translation results obtained from these two routes are compared and adjusted to make the translated sentence the most suitable sentence (step S830).The translated sentence thus obtained is finally output from the translation unit 916 (step S831).

[0086] Note that translation may be performed using only a dictionary without using a trained translation estimation model.

[0087] The voice information creating unit 918 shown in FIG. 12 converts the translation of the original mail text created by the translation unit 916 into voice information that can be audibly reproduced, and creates a voice information file.

[0088] The text analysis unit 920 is configured to include a sentiment analysis unit 922, a misspelling and omission detection unit 924, a politeness level analysis unit 926, a harassment level analysis unit 928, and an impersonation analysis unit 929. Each of these analysis units analyzes the original email body using a trained generative model that has been generated in advance through machine learning in the learning unit 930. The language to be analyzed is the original language, and therefore the analysis is performed assuming that the original email body is written in the source language acquired by the language acquisition unit 904.

[0089] The original email may be analyzed by all the analysis units included in the text analysis unit 920, but in this embodiment, the analysis unit corresponding to the analysis content acquired by the analysis content acquisition unit 906 is selected and the analysis is performed.

[0090] The emotion analysis unit 922 analyzes the original email body acquired by the original email acquisition unit 902, estimates the emotion of the creator, and outputs an emotion level as the analysis result. The emotion level is expressed as "happy," "troubled," "surprised," "neutral," "angry," etc., as shown in Fig. 22.

[0091] FIG. 14(a) is a flowchart showing an outline of the process for generating a trained emotion estimation model used by the emotion analysis unit 922. Using emails stored in the mail server 5 and sentences obtained via the Internet 12, a large number of combinations of training sentences and emotion levels indicating the corresponding emotional states are generated and used as training data (step S840). For example, training sentences containing Japanese words such as "delighted," "fun," and "happiness," or English words such as "happy," "glad," and "pleased," are paired with the emotion level "delighted." Deep learning is performed using the combinations of training sentences and emotion levels created in this way (step S841), and a trained emotion estimation model is constructed (step S842).

[0092] 14(b) is a flowchart showing the general flow of estimating an emotional level from an original text in the emotion analysis unit 922. When an original email and its original language are input to the emotion analysis unit 922 (step S850), the text is divided into predetermined units (step S851), which are then input to a trained emotion estimation model for emotion analysis (step S852). The emotional level estimation result obtained in this way is output from the text analysis unit 920 (step S853).

[0093] It should be noted that emotion estimation may be performed using a dictionary of combinations of words and emotion levels that is stored in advance in the emotion analysis unit 922, rather than using a trained emotion estimation model.

[0094] The typographical error and omission detection unit 924 shown in FIG. 12 analyzes the original email, detects typos and omissions therein, and outputs typographical error and omission information indicating the content of the typos and the positions of the omissions.

[0095] FIG. 15(a) is a flowchart showing an outline of the process for generating a trained typographical error detection model. Using emails stored in the mail server 5, a large number of combinations of training sentences and their corresponding typos and omissions are created and used as training data (step S860). Deep learning is then performed (step S861), and a trained typographical error detection model is constructed (step S862). FIG. 15(b) is a flowchart showing an outline of the process for detecting typos and omissions in an original text by the typographical error detection unit 924. When an original email and an original language are input to the typographical error detection unit 924 (step S870), the original email is divided into predetermined units (step S871) and analyzed using the trained typographical error detection model (step S872). Furthermore, the original email is morphologically analyzed assuming that it is written in the input source language (step S880), and each element is analyzed for typos and omissions using a dictionary of typos and omissions (step S881). Finally, the results obtained using the misspelling and omission detection model are compared with the results obtained using the dictionary, and adjustments are made to correct the misspelling and omission information to obtain more accurate information (step S890).The misspelling and omission information thus obtained is output from the misspelling and omission detection unit 924 (step S891).

[0096] The detection of misspellings and omissions may be performed using only a dictionary of misspellings and omissions, without using a trained misspelling and omission detection model.

[0097] The politeness level analysis unit 926 shown in Fig. 12 analyzes the original email, estimates its politeness level, and outputs the politeness level as the analysis result. The politeness level is expressed as "polite," "normal," "slightly rough," "rough," etc., as shown in Fig. 22.

[0098] FIG. 16(a) is a flowchart showing an outline of the process for generating the trained politeness level estimation model used by the politeness level analysis unit 926. Using emails and other data stored in the mail server 5, a large number of combinations of training sentences and their corresponding politeness levels are generated and used as training data (step S900). For example, training sentences that contain honorifics or humble language in Japanese or that end with "arimasu" (thank you) or "moshiagemasu" (thank you) are paired with the "polite" politeness level. Deep learning is performed using the combinations of training sentences and politeness levels created in this way (step S901), and a trained politeness level estimation model is constructed (step S902). FIG. 16(b) is a flowchart showing an outline of the process for estimating the politeness level from an original text in the politeness level analysis unit 926. When an original text such as an email is input to the politeness level analysis unit 926 (step S910), the text is divided into predetermined units (step S911), and analyzed using the trained politeness level estimation model (step S912). The politeness level estimation result thus obtained is output from the politeness level analysis unit 926 (step S913).

[0099] The politeness level may be estimated using a dictionary of combinations of words and politeness levels that is stored in advance, rather than using a trained politeness level estimation model.

[0100] The harassment level analysis unit 928 shown in Fig. 12 analyzes the original email, estimates the degree of harassment, and outputs the harassment level as the analysis result. The harassment level is expressed as "low," "normal," "high," etc., as shown in Fig. 22.

[0101] FIG. 17(a) is a flowchart showing an outline of the process for generating a trained harassment level estimation model used by the harassment level analysis unit 928. Using emails and other data stored in the mail server 5, a large number of combinations of training sentences and corresponding harassment levels indicating the degree of harassment are generated and used as training data (step S920). For example, training sentences containing highly harassing phrases are paired with the harassment level "high." Deep learning is performed using the combinations of training sentences and harassment levels created in this way (step S921), and a trained harassment level estimation model is constructed (step S922). FIG. 17(b) is a flowchart showing an outline of the process for estimating a harassment level from an original text in the harassment level analysis unit 928. When an original email or other text is input to the harassment level analysis unit 928 (step S930), the text is divided into predetermined units (step S931), and analyzed using the harassment level estimation model (step S932). The estimated result of the harassment level thus obtained is output from the harassment level analysis unit 928 (step S933).

[0102] The spoofing analysis unit 929 shown in Fig. 12 analyzes the original email to determine whether the creator of the original email body acquired by the original email acquisition unit 902 is the original creator corresponding to the creator email address acquired by the creator information acquisition unit 905, and calculates the probability that the creator is not the original creator.The calculated probability that the creator of the original email is not the original creator, i.e., the probability of spoofing, is output.This spoofing probability will be a small value close to 0% if the creator of the original email is estimated to be the original creator.On the other hand, if the creator is estimated to be not the original creator, it will be a large value, and the more likely it is that the original creator is spoofed, the closer it gets to 100%.

[0103] The impersonation analysis unit 929 stores information about the original text creator associated with the email address (not shown), and when an email address is input, the corresponding information about the original text creator is read out, allowing the original text creator to be identified.

[0104] 18(a) is a flowchart showing an outline of the process for generating a trained spoofing probability estimation model used by the spoofing analysis unit 929. Emails created by multiple creators stored in the mail server 5 are extracted and used as training data (step S940). In this case, a trained spoofing probability estimation model targeting these multiple creators is generated. The extracted emails may be all emails from the multiple creators, or just a portion of them.

[0105] Next, as preprocessing, signatures and quotations from other emails are removed from the text of all emails used as training data, extracting only the portions written by the email author (step S941). Next, feature values ​​are detected for the preprocessed email text (step S942). Feature values ​​include the katakana ratio, romaji ratio, kanji ratio, hiragana ratio, noun ratio, adjective ratio, verb ratio, symbol ratio, adverb ratio, conjunction ratio, prefix ratio, and sentence length ratio. Other feature values ​​include TF-IDF (Term Frequency-Inverse Document Frequency). TF-IDF is a method for evaluating the importance of words contained in a document and is calculated as the product of TF (Term Frequency, word frequency) and IDF (Inverse Document Frequency). TF is the frequency of a word in an email text and is calculated by dividing the number of a word contained in the email text by the total number of words contained in the email text. IDF is a value that indicates how rare a word is among multiple target emails, and is calculated as the logarithm of the value obtained by dividing the total number of emails for each author that serve as training data by the number of emails that contain the word. IDF will be a small value if the word appears in many emails, and a large value if the word appears in only a few emails. Note that it is not necessary to use all of the feature values ​​shown above; some of them may be selected and used.

[0106] Deep learning is performed using the feature values ​​calculated in this manner (step S943), and a trained spoofing probability estimation model is constructed (step S944).

[0107] 18(b) is a flowchart showing a general flow of the spoofing analysis unit 929 estimating the spoofing probability from the original text. The original email, its original language, and the email address of the creator to be analyzed for spoofing are input to the spoofing analysis unit 929 (step S950). The spoofing analysis unit 929 stores the original text creator corresponding to this creator email address, and identifies the original text creator based on the input email address (step S951). Next, the text of the original email is divided into predetermined units (step S952), which are input into a trained spoofing probability estimation model for spoofing analysis (step S953). The spoofing analysis unit 929 outputs the spoofing probability estimation result obtained in this way (step S954).

[0108] The new word / buzzword acquisition unit 914 shown in FIG. 12 accepts new words, buzzwords, etc., and reflects them in the translation performed by the translation unit 916, and in the analysis of the original email performed by the sentiment analysis unit 922, the misspelling detection unit 924, the politeness level analysis unit 926, and the harassment level analysis unit 928. [Email display screen] When a user sends a request to read received or sent emails from the client terminal 10 to the WEB / APP server 6, the WEB / APP server 6 reads the email aggregation information 460 of the specified email from the mail server 5 and displays the contents of the email on the client terminal 10.

[0109] FIG. 19 is an example of an email display screen 30 for a received email displayed on the client terminal 10. The original text of this email is in English, and the translation is in Japanese. The From field 32 displays the sender's email address of the received email, and the To field 34 displays the recipient's email address. The Subject field 36 displays the subject. In this figure, "Meeting request" is displayed in the Subject field 36. Of these, "Meeting request" is the translation of the subject, and "Meeting request" represents the original text. Note that if the translation setting 441 at the time of reception (see FIG. 4) is set to not leave the original text, the original subject in the Subject field 36 is not displayed, and in the example of FIG. 19, only the translation of the subject, "Meeting request," is displayed.

[0110] The upper right corner of the email display screen 30 includes an emotion level field 38, which displays the analysis results of the emotion level of the original email. While "normal" is displayed in this figure, this emotion level field 38 can be displayed using status screen displays such as "happy" or "troubled," along with the images shown in FIG. 22. Below the emotion level field 38 is a spoofing probability field 39, which displays the probability of spoofing, i.e., the sender email address displayed in the "From" field 32, not the sender. The spoofing probability is displayed as a value ranging from 0% to 100%, with the closer to 0%, the higher the likelihood that the sender wrote the email themselves, and the closer to 100%, the higher the likelihood that someone other than the sender wrote the email while impersonating them. Furthermore, if an analysis is performed on analysis content other than emotion level or spoofing probability, the results are displayed in the upper right corner, similar to the emotion level field 38 and spoofing probability field 39.

[0111] Following the Subject field 36 is a translation display field 40, which displays the translation of the original email. Below that is an original text display field 42, which displays the original email. At the boundary between the translation display field 40 and the original text display field 42, there is a display "--Original Text--", making it easy to distinguish between the translation and the original text. However, if the translation setting 441 at the time of reception is set to not leave the original text, the original text display field 42 will not be displayed.

[0112] Although not shown in FIG. 19, an audio information file of the translated text may be attached to this mail display screen 30, or a one-time URL for playing back this audio information file may be pasted.

[0113] Although FIG. 19 shows an example of the email display screen 30 for received email, the email display screen for outgoing emails created and sent by the user will also be displayed in a similar manner.

[0114] In addition, the outgoing email document generated in the outgoing email document generation unit 626 of the WEB / APP server 6 shown in Figure 8 may be configured so that the screen displayed to the recipient is almost identical to the screen shown in Figure 19 with the analysis result display in the upper right corner removed. [Email creation screen] 20 shows an example of an email creation screen 50 displayed on the client terminal 10. The From field 52 displays the user's email address as the sender's email address, and the To field 54 displays the recipient's email address. The Subject field 56 displays the subject. The Subject field 56 is followed by a document creation field 60 in which an email document created by the user is entered, and in this example, the email document is written in Japanese. Below the document creation field 60, a sender's email document field 62 is displayed, displaying the email document from the sender to which this email is to be replied.

[0115] When the discard button 66 located at the bottom of the email creation screen 50 is clicked, the created email document is deleted.

[0116] Clicking the preview button 64 translates and analyzes the created email document, and a translation preview screen 70, as shown in FIG. 21, is displayed. The analysis results of the created email are displayed in the upper right corner of the preview screen 70, and the results are displayed in the emotion level field 72, harassment level field 74, and politeness level field 76 using the status screen display shown in FIG. 22. The spoofing probability field 76 also displays an estimated spoofing probability that the text entered in the document creation field 60 was written by someone other than the user whose email address is in the From field 52. Following the Subject field is a translation display field 78, which displays the translation of the text of the created email. In this example, the text is translated from Japanese to English.

[0117] The user can manually correct the contents of this translation display field 78.

[0118] When the user wishes to send the contents of this preview screen 70 to a destination email address, he or she clicks the send button 82 located at the bottom of this screen 70. This operation sends a send request to the WEB / APP server 6, which then generates a sending email document based on the contents of this preview screen 70 and sends it to the destination email address.

[0119] When the user clicks on the back button 84 located at the bottom of this translation preview screen 70, the user returns to the email creation screen 50 shown in FIG. 20, where the user can correct the original email document. [Operation of the email translation analysis system] Next, the operation of the email translation and analysis system 1 according to the first embodiment will be described. [Operations from receiving email to storing it on the mail server] First, the operation from when the InBound SMTP server 3 receives e-mails to when the mail aggregation information 460 of the received e-mails is constructed and stored in the mail server 5 will be described.

[0120] FIG. 23 is a schematic flowchart showing the process from receiving an incoming email to storing this email in the email server 5.

[0121] When the InBound SMTP server 3 receives an e-mail delivered from the external mail server 11, the received e-mail is transferred to the mail gateway server 4 (step S100).

[0122] In the mail gateway server 4, the received mail is accepted by the received mail acceptance unit 402 (step S110). The operation of this server 4 is controlled by the mail gateway server control unit 400, and in order to obtain the original language and translation language of the received mail, a request to provide address information 660 is sent to the WEB / APP server 6. Since the address information 660 requires both user address information 674 corresponding to the destination of the received mail and destination address information 672 corresponding to the sender, the destination email address and sender email address of the received mail are sent to the WEB / APP server 6 along with the request to provide the information.

[0123] When the WEB / APP server 6 receives the address information provision request, it reads out the user address information 674 and the destination address information 672 from the address book storage unit 638 and provides them to the mail gateway server 4 (step S140).

[0124] When the mail gateway server 4 acquires the address information 660, the language identification unit 406 reads the address corresponding language 666 from the user address information 674 and identifies the translation language. The language identification unit 406 also reads the address corresponding language 666 from the destination address information 672 and identifies the original language of the received email (step S111).

[0125] Thereafter, the mail gateway server 4 reads the user's personal setting information 438 stored in the personal setting information storage unit 426 in order to obtain the analysis content of the received mail. From the reception setting 440 (see FIG. 4) of this personal setting information 438, it is detected whether or not analyses such as an emotion level check 442 and a spoofing probability check 443 have been performed (step S112). Furthermore, from the translation setting 441 of the personal setting information 438, it is also possible to read whether or not translation will be performed.

[0126] The mail gateway server 4 inputs the received original email, the original language of the received email, the translation language, the sender email address of the received email, the selected analysis content, etc. via an API provided by the translation analysis server 9. The translation analysis server 9 performs translation and analysis based on the input received email, translation language, etc. (step S120). It also generates an audio information file of the translated text.

[0127] The translation and analysis server 9 outputs the translations and analysis results to the mail gateway server 4, which then creates mail aggregation information 460 of received mail (step S113). This mail aggregation information 460 is sent to the mail server 5 and stored in the received mail aggregation information storage unit 504 (step S130). [Behavior when displaying email aggregation information stored on the email server] Next, a description will be given of the operation when the mail aggregation information 460 of received mails or mails created and sent by the user and stored in the mail server 5 is displayed on the client terminal 10. Fig. 24 is a schematic flowchart showing this operation.

[0128] When a user operates a browser to view an email, a request for providing a mail display screen is sent from the client terminal 10 to the WEB / APP server 6 (step S200). In response to this request, the WEB / APP server 6 provides a mail display screen from the mail display / creation screen providing unit 602 (step S210) and displays it on the client terminal 10 (step S201).

[0129] The user operates this email display screen to specify the email to be displayed (step S202). The specified email may be a received email or an email created and sent by the user, as long as it is stored in the email server 5. In response to the user's specification, a read request for email aggregation information 460 is sent from the client terminal 10 to the WEB / APP server 6.

[0130] When the WEB / APP server 6 receives this read request in the email display / creation screen providing unit 602 (step S211), it sends a request to provide email aggregation information 460 to the email server 5. In response to this request, the email server 5 reads out the specified email aggregation information 460 from the stored email aggregation information 460 (step S220) and provides it to the WEB / APP server 6.

[0131] At this time, if the email specified by the user is an incoming mail, a provision request is sent from the incoming mail information request sending unit 604 of the WEB / APP server 6, and this provision request is accepted by the incoming mail information request accepting unit 510 of the mail server 5. The mail server 5 reads out the mail aggregation information 460 of the specified incoming mail from the received mail aggregation information storage unit 504. The read out mail aggregation information 460 is provided to the WEB / APP server 6 from the received mail aggregation information providing unit 512, and is acquired by the received mail aggregation information acquiring unit 606 of the WEB / APP server 6.

[0132] On the other hand, if the user designation is for an email created and sent by the user, a provision request is sent from the created email information request sending unit 632 of the WEB / APP server 6, and this provision request is accepted by the created email information request accepting unit 514 of the mail server 5. The mail server 5 reads out the email aggregation information 460 of the specified created email from the created email aggregation information storage unit 508. This email aggregation information 460 is provided from the created email aggregation information providing unit 516 to the WEB / APP server 6, and is acquired by the created email aggregation information acquiring unit 634 of the WEB / APP server 6.

[0133] In the WEB / APP server 6, the content of the acquired email aggregation information 460 is provided to the client terminal 10 from the email display / creation screen providing unit 602 (step S212), and is displayed on the email display screen 30 of the terminal 10 (step S203). This screen 30 allows the user to view the specified email. [Behavior until user creates and sends mail document] Next, the operation from when a user creates a mail document to when the mail document is sent to a destination mail address will be described. Figure 25 is a schematic flowchart showing this operation.

[0134] When a user operates a browser to create a mail document, a request to provide a mail creation screen is sent from the client terminal 10 to the WEB / APP server 6 (step S300). In response to this request, the WEB / APP server 6 provides a mail creation screen from the mail display / creation screen providing unit 602 (step S310) and displays it on the client terminal 10 (step S301).

[0135] The user creates an email document on this email creation screen 50 (step S302). To check the translation and analysis results of the email the user has created, the user clicks on the preview button 64 located on the email creation screen 50. A request for a translation preview screen is sent from the client terminal 10 to the WEB / APP server 6.

[0136] In the WEB / APP server 6, the mail display / creation screen providing unit 602 receives the request to provide this preview screen (step S311), and also receives the user's created mail on the mail creation screen (step S312).

[0137] To obtain the original language and translation language of the user-created email, the WEB / APP server 6 reads user address information 674 and destination address information 672 corresponding to the destination address of the user-created email from the address book storage unit 638. The language identification unit 610 reads the address correspondence language 666 from the user address information 674 and identifies the original language of the user-created email. The language identification unit 610 also reads the address correspondence language 666 from the destination address information 672 and identifies the translation language (step S313).

[0138] Thereafter, the WEB / APP server 6 sends a request for providing personal setting information 438 to the mail gateway server 4 in order to obtain the analysis content of the user-created mail.

[0139] When the mail gateway server 4 receives the request to provide personal setting information, it reads out the user's personal setting information 438 from the personal setting information storage unit 426 and provides it to the WEB / APP server 6 (step S330).

[0140] The WEB / APP server 6 detects whether or not analyses such as an emotion level check 447, a typographical error check 448, a politeness level check 449, a harassment level check 450, and an impersonation probability check 451 have been performed from the sending settings 445 (see Figure 4) of this personal setting information 438 (step S314).

[0141] The WEB / APP server 6 inputs the original text of the user-created email, its original language, translation language, the email address of the user who created it, selected analysis content, etc. via the API provided by the translation analysis server 9. The translation analysis server 9 performs translation and analysis based on the input text of the created email, translation language, etc. (Step S340). It also generates an audio information file of the translated text.

[0142] The translation and analysis results are output from the translation analysis server 9 to the WEB / APP server 6, which then creates email aggregation information 460 of user-created emails (step S315).Then, the email display / creation screen providing unit 602 provides the contents of this email aggregation information 460 to the client terminal 10 as a translation preview screen 70 (step S316).

[0143] The user can check the translation and analysis results on the translation preview screen 70 displayed on the client terminal 10 (step S303). When the user clicks the send button 82 located on this preview screen 70, a send request is sent from the client terminal 10 to the WEB / APP server 6.

[0144] In the WEB / APP server 6, when the mail display / creation screen providing unit 602 receives this transmission request, the outgoing mail document generating unit 626 generates an outgoing mail document based on the mail aggregation information 460 of the user-created mail (step S317). The outgoing mail containing the generated outgoing mail document is sent from the outgoing mail sending unit 628 to the OutBound SMTP server 7 (step S318), and is then sent from the OutBound SMTP server 7 to the destination mail address via the Internet 12 (step S360).

[0145] Finally, the mail aggregation information 460 of the user-created mails is sent from the WEB / APP server 6 to the mail server 5, and is stored in the created mail aggregation information storage unit 508 of the mail server 5 (step S350). [Advantages of the First Embodiment] According to the first embodiment, the original language and translation language of the original email are identified using the user address information 674 and destination address information 672 stored in the address book storage unit 638 of the WEB / APP server 6, and the original email is then translated. This eliminates the need for the user to manually identify the language, and the original email is automatically translated. In addition, the original email can be analyzed by the sentence analysis unit 920 of the translation analysis server 9 to estimate the emotional state, etc.

[0146] Furthermore, according to the first embodiment, audio information that can be audibly reproduced is generated for the translation of the original email, so that the content of the translation can be understood aurally.

[0147] Furthermore, according to the first embodiment, the translation analysis server 9 can analyze the original email to detect typos and omissions, estimate the degree of politeness, the degree of harassment, the probability of impersonation, and so on.

[0148] Furthermore, according to the first embodiment, the translation and analysis server 9 uses a trained generative model, so that the original email can be translated and analyzed with higher accuracy.

[0149] Furthermore, according to the first embodiment, the mail aggregation information 460, which aggregates the translations and analysis results of the original emails, is stored in the mail server 5, so that the translations and analysis results can be displayed and viewed on the client terminal 10 at any time.

[0150] Furthermore, according to the first embodiment, the contents of the mail aggregation information 460, which is constructed by aggregating translations and analysis results of user-created mail, can be displayed and confirmed on the client terminal 10 as a translation preview screen 70. Furthermore, when a transmission request is received from the client terminal 10, a transmission mail document can be generated based on this mail aggregation information 460 and delivered as a transmission mail to the destination mail address. [Embodiment 2 of the Invention] Next, a second embodiment of the present invention will be described with reference to Figures 26 to 36. However, elements that are the same as or correspond to those in the first embodiment described above will be given the same reference numerals, and duplicated descriptions will be omitted.

[0151] 26 is a block diagram showing the configuration of a chat system including a chat translation analysis system. Chat translation analysis system 1A, which serves as an "electronic message translation analysis system" according to the present invention, includes a chat gateway server 4A, a WEB / APP server 6A, and a translation analysis server 9A. The chat gateway server 4A and the WEB / APP server 6A are connected via a LAN 8, and the chat gateway server 4A and the WEB / APP server 6A are connected to the translation analysis server 9A via the Internet 12.

[0152] Furthermore, this chat system 100A is configured such that, in addition to the configuration of the chat translation and analysis system 1A, client terminals 10, which allow each user to view, create, and distribute electronic chat messages, are connected via the Internet 12 as a "communication line." Chat users form groups, and conversations are carried out via electronic messages by sharing the electronic messages sent by each user in the group with everyone in the group.

[0153] Users of this chat system 100A access chat software running on the WEB / APP server 6A through a browser installed on the client terminal 10, and view past chats, create chat message documents, and send and receive message documents.

[0154] By using this chat translation and analysis system 1A, it is possible to automatically translate received electronic messages written in a foreign language into the user's language, and to translate message documents created by the user into the language of the communication partner.It is also possible to analyze the emotions, detect typos, the level of politeness, the level of harassment, and the probability of impersonation for the electronic messages created by the user.

[0155] The operation of this chat translation and analysis system 1A can be summarized as follows: When a user in a group creates an electronic message on a client terminal 10, the electronic message is sent to the WEB / APP server 6A. The electronic message sent to the WEB / APP server 6A is then sent to the chat gateway server 4A, where it is translated and analyzed by the translation and analysis server 9A. The chat gateway server 4A aggregates the translations and analysis results and sends them to the WEB / APP server 6A, which then distributes the original text of the electronic message, the translation, the analysis results, etc. to all users in the chat group. [Chat gateway server] The chat gateway server 4A accepts user-created messages sent from the WEB / APP server 6A, inputs the text of these electronic messages to the translation and analysis server 9A for translation and analysis, and creates message aggregation information 460A that aggregates the original text, translation, analysis results, etc. that make up the text of these electronic messages. Message aggregation information 460A created in this way is sent to the WEB / APP server 6A. As shown in the schematic block diagram of chat gateway server 4A in Figure 27, this server 4A is configured to include a chat gateway server control unit 400A, a message receiving unit 402A, an address information acquisition unit 404A, a language identification unit 406A, a message sending unit 408A, a language notification unit 410, a creator information notification unit 411, an analysis content selection unit 412, a message translation receiving unit 414A, a voice information receiving unit 416, an analysis result receiving unit 418, a message aggregation information composition unit 420A, a message aggregation information sending unit 422A, a personal setting information registration / display screen providing unit 424, and a personal setting information storage unit 426.

[0156] The chat gateway server control unit 400A is composed of a CPU, ROM, and RAM (not shown), and receives electronic messages from the WEB / APP server 6A, identifies the original language and translation language of the electronic message body, performs translation and analysis, and controls the operation of composing message aggregation information 460A based on the results. For this reason, this chat gateway server control unit 400A operates as part of the "translation and analysis control unit."

[0157] In this chat translation and analysis system 1A, a chat gateway server control unit 400A of the chat gateway server 4A and a WEB / APP server control unit 600A of the WEB / APP server 6A (described later) cooperate to operate as a "translation and analysis control unit."

[0158] The message receiving unit 402A receives electronic messages sent from the WEB / APP server 6A.

[0159] The address information acquisition unit 404A acquires address information 660A of all members of the chat group sent from the WEB / APP server 6A, and a creator chat address 662A that can identify the user who created the electronic message.

[0160] The language identification unit 406A identifies the original language of the electronic message body from the creator chat address 662A acquired by the address information acquisition unit 404A, and identifies the translation language from the group's address information 660A. If the users who make up the group speak multiple languages, the translation will also be in multiple languages. This language identification unit 406A operates as an "original language identification unit" that identifies the original language of the body of the original message, and as a "translation language identification unit" that identifies the translation language.

[0161] The message sending unit 408A inputs the electronic message text to the translation and analysis server 9A via the Internet 12 for translation and analysis. The language notification unit 410 also inputs the original language and translation language of the electronic message identified by the language identification unit 406A to the translation and analysis server 9A. If there are multiple translation languages, the multiple translation languages ​​are input.

[0162] Creator information notification unit 411 inputs creator chat address 662A to translation analysis server 9A as "creator information" that identifies the creator of the electronic message body. This creator chat address 662A is sent from WEB / APP server 6A and acquired by address information acquisition unit 404A. Since the translation analysis server 9A stores the original creator in association with the chat address, it is possible to identify the original creator from the input creator chat address.

[0163] The analysis content selection unit 412 reads and selects analysis content for the electronic message in the translation analysis server 9A from personal setting information 438A registered in advance by the user, and causes the translation analysis server 9A to analyze the content.

[0164] 28 shows an example of the configuration of personal setting information 438A, which has almost the same configuration as the personal setting information 438 shown in FIG. 4 above. The difference is that the contents of translation setting 441A included in receiving settings 440A are "display translation and original text," "display translation only," and "display original text only." The chat translation analysis system 1A of the second embodiment is configured so that a translation is created before the electronic message created by the user is distributed, and therefore, when receiving a message, the user is allowed to select whether or not to display the created translation.

[0165] Personal setting information 438A registered by the user is stored in personal setting information storage unit 426 shown in Fig. 27. Personal setting information registration / display screen providing unit 424 provides a screen for displaying and registering personal setting information 438A to client terminal 10 via Internet 12. Fig. 29 shows an example of personal setting information registration / display screen 456A. ​​This screen 456A has a configuration similar to personal setting information registration / display screen 456 shown in Fig. 5 above, with only the display of translation setting selection items included in reception settings being different.

[0166] Incidentally, the settings may be made for an individual user as in the personal setting information 438A, but they may also be made on a group basis.

[0167] A message translation receiving unit 414A shown in FIG. 27 receives a translation of the original text, which is the main text of the electronic message, from the translation and analysis server 9A.

[0168] Message aggregation information constructing unit 420A constructs message aggregation information 460A by aggregating the original text, translation, and analysis results of electronic messages. This message aggregation information constructing unit 420A operates as an "aggregated information constructing unit."

[0169] The configuration example of message aggregation information 460A shown in Fig. 30 is almost the same as the email aggregation information 460 shown in Fig. 6. The difference is that the chat address of the creator of the electronic message (sender chat address) is stored in sender chat address field 461A, and the chat addresses of all members of the chat group are stored in group chat address field 462A. Electronic messages and their translations are delivered to all users in this group.

[0170] The message aggregation information sending unit 422A shown in FIG. 27 sends message aggregation information 460A of electronic messages composed by the message aggregation information composition unit 420A to the WEB / APP server 6A. [WEB / APP server] When the WEB / APP server 6A receives a read request from a client terminal 10 to allow chat users to view past electronic messages, it provides the requesting terminal 10 with a chat display screen and a chat creation screen so that the user can create a message document. The server 6A sends the user's electronic message to the chat gateway server 4A, which translates and analyzes the electronic message. The server 6A then obtains message aggregation information 460A, which is based on the translation and analysis results, from the chat gateway server 4A. The server 6A then displays the message aggregation information 460A on the client terminal 10 of the user who created the electronic message, allowing the user to view the translation and analysis results. Furthermore, when a send request is received from the user's client terminal 10, the server 6A generates a message document to be displayed on the chat display screen based on the message aggregation information 460A and distributes it to the chat addresses of all members of the chat group. This allows electronic messages and their translations created by one user to be displayed on the client terminals 10 of all users in the group and shared by all users.

[0171] As shown in the schematic block diagram of the WEB / APP server 6A in Figure 31, this server 6A is configured to include a WEB / APP server control unit 600A, a chat display / creation screen providing unit 602A, a personal setting information acquisition unit 608, a created message sending unit 612A, a created message aggregation information acquisition unit 606A, a sent message document generation unit 626A, a message aggregation information storage unit 504A, an address registration / display screen providing unit 636, an address book storage unit 638A, and an address information providing unit 642.

[0172] The WEB / APP server control unit 600A is composed of a CPU, ROM, and RAM (not shown), and controls the operation of providing a chat display screen configured based on the message aggregation information 460A to the client terminal 10, and of providing a chat creation screen. Therefore, the WEB / APP server control unit 600A corresponds to an "information provision control unit" that controls the operation of providing the message aggregation information 460A via the Internet 12. Note that the chat screen is often configured as a chat display / creation screen that combines the chat display screen and chat creation screen into a single screen.

[0173] Control unit 600A also controls the operation of storing message aggregation information 460A, which is configured based on the translations and analysis results of user-created messages sent from chat gateway server 4A, in message aggregation information storage unit 504A included in WEB / APP server 6A. Therefore, WEB / APP server control unit 600A also operates as part of the "translation analysis control unit."

[0174] Furthermore, the control unit 600A also controls the operation of generating a sending message document based on the message aggregation information 460A of the user-created messages and distributing it to the chat addresses of all the members of the chat group, which are the destination electronic addresses.

[0175] The chat display / creation screen providing unit 602A configures a chat display screen to be displayed on the client terminal 10 based on the message aggregation information 460A in order to view electronic messages from past chats, and provides this to the terminal 10 via the Internet 12. Therefore, the chat display / creation screen providing unit 602A operates as an "aggregated information providing unit" that provides the message aggregation information 460A via the Internet 12.

[0176] When viewing electronic messages from past chats, the user specifies the electronic message to be displayed on the chat display screen. This specification causes the client terminal 10 to request the WEB / APP server 6A via the Internet 12 to read the message aggregation information 460A for the specified electronic message. When this read request is received, the message aggregation information 460A for the specified electronic message is read from the message aggregation information storage unit 504A and provided to the requesting client terminal 10. That is, the contents of the electronic message, such as the translation, original text, and analysis results, are displayed on the client terminal 10 based on the read message aggregation information 460A.

[0177] The chat display / creation screen providing unit 602A also provides a chat creation screen so that the user can create a message document using the client terminal 10. The WEB / APP server 6A accepts the message to be created input by the user via this chat creation screen. For this user-created message, which is the "original message" to be translated and analyzed, the chat display / creation screen providing unit 602A corresponds to the "original message accepting unit."

[0178] As described above, the WEB / APP server 6A sends the received user-created message to the chat gateway server 4A, and obtains the message aggregation information 460A configured by the chat gateway server 4A. The contents of this message aggregation information 460A are provided and displayed on the client terminal 10 of the user who created the created message from the chat display / creation screen provider 602A via the Internet 12. When the user checks the contents on the chat creation screen and issues a transmission request, the transmission request is sent to the WEB / APP server 6A via the Internet 12. This server 6A generates a transmission message document based on the message aggregation information 460A and distributes it to the chat addresses of all members of the group via the chat display / creation screen provider 602A.

[0179] The created message sending unit 612A inputs the user-created message to the chat gateway server 4A via the Internet 12 for translation and analysis. In addition, the address information providing unit 642 transmits address information 660A of all members of the chat group stored in the address book storage unit 638A and the chat address 662A of the user who created the electronic message to the chat gateway server 4A.

[0180] Fig. 32(a) shows an example of the configuration of address information 660A, which is configured to include a chat address 662A as an "electronic address" that identifies a chat user, a name 664 corresponding to the chat address 662A, and an address corresponding language 666 that is the language corresponding to the address 662A. As shown in Fig. 32(b), a group address book 670A is configured from multiple address information 660A registered as users of a group. The group address book 670A is configured for each group and is stored in the address book storage unit 638A (see Fig. 31).

[0181] 31 acquires message aggregation information 460A of user-created messages sent from the chat gateway server 4 A. The contents of the acquired message aggregation information 460A are displayed on the chat creation screen of the user's client terminal 10.

[0182] When a user requests to send a created message, a sending message document generation unit 626A serving as a "sending message generation unit" generates a sending message document consisting of translations and original texts to be distributed to the chat addresses of all members of the chat group, which are the "destination electronic addresses," based on this message aggregation information 460A. The sending message document generated by the chat display / creation screen providing unit 602A serving as a "sending message sending unit" is then distributed to the client terminals 10 of the group via the Internet 12.

[0183] The personal setting information acquisition unit 608 sends a request to the chat gateway server 4A to acquire the personal setting information 438A (see FIG. 28) of all the members of the group. From this personal setting information 438A, the receiving setting 440A of each user when delivering a sending message document to the group is read and reflected on the chat display screen displayed to each user.

[0184] Furthermore, message aggregation information 460A of user-created messages sent from the chat gateway server 4A is stored in a message aggregation information storage unit 504A shown in Fig. 31. This message aggregation information storage unit 504A functions as an "aggregated information storage unit."

[0185] The address registration / display screen providing unit 636 displays a registration screen for address information 660A and a display screen for registered address information 660A on the client terminal 10. In the example of address registration screen 680A shown in Fig. 33, there is a group field 681A on the left side that displays group users, and the name, chat address, language, etc. to be registered in the group can be entered in a name field 682, a chat address field 684A, a language field 686, etc. [Translation analysis server] The translation and analysis server 9A receives via the Internet 12 the original text of an electronic message, its original language, translation language, analysis content, and the creator chat address as "creator information" that identifies the creator of the original message, and translates and analyzes the received original message. In this chat translation and analysis system 1A, the original message and other information are input from the chat gateway server 4A. This translation and analysis server 9A sends the translation, analysis results, and other information to the chat gateway server 4A, which is the input source.

[0186] Input and output between the chat gateway server 4A, which is the input source, and the translation analysis server 9A is performed via an API provided by the translation analysis server 9A.

[0187] As shown in the schematic block diagram of the translation analysis server 9A in FIG. 34, this server 9A is configured to include a translation analysis server control unit 900, an original message acquisition unit 902A, a language acquisition unit 904, a creator information acquisition unit 905A, an analysis content acquisition unit 906, a message translation sending unit 908A, an audio information sending unit 910, an analysis result sending unit 912, a new word / buzzword acquisition unit 914, a translation unit 916, an audio information creation unit 918, a sentence analysis unit 920, and a learning unit 930.

[0188] This translation analysis server 9A has almost the same configuration as the translation analysis server 9 shown in FIG.

[0189] The source message acquisition component 902A receives the source electronic message to be translated and analyzed.

[0190] The creator information acquisition unit 905A acquires a creator chat address used to identify the creator of the original message.

[0191] The translated message sending unit 908A sends the translation of the original message body to the chat gateway server 4A that is the input source of the original message. [Chat display / creation screen] A user operates his / her own client terminal 10 to connect to the client terminals 10 of other users in the group via the WEB / APP server 6A and exchange electronic messages.

[0192] 35 is a diagram showing an example of a chat display / creation screen 90 displayed on the client terminal 10. At the left end of this screen 90 is a group field 92 that displays group users, and at the center of the screen is a chat display screen field 94 that displays outgoing message documents of electronic messages delivered from the WEB / APP server 6A. Also, at the bottom center of this screen 90 is a chat creation screen field 96 where the user can create and input a message document.

[0193] When the user inputs a message text into this chat creation screen field 96 and clicks a preview button (not shown), the created message text is translated and analyzed, and the content is displayed on a translation preview screen (not shown).

[0194] After checking the displayed content, the user clicks the send button (not shown) located at the bottom of the preview screen to send an electronic message with that content to everyone in the group. This operation sends a send request to the WEB / APP server 6A, which then generates a message document to send based on the content of the preview screen and distributes it to the chat addresses of the group users. [Chat translation analysis system in action] Next, the operation of the chat translation and analysis system 1A according to the second embodiment will be described.

[0195] Here, we will explain the operation from when a user creates a message document to when that message document is distributed to the chat addresses of all users in the group. Figure 36 is a schematic flowchart showing this operation.

[0196] When a chat is started, the WEB / APP server 6A provides the chat display / creation screen from the chat display / creation screen providing unit 602A to the client terminals 10 of the group, and causes each client terminal 10 to display the chat display / creation screen 90 (steps S501 and S540).

[0197] The user creates a message document on this chat display / creation screen 90 (step S502). In order to check the translation and analysis results of the message created by the user, when the user clicks a preview button (not shown) arranged on the chat display / creation screen 90, a request for providing a translation preview screen is sent from the client terminal 10 to the WEB / APP server 6A.

[0198] In the WEB / APP server 6A, the chat display / creation screen providing unit 602A receives the request to provide this preview screen (step S511), and also receives the message created by the user on the chat display / creation screen 90 (step S512).

[0199] The WEB / APP server 6A transmits the received created message from the created message sending unit 612A to the chat gateway server 4A, and the chat gateway server 4A receives this created message in the message receiving unit 402A (step S520).

[0200] Furthermore, the WEB / APP server 6A transmits address information 660A of all members of the chat group stored in the address book storage unit 638A and the chat address 662A of the user who created the created message from the address information providing unit 642 to the chat gateway server 4A (step S513). The chat gateway server 4A acquires this address information 660A and the like using the address information acquisition unit 404A.

[0201] Next, the language identification unit 406A of the chat gateway server 4A identifies the original language and translation language of the created message based on the acquired chat group address information 660A and creator chat address 662A (step S521).

[0202] Next, in order to acquire the analysis content of the user-created message, the chat gateway server 4A reads the user's personal setting information 438A stored in the personal setting information storage unit 426. From the sending setting 445 (see FIG. 28) of this personal setting information 438A, it is detected whether or not analyses such as an emotion level check 447 and an impersonation probability check 451 have been performed (step S522). Furthermore, from the translation setting 446 of the personal setting information 438A, it is also possible to read whether or not translation will be performed.

[0203] The chat gateway server 4A inputs the user-created message, the original language of the created message, the translation language, the creator's chat address, the selected analysis content, etc. via the API provided by the translation analysis server 9A. The translation analysis server 9A performs translation and analysis based on the input created message, translation language, etc. (step S530). It also generates an audio information file of the translated text.

[0204] The translation and analysis results are output from the translation and analysis server 9A to the chat gateway server 4A, and upon receiving them, the chat gateway server 4A constructs message aggregation information 460A of user-created messages in the message aggregation information construction unit 420A (step S523).

[0205] This message aggregation information 460A is sent from the chat gateway server 4A to the WEB / APP server 6A.

[0206] The WEB / APP server 6A provides the content of the received message aggregation information 460A from the chat display / creation screen providing unit 602A to the creator's client terminal 10 as a translation preview screen (not shown) (step S514).

[0207] The user can check the translation and analysis results on the translation preview screen displayed on the client terminal 10 (step S503). When the user clicks a send button (not shown) located on this preview screen, a send request is sent from the client terminal 10 to the WEB / APP server 6A.

[0208] In the WEB / APP server 6A, when the chat display / creation screen providing unit 602A receives this transmission request, the outgoing message document generating unit 626A generates an outgoing message document based on the message aggregation information 460A of the user-created messages (step S515). The generated outgoing message document is distributed from the chat display / creation screen providing unit 602A to the chat addresses of all members of the chat group via the Internet 12 (step S516).

[0209] The chat display / creation screen delivered from the WEB / APP server 6A is displayed on each client terminal 10 of the group, allowing all members of the group to check the electronic messages sent by one user (steps S504 and S541).

[0210] Finally, message aggregation information 460A of the user-created messages is stored in the message aggregation information storage unit 504A of the WEB / APP server 6A.

[0211] The above operations are repeated while the chat continues. [Effects of the Second Embodiment] According to the second embodiment, it is possible to obtain substantially the same effects as those obtained by the first embodiment.

[0212] According to the second embodiment, the original language of the original message and the translation language are identified using address information 660A in a group address book 670A stored in an address book storage unit 638A of the WEB / APP server 6A, and the original message is then translated. This eliminates the need for the user to manually identify the language, and the original message is automatically translated. Furthermore, the original message can be analyzed by the sentence analysis unit 920 of the translation analysis server 9A, allowing estimation of the emotional state, etc. [Explanation of symbols]

[0213] 100 Email Systems 1. E-mail translation and analysis system (electronic message translation and analysis system) 2 organization 3 Inbound SMTP Server 4 Mail Gateway Server 5. Mail Server 6,6A WEB / APP server 7 Outbound SMTP Server 8 LAN 9,9A Translation analysis server 101, 102, 103, 10 client terminals 11 External mail servers 12 Internet (communication line) 400 Mail gateway server control unit (translation analysis control unit) 402 Incoming mail reception section (original message reception section) 406 Language Identification Unit (Source Language Identification Unit, Translation Language Identification Unit) 412 Analysis Content Selection Section 420 Received mail aggregation information configuration part (aggregation information configuration part) 426 Personal setting information storage unit 438,438A Personal setting information 460 Email aggregation information (message aggregation information) 504 Received email aggregation information storage unit (aggregation information storage unit) 508 Created email aggregation information storage unit (aggregation information storage unit) 600 WEB / APP server control unit (translation analysis control unit, information provision control unit) 602 Email display / creation screen provider (original message receiver, aggregated information provider) 610 Language Identification Unit (Source Language Identification Unit, Translation Language Identification Unit) 616 Analysis Content Selection Section 624 Created email aggregation information configuration part (aggregation information configuration part) 626 Sending mail document generation unit (sending message generation unit) 628 Outgoing mail sending unit (outgoing message sending unit) 638,638A Address book memory section 660,660A Address Information 900 Translation analysis server control unit 916 Translation Department 918 Audio Information Creation Department 920 Sentence Analysis Department 922 Emotion Analysis Department 924 Misspellings and Omissions Detection Unit 926 Politeness Level Analysis Department 928 Harassment Level Analysis Department 929 Impersonation Analysis Department 930 Learning Department 100A Chat System 1A Chat translation analysis system (electronic message translation analysis system) 4A Chat Gateway Server 400A Chat gateway server control unit (translation analysis control unit) 406A Language Identification Unit (Source Language Identification Unit, Translation Language Identification Unit) 420A Message aggregation information configuration part (aggregation information configuration part) 460A Message Aggregation Information 600A WEB / APP server control unit (information provision control unit, translation analysis control unit) 602A Chat display / creation screen providing unit (aggregated information providing unit, original message receiving unit, sending message sending unit) 626A Transmission message document generation unit (transmission message generation unit) 504A Message aggregation information storage unit (aggregation information storage unit)

Claims

1. 1. An electronic message translation and analysis system for translating and analyzing the body of an electronic message, comprising: an original message receiving unit for receiving the electronic message to be translated and analyzed; a translation unit that translates the body of the received electronic message, which is written in a source language, into a target language; an impersonation analysis unit that analyzes the original text using at least the kanji ratio, the hiragana ratio, and TF-IDF to estimate the probability of impersonation of the creator of the original text; and a providing unit that provides the translation of the original text translated by the translation unit and the probability of spoofing estimated by the spoofing analysis unit to the terminal that is the destination of the electronic message or the terminal that created the electronic message.

2. 1. An electronic message translation and analysis system for translating and analyzing the body of an electronic message, comprising: an original message receiving unit for receiving the electronic message to be translated and analyzed; a translation unit that translates the body of the electronic message, which is written in a source language, into a target language; a text analysis unit including an impersonation analysis unit that analyzes the original text using at least the kanji ratio, the hiragana ratio, and TF-IDF to estimate the probability of impersonation of the original text creator; a message creation screen providing unit that provides an electronic message creation screen to a client terminal via a communication line; After a message to be sent is created as the electronic message on the electronic message creation screen displayed on the client terminal, when a pre-sending confirmation request sent from the client terminal is received by the message creation screen providing unit, the original message receiving unit receives the outgoing message created by the client terminal, the translation unit translates the received outgoing message into a translation, and the sentence analysis unit estimates the probability of spoofing based on the received outgoing message, a pre-transmission confirmation screen on which the estimated probability of spoofing is displayed is provided to the client terminal; and a message creation screen providing unit that provides a pre-send confirmation screen before sending the message. The message creation screen providing unit then receives a request to send the message from the client terminal, and sends the translation of the message to a destination electronic address as a message document.

3. the pre-sending confirmation screen provided to the client terminal includes the analysis result of the sentence analysis unit and the translation of the message to be sent translated by the translation unit; When the translation included in the pre-sending confirmation screen displayed on the client terminal is corrected from the client terminal, the translation translated by the translation unit is rewritten with the corrected translation corrected on the pre-sending confirmation screen, 3. The electronic message translation and analysis system according to claim 2, wherein when the request to send the outgoing message from the client terminal is accepted by the message creation screen providing unit, the rewritten revised translation is sent as the outgoing message document to the destination electronic address of the outgoing message document.

4. 3. The electronic message translation and analysis system according to claim 2, wherein the outgoing message document sent to the destination electronic address is configured to display the analysis results of the sentence analysis unit in addition to the translated sentence.

5. 3. The electronic message translation and analysis system according to claim 2, wherein the outgoing message document sent to the destination electronic address does not include the analysis result of the sentence analysis unit and is composed of the translated sentence.

6. 1. An electronic message translation and analysis system for translating and analyzing the body of an electronic message, comprising: an original message receiving unit that receives an original message, which is the electronic message to be translated and analyzed; a translation unit that translates the original text of the original message, which is written in an original language, into a translation language; a text analysis unit including an impersonation analysis unit that analyzes the original text using at least the kanji ratio, the hiragana ratio, and TF-IDF to estimate the probability of impersonation of the original text creator; an aggregated information constructing unit that constructs message aggregated information by aggregating the original text, the translation of the original text translated by the translation unit, and the probability of spoofing estimated by the sentence analysis unit; a translation and analysis control unit that controls an operation of translating and analyzing the original message received by the original message receiving unit to configure the message aggregation information, The translation analysis control unit inputting the original text into the translation unit to generate the translation; inputting the original text into the text analysis unit to estimate the probability of spoofing; The electronic message translation analysis system is characterized in that the original text, the created translation, and the estimated probability of spoofing are input to the aggregate information construction unit to construct the message aggregate information.

7. 7. The electronic message translation and analysis system according to claim 1, wherein the electronic message is an email.

8. 7. The electronic message translation and analysis system according to claim 1, wherein the electronic message is a chat message.

Images