System and Method
The system uses a large-scale language model to determine and modify transmission content in communication tools, addressing inappropriate content review by third parties and enhancing message security and accuracy.
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Current Assignee / Owner
- NTT DOCOMO INC
- Filing Date
- 2024-11-05
- Publication Date
- 2026-05-19
AI Technical Summary
Existing communication tools face issues with inappropriate content being reviewed by human third parties, risking exposure of personal information to unknown recipients.
A system utilizing a large-scale language model to determine and modify transmission content based on prompts, ensuring appropriate content review without human intervention, including a receiving unit, first and second prompt creation units, and a transmitting unit to manage content transmission.
Enables secure and accurate content review in communication tools, preventing third-party access to transmission content and improving message content determination and revision accuracy.
Smart Images

Figure 2026081435000001_ABST
Abstract
Description
Technical Field
[0001] One aspect of the present disclosure relates to systems and methods.
Background Art
[0002] Patent Document 1 discloses a method and a system that enable a message to be evaluated before being transmitted to a recipient.
Prior Art Documents
Patent Documents
[0003]
Patent Document 1
Summary of the Invention
Problems to be Solved by the Invention
[0004] In communication tools (such as various SNSs and matching services), efforts are made to promote soundness by proofreading whether the content transmitted from the sender to the recipient (transmission content) contains inappropriate content or is unpleasant for the recipient. This can keep the communication tool environment healthy.
[0005] Here, in the technologies as described above, there is a problem in that the transmission content is reviewed by a human third party (service provider). Specifically, there is a concern that information such as personality information that can be exchanged between parties may be grasped by a third party and exposed to an unspecified number of people due to leakage.
[0006] One aspect of the present disclosure has been made in view of the above circumstances, and an object thereof is to provide a system and a method for realizing appropriate proofreading of transmission content while avoiding the third party from grasping the transmission content in a communication tool.
Means for Solving the Problems
[0007] A system relating to one aspect of the present invention includes: a receiving unit that receives transmission content, which is the content to be sent from a sender to a receiver; a first prompt creation unit that creates a first prompt for a large-scale language model to determine whether or not to send based on the transmission content and information about the sender and receiver; a second prompt creation unit that creates a second prompt for a large-scale language model to propose modified transmission content, which is the modified transmission content, based on the transmission feasibility determination result output from the large-scale language model in response to the first prompt; an input unit that inputs the first prompt to the large-scale language model and the second prompt to the large-scale language model; and a transmitting unit that sends the transmission feasibility determination result output from the large-scale language model in response to the first prompt and the modified transmission content output from the large-scale language model in response to the second prompt to the sender.
[0008] In such a system, a first prompt is input to the large-scale language model to determine whether or not to send a message, based on the message content and information about the sender and recipient. A second prompt is input to the large-scale language model to suggest revised message content, based on the result of the message content determination in response to the first prompt. As a result, the large-scale language model is used to determine whether or not to send a message and, if necessary, to suggest revised message content. This allows for message content review and determination of whether or not to send a message, as well as creation of revised message content, without the message content being known to third parties. The accuracy of the message content determination and creation of revised message content (i.e., message content review) can be improved by repeatedly reviewing the training and output results of the large-scale language model in advance. As described above, according to one aspect of this disclosure, it is possible to achieve appropriate review of message content in a communication tool while avoiding the message content being known to third parties. [Effects of the Invention]
[0009] According to one aspect of this disclosure, communication tools can enable proper review of transmitted content while preventing third parties from understanding the content being transmitted. [Brief explanation of the drawing]
[0010] [Figure 1] Figure 1 is a diagram showing the device configuration of the communication system according to this embodiment. [Figure 2] Figure 2 illustrates how multiple accounts belonging to the same person can be linked. [Figure 3] Figure 3 is a flowchart showing an example of a process performed by the communication system. [Figure 4] Figure 4 is a flowchart showing an example of a process performed by the communication system. [Figure 5] Figure 5 is a flowchart showing an example of a process performed by the communication system. [Figure 6] Figure 6 shows an example of the hardware configuration of the prompt generation system and server device. [Modes for carrying out the invention]
[0011] Embodiments of the present invention will be described in detail below with reference to the attached drawings. In the description of the drawings, the same reference numerals are used for the same or equivalent elements, and redundant explanations are omitted.
[0012] Figure 1 shows the device configuration of the communication system 1 according to this embodiment. The communication system 1 is a system that manages the sending and receiving of content in various communication tools. Here, communication tools refer to tools such as various SNS (Social Networking Services) and matching services. Sending and receiving content refers to the sending and receiving of content between a sender and a receiver. The sender is a user who sends content, for example, a user who sends content using the sending terminal 10. The receiver is a user who receives content, for example, a user who receives content using the receiving terminal 50. The sending terminal 10 and the receiving terminal 50 may be devices such as a personal computer, smartphone, tablet terminal, feature phone, server device, or game machine. Note that sending content does not only mean sending content one-to-one, but may also include posting or publishing content on a bulletin board, etc. In this case, the receiver may be one or more users who can access the bulletin board, etc. Below, an example in which content is directly sent from the sender's sending terminal 10 to the receiving terminal 50 of one receiver will be described.
[0013] The term "content" here includes all information that can be sent and received between the sender and receiver, and may include messages, images (pictures, photographs), videos, audio, etc., and may be sent and received in various file formats.
[0014] In communication system 1, when content is transmitted from the sender's sending terminal 10 to the receiver's receiving terminal 50, a determination is made as to whether the content should be sent to the receiving terminal 50, and the transmitted content is modified (i.e., proofread). By proofreading to ensure that the transmitted content does not contain inappropriate material or content that may offend the recipient, the communication environment in the communication tool can be kept healthy.
[0015] As shown in Figure 1, the communication system 1 includes a transmitting terminal 10, a receiving terminal 50, a prompt creation system 20, and a server device 30, all configured to communicate with each other via a network including a wireless communication network and a fixed communication network. The prompt creation system 20 and the server device 30 may be configured as a single integrated device. The prompt creation system 20 creates prompts based on information received from the transmitting terminal 10. A prompt is information indicating an instruction or question input to a large-scale language model in an interactive system such as a dialogue with a large-scale language model or a command-line interface (CLI).
[0016] The server device 30 is a device that holds a large-scale language model and enables the determination of whether or not content can be transmitted using the large-scale language model and the modification of transmitted content. The large-scale language model may be a model that, in response to the input of a prompt containing input information, generates an output according to one or a combination of the instructions, context, question, and output format indicated by the prompt, and returns the output as response information. In the following description, the large-scale language model will be described as an LLM (Large Language Model) 31. The LLM 31 is a proofreading LLM that proofreads the transmitted content.
[0017] LLM31 may be configured to include, for example, a user interface (UI) for interaction with a user, and may be an interactive AI that enables text or voice chats with the user. Examples of such generative AI models include ChatGPT, GPT (registered trademark)-3.5, GPT-4V, PaLM2, and the like. In the present embodiment, an example will be described in which the server device 30 determines whether content can be transmitted and modifies the transmitted content using one LLM31. However, the server device 30 may determine whether content can be transmitted and modify the transmitted content using multiple generative AI models such as multiple LLMs. Also, although only one server device 30 is illustrated in FIG. 1, the communication system 1 may include multiple server devices 30.
[0018] The prompt creation system 20 is configured to include, as functional components, a transmission / reception unit 21 (reception unit, transmission unit), a first prompt creation unit 22, a second prompt creation unit 23, an input unit 24, and a storage unit 25. Hereinafter, the functions of each functional unit of the prompt creation system 20 will be described in detail.
[0019] The transmission / reception unit 21 transmits and receives information to and from the sender's transmission terminal 10. The transmission / reception unit 21 receives the transmission content, which is the content of the content scheduled to be transmitted from the sender's transmission terminal 10 to the receiver's reception terminal 50. Specifically, when the content is transmitted from the transmission terminal 10 to the reception terminal 50, the transmission / reception unit 21 receives the content from the transmission terminal 10 first (before the reception terminal 50) to receive the transmission content, which is the content of the content scheduled to be transmitted from the transmission terminal 10 to the reception terminal 50. The transmission / reception unit 21 outputs the received transmission content, as well as the information on the sender and receiver of the content related to the received transmission content, to the first prompt creation unit 22 and the second prompt creation unit 23.
[0020] The first prompt creation unit 22 creates a first prompt for the LLM 31 to determine whether or not to send the message, based on the message content and information about the sender and recipient. In this way, by having the first prompt automatically created by the prompt creation system 20 rather than by the sender, it is possible to avoid problems such as the sender intentionally lowering the level of proofreading, which can occur when the sender creates the prompt, and to perform proper proofreading.
[0021] The first prompt creation unit 22 creates a first prompt that instructs the LLM 31 to determine whether the content to be transmitted contains the meaning or details regarding the words pre-stored in a specific table. Here, the table is stored, for example, in the storage unit 25. In the table, for example, first information, second information, third information, and fourth information are defined. The first information is information defined from the perspective of social common sense or compliance and policies. The second information is the registration information of each user including the sender and the receiver. The registration information is, for example, the user's attribute information such as age, biological gender, work history, affiliated company, nationality, occupation, etc. The third information is information that can be grasped such as past transmission and reception information (exchanged information) for each combination of the sender and the receiver. The fourth information is information such as the regulations determined by the master (company, management position, oneself, etc.) of each user including the sender and the receiver, the branding set by oneself, the information pre-registered as a policy, etc. By determining that transmission is not possible when the information (meaning or details) regarding the words pre-stored in such a specific table is included in the transmission content, appropriate proofreading can be performed. The first prompt creation unit 22 can create, for example, a prompt such as "Please determine based on the previous learning results whether the transmission content contains the meaning and details regarding the words defined in a specific table." In addition, as the merit of performing the determination using the LLM 31 (the merit compared with the case of collating with an ordinary database), it can be determined whether inappropriate content is included by judging not simply by keywords but as a sentence. For example, when determining a message such as "The weather is nice today [shine]", if it is determined by the NG keywords in the database, it may be determined as an inappropriate sentence just because the word [shine] is included. In this regard, by using the LLM 31 in which the sentence has been learned, it can be determined as a non-inappropriate sentence "The weather is nice today shine" by judging the sentence as a whole. Thus, the LLM 31 can determine whether a word is misrecognized as a prohibited word due to the connection with another word from the connection of the sentence.
[0022] The first prompt generation unit 22 may create a first prompt instructing the LLM 31 to determine whether the transmitted content contains confidential information. Confidential information includes, for example, trade secrets that are not publicly known, private information that should not be disclosed to other companies, etc. Confidential information may be defined in the specific table described above, or it may be determined from the wording. The first prompt generation unit 22 may, for example, create a prompt that says, "Based on the learning results so far, determine whether the transmitted content contains trade secrets that are not publicly known, or information that should not be disclosed to other companies."
[0023] The first prompt generation unit 22 may create a first prompt that instructs the LLM 31 to determine whether the content to be sent is appropriate as content to be sent from the sender to the receiver. Such a determination may be made, for example, based on the third information described above. The first prompt generation unit 22 can create a prompt such as, "Mr. / Ms. XX is trying to send the information indicated by 'Content to be sent' to Ms. YY. Please determine whether this transmission is appropriate based on the learning results so far."
[0024] The first prompt generation unit 22 may create a first prompt instructing the LLM 31 to determine whether the transmitted content is appropriate, taking into account the transmitted and received content between the sender and a user other than the receiver. Such a determination may be made, for example, based on the third information described above. It is conceivable that a sender might transmit content that they received from a user other than the receiver under the condition that it is prohibited from being shared with a third party, in violation of that condition, to a new recipient. In this regard, the creation of the first prompt makes it possible to determine that transmission is not permitted if the transmitted and received content between the sender and the other user is being passed on as is. In other words, because the transmitted content that the sender has received in the past is managed as direct data or as encrypted data, when the sender attempts to transmit the transmitted content, it can be compared with the data managed by the LLM 31, which can determine that it is information that has been passed on, and prevent the information from being transmitted.
[0025] The first prompt creation unit 22 may create a first prompt instructing the LLM 31 to determine whether another account exists belonging to the same person as the sender, and if so, to determine whether the content is appropriate considering the transmission method (transmission content) of the other account. The determination of whether another account exists belonging to the same person as the sender may be made, for example, based on the second information (registration information) described above. It is conceivable that information may be leaked by having multiple accounts on the same communication tool. For example, it is conceivable that content received with the condition that it is prohibited from being shared with third parties may be sent from a different account. In this regard, by making it mandatory to input information strongly linked to the individual when creating an account, it is possible to link multiple accounts and prevent the above situation. Specifically, when creating an account, information strongly linked to the individual (identity verification document information, employment information) can be set, and the leakage of information from accounts with similar attributes can be prevented. Figure 2 is a diagram illustrating the linking of multiple accounts by the same person. As shown in Figure 2, even if the senders appear to be different at first glance, the input of the aforementioned identity verification document information makes it possible to identify that each sender is an account belonging to the same person.
[0026] The output of LLM31 for the first prompt is repeatedly reviewed until an LLM31 that produces the appropriate output is completed. It is preferable to use an LLM31 that has undergone repeated review and training.
[0027] The first prompt generation unit 22 may create a first prompt that outputs whether the message cannot be sent, indicating the level at which it is determined that transmission is impossible: Level 1, where transmission is possible by making some modifications; or Level 2, where transmission is impossible even with modifications. Level 1 may be divided into, for example, Level 1-1 and Level 1-2. The degree of impossibility of transmission is considered to be increasing in the order of Level 2, Level 1-2, and Level 1-1. Level 2 contains information that should not be transmitted and cannot be replaced with other words (i.e., modified). Specifically, this includes trade secrets, information that should not be leaked to other companies, rumors, etc. Level 1-2 contains content that is problematic in itself and it is desirable to replace it with other words. Specifically, this includes excessive expressions that are inappropriate for the situation, expressions that are suspected of being in violation of compliance, expressions that are thought to include harassment, etc. Level 1-1 contains content that is not problematic in itself, but it is considered undesirable to transmit as is. Specifically, this includes incorrect honorific expressions to superiors, incorrect Japanese expressions, misspellings of kanji or spelling, etc. For transmissions determined to be Level 2 by LLM31, it is recommended to cancel the transmission of that content. For transmissions determined to be Level 1-1 or Level 1-2, modifications will be made according to the second prompt. Furthermore, for transmissions determined to be not unsendable (not Level 1 or Level 2), it is recommended to send them as they are.
[0028] The first prompt generation unit 22 outputs the generated first prompt to the input unit 24. The input unit 24 inputs the first prompt to the LLM 31 and also obtains the output result from the LLM 31 (transmission feasibility determination result) and stores it in the storage unit 25.
[0029] The second prompt generation unit 23 obtains the transmission feasibility determination result output from the LLM 31 in response to the first prompt from the storage unit 25, and creates a second prompt to cause the LLM 31 to propose the modified transmission content, which is the modified transmission content, based on the determination result. The second prompt generation unit 23 may create the second prompt only if the transmission feasibility determination result indicates that the transmission is not feasible at the first level (level 1-1 or level 1-2). The modified transmission content output from the LLM 31 in response to the second prompt replaces the part that was determined to be untransmittable with, for example, another similar expression that makes transmission possible.
[0030] The second prompt generation unit 23 may create a second prompt instructing the LLM 31 to propose revised content so that the structure of the content in the transmitted content does not change before and after the revision. The second prompt generation unit 23 may also create a second prompt instructing the LLM 31 to clearly indicate (distinguish) the revised parts in the revised content. For example, the second prompt generation unit 23 can create a prompt such as, "For the content that needs to be revised in the first prompt, please create a revised version without changing the structure of the sentences. Also, if there are inappropriate sentences or meanings that make it unsendable, please clearly indicate those parts (the parts to be replaced)." The revised content output from the LLM 31 in response to the second prompt may be completely new content, or it may maintain the original content as much as possible.
[0031] The second prompt generation unit 23 outputs the created second prompt to the input unit 24. The input unit 24 inputs the second prompt to the LLM 31 and also obtains the output result from the LLM 31 (modified transmission content) and stores it in the storage unit 25.
[0032] The transmitting / receiving unit 21 transmits the transmission feasibility determination result output from the LLM 31 in response to the first prompt, and the modified transmission content output from the LLM 31 in response to the second prompt, to the sender's transmitting terminal 10. The transmission feasibility determination result and the modified transmission content are stored in the storage unit 25. The transmitting / receiving unit 21 receives the transmission feasibility determination result and the modified transmission content by referring to the storage unit 25, and transmits this information to the transmitting terminal 10.
[0033] From the sender's perspective, it is preferable that the corrected message output from LLM31, the original message intended to be sent (original text), and the received message received from the receiving terminal 50 can all be viewed on the sending terminal 10. The advantage of being able to view the original text is that it allows for confirmation of corrections and confirmation of the original message that was intended to be conveyed. However, care must be taken in displaying the corrected message (actual message) and the original text, as this may lead to discrepancies between what was actually conveyed to the recipient and what was not. For example, it is preferable that the corrected message (actual message) and the received message be easily displayed, while the original text can only be viewed by clicking or tapping on the message field of the message to navigate to a separate page, or by displaying it below the message in a way that is only visible to the sender.
[0034] The transmitting / receiving unit 21 may display a warning message instructing the sender to cancel the transmission of content that has been determined by LLM 31 to be at Level 2. If the sender decides after seeing the warning message that it is safe to send the content, they may perform the transmission operation again. In this case, the sender may be notified that the content will not be sent directly to the recipient but will be disclosed to the administrator. This sender's decision is fed back to LLM 31, which can be used to update LLM 31. In this case, there is a concern that the content of the transmission and reception may be disclosed to a third party. Therefore, it is conceivable that LLM 31 may convert the content into example sentences with the same but different circumstances and provide them as training data.
[0035] Furthermore, communication between users and between users and LLM31 may be serverless and implemented using methods without a central administrator, such as blockchain. In that case, the data output by the algorithm may also be handled by blockchain. This would leave no trace on the server and prevent information leaks.
[0036] Next, we will describe the processes performed by communication system 1. Figure 3 is a flowchart showing an example of the processes performed by communication system 1.
[0037] As shown in Figure 3, when the sender's sending terminal 10 prepares to send content (step S1), the content is received and the identity of the sender and receiver is confirmed (step S2). Then, it is determined whether or not there are conditions for content that cannot be sent corresponding to the sender and receiver (step S3).
[0038] If present in step S3, the pre-specified conditions are incorporated (step S4), and if present, the relationship between the sender and receiver is checked to determine if there are any problems in sending the content (text) (step S5).
[0039] Then, it is determined whether the message to be sent contains any content that should not be sent, specifically, any Level 2 wording (Step S6). If it does, a message indicating that the message contains content that should not be sent is displayed on the sending terminal 10 (Step S9).
[0040] On the other hand, if the second level of wording is not included, it is determined whether the message content should be corrected to be better (step S7). If it is, the corrected message content is output from LLM31, and the original message and the suggested corrected message are displayed on the screen of the sending terminal 10 (step S8). Then, the actual message content to be sent is displayed on the sending terminal 10 (step S10), and the actual transmission is performed (step S11).
[0041] Figure 4 is a flowchart illustrating an example of the process performed by the communication system. Steps S101-S106 and S108-S112 in the flowchart of Figure 4 correspond to steps S1-S11 in the flowchart of Figure 3, respectively. In the process of Figure 4, a process (step S107) is also performed to determine whether the transmitted content includes information obtained from other third parties on the same platform. If such information is found, a message indicating that the transmitted content should not be sent is displayed (step S110).
[0042] Figure 5 is a flowchart illustrating an example of a process performed by the communication system. Steps S202 to S213 in the flowchart of Figure 5 correspond to steps S101 to S112 in the flowchart of Figure 4, respectively. The process in Figure 5 also includes an initial setting of recipient information (step S201). The recipient information here does not necessarily have to be information about a real person, and the recipient's account does not need to exist. Many chat tools are used as notepads to create drafts or groups containing only oneself. In this case, by setting a fictitious person who does not exist and sending the message, it is possible to proofread the content using LLM31. For example, suppose the recipient is set to "Our company president" or "The president of a business partner," the situation to "Business," the sending method to "Email," the purpose to send to "Schedule a meeting," and the details to "Do you know about (confidential information)?" In this case, based on the conditions, LLM31 will output a basic message for scheduling a meeting, including an external party. Furthermore, because the expression "Do you know?" is inappropriate, LLM31 suggests changing it to "Are you aware?" or similar. In addition, a message prohibiting transmission is displayed because it contains confidential company information. Receiving such information allows the sender to appropriately proofread the content before sending it.
[0043] Next, the effects and benefits of the communication system 1 according to this embodiment will be described.
[0044] Communication system 1 includes a transmission / reception unit 21 that receives transmission content, which is the content to be sent from the sender to the receiver; a first prompt creation unit 22 that creates a first prompt for the LLM 31 to determine whether or not to send based on the transmission content and information about the sender and receiver; a second prompt creation unit 23 that creates a second prompt for the LLM 31 to propose revised transmission content, which is the revised transmission content, based on the transmission feasibility determination result output from the LLM 31 in response to the first prompt; an input unit 24 that inputs the first prompt and the second prompt to the LLM 31; and a transmission / reception unit 21 that sends the transmission feasibility determination result output from the LLM 31 in response to the first prompt and the revised transmission content output from the LLM 31 in response to the second prompt to the sender.
[0045] In this communication system 1, a first prompt, created based on the content to be sent and information about the sender and recipient, is input to LLM31 to cause it to determine whether or not to send the message. A second prompt, created based on the result of the transmission feasibility determination in response to the first prompt, is input to the LLM31 model to cause LLM31 to propose revised content to be sent. As a result, LLM31 is used to determine whether or not to send the message and, if necessary, to propose revised content to be sent. This allows for the determination of whether or not to send the message and the creation of revised content, i.e., proofreading of the message content, without the content being known to third parties. By repeatedly performing training and reviewing the output results of LLM31 in advance, the accuracy of determining whether or not to send the message and creating revised content (i.e., proofreading of the message content) can be improved. In summary, communication system 1 enables appropriate proofreading of message content in a communication tool while avoiding the content being known to third parties.
[0046] The first prompt generation unit 22 creates a first prompt that outputs whether the level at which transmission is not possible is Level 1, where transmission can be made possible by making some modifications, or Level 2, where transmission cannot be made even with modifications. The second prompt generation unit 23 may create a second prompt only if the transmission failure level in the transmission feasibility determination result is Level 1. By separating the transmission failure levels in this way, appropriate actions can be taken according to each level (modification if possible, notification without modification if not possible, etc.), and the appropriate action can be taken for the content to be transmitted.
[0047] The first prompt generation unit 22 may create a first prompt that instructs the LLM 31 to determine whether the content to be sent contains meaning or details related to words pre-stored in a specific table. This makes it possible to determine more reliably and easily whether or not transmission is not possible by taking into account the information pre-stored in a specific table.
[0048] The first prompt generation unit 22 may create a first prompt instructing the LLM 31 to determine whether the transmitted content contains confidential information. This makes it possible to appropriately prevent the transmission of transmitted content that contains confidential information.
[0049] The first prompt generation unit 22 may create a first prompt that instructs the LLM 31 to determine whether the content to be transmitted is appropriate as content to be sent from the sender to the receiver. This makes it possible to determine whether the content to be transmitted is appropriate based on the combination of sender, receiver and content to be transmitted.
[0050] The second prompt generation unit 23 may create a second prompt instructing the LLM 31 to propose a revised transmission content so as not to change the content structure of the transmission content. This ensures that the transmission content is natural to the sender and allows for the easy and reliable creation of the revised transmission content.
[0051] The second prompt generation unit 23 may create a second prompt instructing the LLM 31 to clearly indicate the modified parts in the modified content to be sent. This makes it possible to clearly indicate the modified parts, allowing the sender to easily understand the changes.
[0052] The first prompt generation unit 22 may create a first prompt that instructs the LLM 31 to determine whether the transmission content is appropriate, taking into account the content transmitted and received between the sender and a user other than the receiver. This allows for appropriate rejection of transmission, for example, if the sender is leaking confidential information received from another user.
[0053] The first prompt generation unit 22 may create a first prompt that instructs the LLM 31 to determine whether another account exists belonging to the same person as the sender, and if so, to determine whether the content of the transmission is appropriate considering the transmission method of that other account. This makes it possible to appropriately prevent transmission in cases where the same person uses multiple accounts to leak confidential information, etc.
[0054] The system and method of this disclosure have the following configuration.
[0055] [1] A receiving unit that receives the content, which is the content that is to be sent from the sender to the recipient, A first prompt creation unit creates a first prompt for a large-scale language model to determine whether or not to send the message, based on the message content and the information of the sender and the recipient. A second prompt creation unit creates a second prompt to cause the large-scale language model to propose the modified transmission content, which is the modified transmission content, based on the transmission feasibility determination result output from the large-scale language model in response to the first prompt, An input unit that inputs the first prompt to the large-scale language model and the second prompt to the large-scale language model, A system comprising: a transmission unit that transmits to the sender the determination result of whether or not transmission is permitted, output from the large-scale language model in response to the first prompt, and the modified transmission content, output from the large-scale language model in response to the second prompt.
[0056] [2] The first prompt generation unit generates a first prompt that outputs whether it is a first level where transmission becomes possible by making some modifications, or a second level where transmission becomes impossible even with modifications, as the level at which transmission is determined to be impossible. The system according to [1], wherein the second prompt generation unit generates the second prompt only when the level of transmission failure in the transmission feasibility determination result is the first level.
[0057] [3] The system according to [1] or [2], wherein the first prompt generation unit generates the first prompt which instructs the large-scale language model to determine whether the transmitted content contains meaning or details related to words pre-stored in a specific table.
[0058] [4] The system according to any one of [1] to [3], wherein the first prompt generation unit creates the first prompt which instructs the large-scale language model to determine whether the transmitted content contains confidential information.
[0059] [5] The system according to any one of [1] to [5], wherein the first prompt generation unit creates the first prompt which instructs the large-scale language model to determine whether the content to be transmitted is appropriate as content to be transmitted from the sender to the receiver.
[0060] [6] The system according to any one of [1] to [5], wherein the second prompt generation unit creates the second prompt which instructs the large-scale language model to propose the modified transmission content so that the content structure in the transmission content does not change.
[0061] [7] The system according to any one of [1] to [6], wherein the second prompt generation unit creates the second prompt which instructs the large-scale language model to specify the modified parts in the modified transmitted content.
[0062] [8] The system according to any one of [1] to [7], wherein the first prompt generation unit creates the first prompt which instructs the large-scale language model to determine whether the transmitted content is appropriate, taking into account the transmitted and received content between the sender and a user other than the receiver.
[0063] [9] The system according to any one of [1] to [8], wherein the first prompt generation unit creates the first prompt which instructs the large-scale language model to determine whether or not another account exists belonging to the same person as the sender, and if so, to determine whether or not the transmitted content is appropriate considering the transmission method of the other account.
[0064]
[10] Receiving the content that is intended to be sent from the sender to the recipient, Based on the transmitted content and the information of the sender and receiver, a first prompt is created to cause a large-scale language model to determine whether or not to send the message. Based on the determination result of whether or not transmission is permissible, output from the large-scale language model in response to the first prompt, a second prompt is created to cause the large-scale language model to propose the modified transmission content, which is the modified transmission content. The first prompt is input to the large-scale language model, and the second prompt is input to the large-scale language model, A method comprising sending to the sender the result of determining whether transmission is permissible or not, output from the large-scale language model in response to the first prompt, and the modified transmission content, output from the large-scale language model in response to the second prompt.
[0065] The block diagram used in the description of the above embodiment shows functional units. These functional blocks (components) are realized by any combination of at least one of hardware and software. Furthermore, the method of realizing each functional block is not particularly limited. That is, each functional block may be realized using one device that is physically or logically coupled, or it may be realized using two or more physically or logically separated devices that are directly or indirectly connected (for example, using wired or wireless connections). A functional block may be realized by combining the above one device or the above multiple devices with software.
[0066] Functions include, but are not limited to, judgment, decision, judgment, calculation, calculation, processing, derivation, investigation, exploration, confirmation, reception, transmission, output, access, resolution, selection, selection, establishment, comparison, assumption, expectation, assumption, broadcasting, notifying, communicating, forwarding, configuring, reconfiguring, allocating (mapping), and assigning. For example, a functional block (configuration part) that enables transmission is called a transmitting unit or transmitter. As mentioned above, the method of implementation is not particularly limited.
[0067] For example, the prompt creation system 20 and server device 30 that constitute the communication system 1 in one embodiment of the present disclosure may function as a computer that processes the control method of the present disclosure. Figure 6 is a diagram showing an example of the hardware configuration of the prompt creation system 20 and server device 30 according to this embodiment. The prompt creation system 20 and server device 30 may be physically configured as a computer device including a processor 1001, memory 1002, storage 1003, communication device 1004, input device 1005, output device 1006, bus 1007, etc. Note that the prompt creation system 20 and server device 30 may be configured as a computer device including at least one processor such as a CPU or GPU, may be configured as a computer device including multiple processors, or may be configured to include multiple computer devices.
[0068] In the following explanation, the term "device" can be replaced with "circuit," "device," "unit," etc. The hardware configuration of the prompt creation system 20 and the server device 30 may include one or more of the devices shown in the figure, or it may be configured to omit some of the devices.
[0069] Each function in the prompt creation system 20 and the server device 30 is realized by loading predetermined software (programs) onto hardware such as the processor 1001 and memory 1002, which causes the processor 1001 to perform calculations, control communication by the communication device 1004, and control at least one of data reading and writing in the memory 1002 and storage 1003.
[0070] The processor 1001 controls the entire computer, for example, by running an operating system. The processor 1001 may be composed of a central processing unit (CPU) that includes interfaces with peripheral devices, control units, arithmetic units, registers, etc. For example, the above-mentioned transmitting and receiving unit 21 may be implemented by the processor 1001.
[0071] Furthermore, the processor 1001 reads programs (program code), software modules, data, etc., from at least one of the storage 1003 and the communication device 1004 into the memory 1002 and executes various processes accordingly. The program used is one that causes the computer to execute at least a part of the operations described in the above embodiment. For example, the transmitting and receiving unit 21 may be implemented by a control program stored in the memory 1002 and running on the processor 1001, and other functional blocks may be implemented similarly. The above-described various processes have been explained as being executed by one processor 1001, but they may be executed simultaneously or sequentially by two or more processors 1001. The processor 1001 may be implemented by one or more chips. The program may also be transmitted from a network via a telecommunications line.
[0072] Memory 1002 is a computer-readable recording medium and may consist of at least one of the following: ROM (Read Only Memory), EPROM (Erasable Programmable ROM), EEPROM (Electrically Erasable Programmable ROM), RAM (Random Access Memory), etc. Memory 1002 may also be called a register, cache, main memory, etc. Memory 1002 can store executable programs (program code), software modules, etc., for implementing a control method according to one embodiment of the present disclosure.
[0073] Storage 1003 is a computer-readable recording medium and may consist of at least one of the following: an optical disc such as a CD-ROM (Compact Disc ROM), a hard disk drive, a flexible disk, a magneto-optical disk (e.g., a compact disc, a digital multipurpose disc, a Blu-ray® disc), a smart card, flash memory (e.g., a card, a stick, a key drive), a floppy® disk, a magnetic strip, etc. Storage 1003 may also be called an auxiliary storage device. The above-mentioned storage medium may be, for example, a database, server, or other suitable medium including at least one of memory 1002 and storage 1003.
[0074] The communication device 1004 is hardware (transceiver / receiver device) for communicating between computers via at least one of a wired network and a wireless network, and is also referred to as a network device, network controller, network card, communication module, etc. The communication device 1004 may be configured to include a high-frequency switch, duplexer, filter, frequency synthesizer, etc., in order to implement at least one of frequency division duplex (FDD) and time division duplex (TDD). For example, the above-mentioned transmit / receive unit 21 may be implemented by the communication device 1004.
[0075] The input device 1005 is an input device that accepts input from an external source (e.g., a keyboard, mouse, microphone, switch, button, sensor, etc.). The output device 1006 is an output device that outputs to an external source (e.g., a display, speaker, LED lamp, etc.). The input device 1005 and the output device 1006 may be configured as an integrated unit (e.g., a touch panel).
[0076] Furthermore, each device, such as the processor 1001 and memory 1002, is connected by a bus 1007 for communicating information. The bus 1007 may be configured using a single bus, or different buses may be configured for each device.
[0077] Furthermore, the prompt generation system 20 and the server device 30 may be configured to include hardware such as a microprocessor, a digital signal processor (DSP), an ASIC (Application Specific Integrated Circuit), a PLD (Programmable Logic Device), and an FPGA (Field Programmable Gate Array), and some or all of each functional block may be realized by such hardware. For example, the processor 1001 may be implemented using at least one of these hardware components.
[0078] Information notification is not limited to the embodiments described herein and may be carried out by other means. For example, information notification may be carried out by physical layer signaling (e.g., DCI (Downlink Control Information), UCI (Uplink Control Information)), upper layer signaling (e.g., RRC (Radio Resource Control) signaling, MAC (Medium Access Control) signaling, broadcast information (MIB (Master Information Block), SIB (System Information Block))), other signals, or combinations thereof. RRC signaling may also be called RRC messages, and may be, for example, RRC Connection Setup messages, RRC Connection Reconfiguration messages, etc.
[0079] The processing procedures, sequences, flowcharts, etc., of each aspect / embodiment described herein may be reordered, provided they are consistent with each other. For example, the methods described herein present various step elements in an exemplary order and are not limited to that specific order.
[0080] Input and output information may be stored in a specific location (e.g., memory) or managed using a management table. Input and output information may be overwritten, updated, or appended to. Output information may be deleted. Input information may be transmitted to other devices.
[0081] The determination may be made by a value represented by 1 bit (0 or 1), by a boolean value (true or false), or by a numerical comparison (for example, a comparison with a predetermined value).
[0082] Each aspect / embodiment described herein may be used individually, in combination, or switched between as needed during implementation. Furthermore, notification of specific information (e.g., notification that "X is") is not limited to explicit notification, but may also be implicit (e.g., by not providing such notification).
[0083] Although the present disclosure has been described in detail above, it will be clear to those skilled in the art that the present disclosure is not limited to the embodiments described herein. The present disclosure can be implemented in modified and altered forms without departing from the intent and scope of the present disclosure as defined by the claims. Accordingly, the descriptions in the present disclosure are illustrative and not intended to be restrictive in any way.
[0084] Software should be broadly interpreted to mean instructions, instruction sets, code, code segments, program code, programs, subprograms, software modules, applications, software applications, software packages, routines, subroutines, objects, executable files, execution threads, procedures, functions, and so on, whether they are called software, firmware, middleware, microcode, hardware description languages, or by any other name.
[0085] Furthermore, software, instructions, information, etc., may be transmitted and received via a transmission medium. For example, if software is transmitted from a website, server, or other remote source using at least one of wired technologies (such as coaxial cable, fiber optic cable, twisted pair, or digital subscriber line (DSL)) and wireless technologies (such as infrared or microwave), then at least one of these wired and wireless technologies is included in the definition of a transmission medium.
[0086] The information, signals, etc. described in this disclosure may be represented using any of the various different techniques. For example, the data, instructions, commands, information, signals, bits, symbols, chips, etc. that may be referred to throughout the above description may be represented by voltage, current, electromagnetic waves, magnetic fields or magnetic particles, optical fields or photons, or any combination thereof.
[0087] In addition, terms used in this disclosure and terms necessary for understanding this disclosure may be replaced with terms having the same or similar meanings. For example, at least one of the channel and symbol may be a signal (signaling). Also, a signal may be a message. Furthermore, a component carrier (CC) may be called a carrier frequency, cell, frequency carrier, etc.
[0088] Furthermore, the information, parameters, etc., described in this disclosure may be expressed using absolute values, relative values from a given value, or corresponding other information. For example, wireless resources may be indicated by an index.
[0089] The names used for the parameters described above are not restrictive in any way. Furthermore, the formulas and other expressions using these parameters may differ from those expressly disclosed in this disclosure. Various channels (e.g., PUCCH, PDCCH, etc.) and information elements can be identified by any suitable name, and therefore, the various names assigned to these various channels and information elements are not restrictive in any way.
[0090] In this disclosure, terms such as "Mobile Station (MS)," "user terminal," "User Equipment (UE)," and "terminal" may be used interchangeably.
[0091] A mobile station may also be referred to by those skilled in the art as a subscriber station, mobile unit, subscriber unit, wireless unit, remote unit, mobile device, wireless device, wireless communication device, remote device, mobile subscriber station, access terminal, mobile terminal, wireless terminal, remote terminal, handset, user agent, mobile client, client, or some other appropriate term.
[0092] As used in this disclosure, the terms “determining” and “determining” may encompass a wide variety of actions. “Determining” may include, for example, judging, calculating, computing, processing, deriving, investigating, looking up, searching, or inquiring (e.g., searching in a table, database, or other data structure), or ascertaining. “Determining” may also include, for example, receiving (e.g., receiving information), transmitting (e.g., sending information), inputting, outputting, or accessing (e.g., accessing data in memory). Furthermore, "judgment" and "decision" can include considering something as having been "judged" or "decided" after resolving, selecting, choosing, establishing, comparing, etc. In other words, "judgment" and "decision" can include considering something as having been "judged" or "decided" after some action. Also, "judgment (decision)" can be reinterpreted as "assuming," "expecting," or "considering."
[0093] The terms “connected,” “coupled,” or any variation thereof, mean any direct or indirect connection or coupling between two or more elements, and may include the presence of one or more intermediate elements between two elements that are “connected” or “coupled” with each other. The coupling or connection between elements may be physical, logical, or a combination thereof. For example, “connection” may be reinterpreted as “access.” As used in this disclosure, two elements may be considered to be “connected” or “coupled” with each other using at least one of one or more wires, cables, and printed electrical connections, and, in some non-limiting and non-exclusive examples, electromagnetic energy having wavelengths in the radio frequency domain, microwave domain, and optical (both visible and invisible) domain.
[0094] In this disclosure, the phrase "based on" does not mean "based solely on" unless otherwise specified. In other words, the phrase "based on" means both "based solely on" and "based at least on."
[0095] Any reference to elements using designations such as “first,” “second,” etc., as used in this disclosure does not generally limit the quantity or order of those elements. These designations may be used in this disclosure as a convenient way to distinguish between two or more elements. Accordingly, references to first and second elements do not imply that only two elements may be adopted, or that the first element must precede the second element in any way.
[0096] Where the terms “include,” “including,” and their variations are used in this disclosure, these terms are intended to be inclusive, as is the term “comprising.” Furthermore, the term “or” as used in this disclosure is not intended to mean exclusive OR.
[0097] In this disclosure, if articles are added by translation, such as a, an, and the in English, this disclosure may include the fact that the noun following these articles is plural.
[0098] In this disclosure, the term "A and B are different" may mean "A and B are different from each other." The term may also mean "A and B are each different from C." Terms such as "separate" and "combine" may be interpreted similarly to "different." [Explanation of Symbols]
[0099] 21...Sending and receiving unit (receiving unit, transmitting unit), 22...First prompt creation unit, 23...Second prompt creation unit, 24...Input unit, 31...LLM (Large-Scale Language Model).
Claims
1. A receiving unit that receives the content, which is the content that is to be sent from the sender to the recipient, A first prompt creation unit creates a first prompt for a large-scale language model to determine whether or not to send the message, based on the message content and information about the sender and the recipient. A second prompt creation unit creates a second prompt to cause the large-scale language model to propose the modified transmission content, which is the modified transmission content, based on the transmission feasibility determination result output from the large-scale language model in response to the first prompt, An input unit that inputs the first prompt to the large-scale language model and the second prompt to the large-scale language model, A system comprising: a transmission unit that transmits to the sender the determination result of whether or not transmission is permitted, output from the large-scale language model in response to the first prompt, and the modified transmission content, output from the large-scale language model in response to the second prompt.
2. The first prompt generation unit generates a first prompt that outputs whether it is a first level where transmission becomes possible by making some modifications, or a second level where transmission becomes impossible even with modifications, as the level at which transmission is determined to be impossible. The system according to claim 1, wherein the second prompt generation unit generates the second prompt only when the level of transmission failure in the transmission feasibility determination result is the first level.
3. The system according to claim 1, wherein the first prompt generation unit generates the first prompt which instructs the large-scale language model to determine whether the transmitted content contains meaning or details relating to words pre-stored in a specific table.
4. The system according to claim 1, wherein the first prompt generation unit generates the first prompt which instructs the large-scale language model to determine whether the transmitted content contains confidential information.
5. The system according to claim 1, wherein the first prompt generation unit generates the first prompt which instructs the large-scale language model to determine whether the content to be transmitted is appropriate as content to be transmitted from the sender to the receiver.
6. The system according to claim 1, wherein the second prompt generation unit generates a second prompt that instructs the large-scale language model to propose the modified transmission content so as not to change the content structure in the transmission content.
7. The system according to claim 1, wherein the second prompt generation unit generates the second prompt which instructs the large-scale language model to specify the modified parts in the modified transmitted content.
8. The system according to claim 1, wherein the first prompt generation unit generates the first prompt which instructs the large-scale language model to determine whether the transmitted content is appropriate, taking into consideration the transmitted and received content between the sender and a user other than the receiver.
9. The system according to claim 1, wherein the first prompt generation unit generates a first prompt that instructs the large-scale language model to determine whether or not another account exists belonging to the same person as the sender, and if such an account exists, to determine whether or not the transmitted content is appropriate considering the transmission method of the other account.
10. Receiving the content that is intended to be sent from the sender to the recipient, Based on the transmitted content and the information of the sender and receiver, a first prompt is created to cause a large-scale language model to determine whether or not to transmit the message. Based on the determination result of whether or not transmission is permitted output from the large-scale language model in response to the first prompt, a second prompt is created to cause the large-scale language model to propose the modified transmission content, which is the modified transmission content. The first prompt is input to the large-scale language model, and the second prompt is input to the large-scale language model, A method comprising sending to the sender the result of determining whether transmission is permissible or not, output from the large-scale language model in response to the first prompt, and the modified transmission content, output from the large-scale language model in response to the second prompt.