Communication method and apparatus, and storage medium

Abstract

Provided in the present application are a communication method and apparatus, and a storage medium, which aim to improve the accuracy and coherence of a response from a call agent, so as to improve the usage experience of a user. The method comprises: acquiring text data of an nth round of interaction of a first call of a user and text data of (n-1) rounds of interaction prior to the nth round of interaction; sending first information to a DCAS, wherein the first information is used for requesting the acquisition of historical call information of the user; receiving second information from the DCAS, wherein the second information comprises first historical call information of the user; sending third information to an LLM, wherein the third information comprises the text data of the nth round of interaction, the text data of the (n-1) rounds of interaction, and the first historical call information; and receiving fourth information from the LLM, wherein the fourth information comprises an inference result corresponding to the nth round of interaction.

Description

Communication methods, devices and storage media

[0001] This application claims priority to Chinese Patent Application No. 202510237936.4, filed on February 28, 2025, entitled "Communication Method, Apparatus and Storage Medium", the entire contents of which are incorporated herein by reference. Technical Field

[0002] This application relates to the field of communications, and more particularly to a communication method, apparatus, and storage medium. Background Technology

[0003] An agent is a fundamental concept in the field of artificial intelligence. It refers to a system capable of operating autonomously and interacting with its environment. The initial purpose of developing agents was to simulate the intelligent behavior of humans or other organisms, aiming to automate problem-solving or task execution. Large language models (LLMs) possess powerful learning and planning capabilities, enabling them to handle more complex and abstract tasks, and demonstrating superior performance in natural language understanding, reasoning, and decision-making. Within the context of LLMs, agents can possess the ability to autonomously understand, plan, make decisions, and execute complex tasks.

[0004] Call assistants represent a new business scenario under the new call network architecture, and they are implemented based on call agents. However, in the current implementation, each call initiated by a user to the call assistant is treated as a completely new call, limiting the information available for the call agent to perform inference, which negatively impacts the user experience. Summary of the Invention

[0005] This application provides a communication method, apparatus, and storage medium to improve the accuracy and consistency of responses from intelligent agents during calls, thereby enhancing the user experience.

[0006] Firstly, a communication method is provided that can be applied to a subscriber AI agent function (SAAF) or a module in the SAAF, such as a circuit or chip in the SAAF. The following description uses the application of this method to SAAF as an example.

[0007] The method includes: acquiring text data of the nth round of interaction of a user's first call and text data of the (n-1)th round of interaction prior to the nth round; sending first information to a data channel application server (DCAS), the first information being used to request the acquisition of the user's historical call information; receiving second information from the DCAS, the second information including the user's first historical call information; sending third information to an LLM, the third information including text data of the nth round of interaction, text data of the (n-1)th round of interaction, and the first historical call information; and receiving fourth information from the LLM, the fourth information including the inference result corresponding to the nth round of interaction.

[0008] Here, the first call refers to the current call between the user and the call assistant, and the nth round of interaction refers to the current round of interaction between the user and the call assistant. A call may include one or more rounds of interaction.

[0009] In this application, the DCAS stores the user's historical call information, which includes text data of historical calls initiated by the user before the first call. After obtaining the text data of the nth round of interaction of the user's first call, SAAF can obtain the user's first historical call information from DCAS. Then, SAAF can use the text data of the nth round of interaction, the text data of the previous nl rounds of interaction, and the first historical call information as input to LLM, so that LLM can combine the first historical call information to more accurately understand the text data of the nth round of interaction and output more accurate and coherent reasoning results.

[0010] In conjunction with the first aspect, in some implementations of the first aspect, the first information includes the range of the user's historical calls to be obtained, and the first historical call information is the user's historical call information within the range of historical calls.

[0011] In this application, the first information includes the range of the user's historical calls to be obtained, so that DCAS can determine the first historical call information within the range of the historical calls from the user's historical call information and send the first historical call information to SAAF. This is beneficial for obtaining effective historical call information, avoiding redundancy and burden on LLM inference, and reducing signaling overhead.

[0012] In conjunction with the first aspect, in some implementations of the first aspect, the first information includes text data of the nth round of interaction, which is used to determine the first historical call information.

[0013] In this application, the first information includes the text data of the nth round of interaction, so that DCAS can determine the first historical call information that is semantically closest to the text data of the nth round of interaction from the user's historical call information, and send the first historical call information to SAAF. This is beneficial for obtaining effective historical call information, avoiding redundancy and burden on LLM inference, and reducing signaling overhead.

[0014] In conjunction with the first aspect, in some implementations of the first aspect, the first information includes the user's identifier, which is used to determine the first historical call information from the historical call information of multiple users.

[0015] It is understandable that DCAS may store historical call information for multiple users. Therefore, the first information includes the user's identifier so that DCAS can determine the first historical call information related to the user from the historical call information of multiple users. This is beneficial for obtaining valid historical call information and for reducing signaling overhead.

[0016] In conjunction with the first aspect, in some implementations of the first aspect, the first information includes a first parameter, which is used to restore the user's historical call information. The first parameter can also be called a restoration parameter.

[0017] Understandably, in order to improve data security, the historical call information stored in DCAS can be protected historical call information. Accordingly, when SAAF obtains the historical call information of a user, it can carry a first parameter to instruct DCAS to restore the historical call information of the user that is requested.

[0018] In conjunction with the first aspect, in some implementations of the first aspect, the restoration process includes decryption, decoding, or deobfuscation.

[0019] In conjunction with the first aspect, in some implementations of the first aspect, the method further includes: sending a fifth message to the DCAS, the fifth message being used to request the storage of the text data of the user's first call; and receiving a sixth message from the DCAS, the sixth message being used to indicate that the text data of the first call has been stored.

[0020] After the first call ends, SAAF can request DCAS to store the text data of the first call so that it can retrieve the text data of the first call from DCAS in subsequent calls, helping LLM to better understand the user's intent and provide more accurate and coherent reasoning results.

[0021] In conjunction with the first aspect, in some implementations of the first aspect, the fifth piece of information includes the user's identifier. In this way, DCAS can associate the user's identifier with the text data of the first call, enabling more efficient retrieval of the user's first call text data subsequently.

[0022] In conjunction with the first aspect, in some implementations of the first aspect, the fifth information includes a second parameter, which is used to protect the text data of the first call. The second parameter can also be referred to as a protection parameter.

[0023] In this application, the text data of the first call can be protected by the second parameter, which helps to securely store the text data of the user's first call and improves data security.

[0024] In conjunction with the first aspect, in some implementations of the first aspect, the protection process includes encryption, encoding, or obfuscation.

[0025] Encryption makes text data difficult to crack even if it is leaked during storage or transmission. Encoding converts text data into a specific encoding format, increasing its complexity and security. Obfuscation transforms text data in a way that makes it difficult to directly recognize and understand.

[0026] Secondly, a communication method is provided that can be applied to DCAS or modules in DCAS, such as circuits or chips in DCAS. The following description uses the application of this method to DCAS as an example.

[0027] The method includes: receiving first information from SAAF, the first information being used to request the acquisition of the user's historical call information; and sending second information to SAAF, the second information including the user's first historical call information.

[0028] In this application, the DCAS stores the user's historical call information. After obtaining the text data of the nth round of interaction of the user's first call, the SAAF can obtain the user's first historical call information from the DCAS. Then, the SAAF can use the text data of the nth round of interaction, the text data of the previous nl rounds of interaction, and the first historical call information as input to the LLM, so that the LLM can combine the first historical call information to more accurately understand the text data of the nth round of interaction and output more accurate and coherent reasoning results.

[0029] In conjunction with the second aspect, in some implementations of the second aspect, the first information includes the range of the user's historical calls to be obtained, and the first historical call information is the user's historical call information within the range of historical calls.

[0030] In conjunction with the second aspect, in some implementations of the second aspect, the first information includes text data of the nth round of interaction of the user's first call, which is used to determine the first historical call information.

[0031] In conjunction with the second aspect, in some implementations of the second aspect, the first information includes the user's identifier, which is used to determine the first historical call information from the historical call information of multiple users.

[0032] In conjunction with the second aspect, in some implementations of the second aspect, the first information includes a first parameter, which is used to restore the user's historical call information. The first parameter can also be called a restoration parameter.

[0033] In conjunction with the second aspect, in some implementations of the second aspect, the restoration process includes decryption, decoding, or deobfuscation.

[0034] In conjunction with the second aspect, in some implementations of the second aspect, the method further includes: receiving fifth information from SAAF, the fifth information being used to request the storage of text data of the user's first call; storing the text data of the first call based on the fifth information; and sending a sixth information to SAAF, the sixth information being used to indicate that the text data of the first call has been stored.

[0035] In conjunction with the second aspect, in some implementations of the second aspect, the fifth information includes the user's identifier. Based on the fifth information, storing the text data of the first call includes: storing the text data of the first call in key-value pairs, using the user's identifier as the key and the text data of the first call as the value. In this application, DCAS persistently stores the text data of the first call in key-value pairs, making the historical call information stored in DCAS persistently available.

[0036] In conjunction with the second aspect, in some implementations of the second aspect, the fifth information includes a second parameter, which is used to protect the text data of the first call. Based on the fifth information, storing the text data of the first call includes storing the text data of the first call after protecting it based on the second parameter. This helps improve data security and prevents the user's historical call information from being leaked or tampered with.

[0037] In conjunction with the second aspect, in some implementations of the second aspect, the protection process includes encryption, encoding, or obfuscation.

[0038] It should be understood that the second aspect of this application corresponds to the technical solution of the first aspect of this application, and the beneficial effects achieved by each aspect and the corresponding feasible implementation are similar, and will not be repeated here.

[0039] Thirdly, a communication apparatus is provided for executing the method in any possible implementation of any of the above aspects. Specifically, the apparatus includes a module for executing the method in any possible implementation of any of the above aspects.

[0040] In one design, the device may include modules that perform the methods / operations / steps / actions described in any of the above aspects. These modules may be hardware circuits, software, or a combination of hardware circuits and software.

[0041] In another design, the device is a communication chip, which may include input circuitry or interface for transmitting information or data, and output circuitry or interface for receiving information or data.

[0042] In another design, the device is a SAAF or DCAS, which may include a transmitter for sending information or data and a receiver for receiving information or data.

[0043] In another design, the device is used to perform the method in any of the possible implementations of any of the above aspects, and the device can be configured in SAAF or DCAS.

[0044] Fourthly, a communication device is provided, comprising at least one processor for calling and running a computer program from a memory, such that the device performs the method in any possible implementation of any of the preceding aspects.

[0045] Optionally, the device further includes a memory for storing instructions and data. The memory is coupled to the processor, which, when executing the instructions stored in the memory, can implement the methods described in the foregoing aspects.

[0046] Optionally, the device may also include a transmitter and a receiver, which may be separate or integrated together and referred to as a transceiver.

[0047] Fifthly, a computer program product is provided, the computer program product comprising: a computer program (also referred to as code or instructions), which, when the computer program is run, causes a computer to perform a method in any possible implementation of any of the above aspects.

[0048] In a sixth aspect, a computer-readable storage medium is provided that stores a computer program (also referred to as code or instructions) that, when executed on a computer, causes the computer to perform the methods in any possible implementation of any of the above aspects.

[0049] In a seventh aspect, this application provides a chip system including at least one processor for supporting the implementation of the functions involved in any possible implementation of any of the above aspects, such as receiving or processing data involved in the above methods.

[0050] In one possible design, the chip system also includes a memory for storing program instructions and data, which may be located within or outside the processor.

[0051] Optionally, the chip system may consist of chips or may include chips and other discrete components.

[0052] Eighthly, this application provides a communication system including a SAAF for implementing the method described in the first aspect and any possible implementation thereof, and a DCAS for implementing the method described in the second aspect and any possible implementation thereof. Optionally, the communication system further includes an LLM.

[0053] It should be understood that the third to eighth aspects of this application correspond to the technical solutions of the first to second aspects of this application, and the beneficial effects achieved by each aspect and the corresponding feasible implementation are similar, and will not be repeated here. Attached Figure Description

[0054] Figure 1 is a schematic diagram of the composition of an LLM-based intelligent agent;

[0055] Figure 2 is a schematic diagram of a call assistant service;

[0056] Figure 3 is a schematic diagram of the network architecture of a call agent provided in an embodiment of this application;

[0057] Figure 4 is a schematic diagram of a user interacting with a call assistant according to an embodiment of this application;

[0058] Figure 5 is a schematic flowchart of a communication method provided in an embodiment of this application;

[0059] Figures 6A and 6B are schematic diagrams of user interaction with a call assistant provided in the embodiments of this application;

[0060] Figures 7 and 8 are schematic block diagrams of the communication device provided in the embodiments of this application. Detailed Implementation

[0061] The technical solutions in this application will now be described with reference to the accompanying drawings.

[0062] Before introducing the technical solutions provided in the embodiments of this application, the following points should be made first.

[0063] First, in the embodiments shown below, the terms and English abbreviations, such as intelligent agent, large language model (LLM), data channel application server (DCAS), user intelligent agent function (SAAF), etc., are all exemplary examples given for ease of description and should not constitute any limitation on this application. This application does not preclude the possibility of defining other terms that can achieve the same or similar functions in existing or future agreements.

[0064] Second, in the embodiments shown below, the terms "first," "second," and various numerical designations are merely for descriptive convenience to distinguish identical or similar items with substantially the same function and purpose. For example, "first information" and "second information" are only used to distinguish different information and do not limit their order, nor are they used to limit the scope of the embodiments of this application. Those skilled in the art will understand that the terms "first," "second," etc., do not limit the quantity or execution order, and that "first," "second," etc., are not necessarily different.

[0065] Third, "at least one" means one or more, while "more than one" means two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can mean: A alone, A and B simultaneously, or B alone, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, and c can mean: a, or b, or c, or a and b, or a and c, or b and c, or a, b, and c, where a, b, and c can be single or multiple.

[0066] Fourth, in this application, "instruction" can include direct and indirect instructions, explicit and implicit instructions, and instructions used for determination. When describing certain instruction information to indicate A, it can be understood that the instruction information carries A, directly indicates A, or indirectly indicates A. In this application, the information indicated by the instruction information is called the information to be instructed. In specific implementation, there are many ways to indicate the information to be instructed, such as, but not limited to, directly indicating the information to be instructed, such as the information to be instructed itself or its index. It can also indirectly indicate the information to be instructed by indicating other information, where there is a correlation between the other information and the information to be instructed. It can also indicate only a part of the information to be instructed, while the other parts are known or pre-agreed upon. For example, the instruction of specific information can be achieved by using a pre-agreed (e.g., protocol-defined) arrangement of various information, thereby reducing instruction overhead to some extent. Furthermore, the information to be instructed can be sent as a whole or divided into multiple sub-information to be sent separately, and the sending period and / or timing of these sub-information can be the same or different. This application does not limit the specific method of instruction. It is understood that, for the sender of the instruction information, the instruction information can be used to indicate the information to be indicated, and for the receiver of the instruction information, the instruction information can be used to determine the information to be indicated.

[0067] The information in this application is used to indicate one or more contents, or it may be replaced with the information indicating one or more contents, or the information including one or more contents.

[0068] Fifth, in this application, "when," "if," and "if" all refer to the device taking corresponding actions under certain objective circumstances, not to a time limit, nor do they require the device to perform a judgment action, nor do they imply any other limitations. Unless otherwise specified, "if" and "if" are interchangeable, and "when" and "under the circumstances" are interchangeable. "When" is interchangeable with "if" / "if."

[0069] Sixth, in this application, the words "exemplarily" or "for example" are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design that is described as "exemplarily" or "for example" in this application should not be construed as being more preferred or advantageous than other embodiments or design options. Specifically, the use of the words "exemplarily" or "for example" is intended to present the relevant concepts in a specific manner.

[0070] Seventh, "Sending information / data" only indicates the direction of information / data transmission, including direct transmission via the device's communication interface (such as an air interface). "Sending" can also be understood as the "output" of the module interface. "Sending" can include indirect transmission by the processing unit through the communication interface, that is, after the processing unit outputs information / data through the module interface, it is transmitted to the device's communication interface and then sent out. "Receiving information / data" only indicates the direction of information / data transmission, including direct reception via the communication interface. "Receiving" can also be understood as the "input" of the module interface. "Receiving information / data" can include indirect reception by the processing unit through the communication interface, that is, after the communication interface receives information / data, it is transmitted to the module interface of the processing unit and then input to the processing unit. "Sending information / data to… (such as a terminal device)" can be understood as the destination of the information being the terminal device. It can include sending information / data directly or indirectly to the terminal device. "Receiving information / data from… (such as a terminal device)" can be understood as the source of the information being the terminal device, and can include receiving information / data directly or indirectly from the terminal device. Information / data may undergo necessary processing, such as format changes, between the source and destination, but the destination can understand the valid information / data from the source. Similar statements in this application can be understood in a similar way, and will not be repeated here.

[0071] In other words, sending and receiving can occur between devices, such as between terminal devices and network devices; or they can occur within a device, such as between components, modules, chips, software modules, or hardware modules within a device via a bus, wiring, or interface.

[0072] Eighth, in this application, the solutions in each embodiment can be used in reasonable combinations, and the explanations or descriptions of various terms, similar operations, or steps appearing in the embodiments can be referenced or explained to each other in the various embodiments, without limitation.

[0073] For ease of understanding, the relevant technologies and concepts involved in this application are introduced below.

[0074] 1. LLM-based intelligent agents.

[0075] An intelligent agent is a fundamental concept in the field of artificial intelligence, referring to a system that can operate autonomously and interact with its environment. The initial goal of developing intelligent agents was to simulate the intelligent behavior of humans or other organisms, aiming to automate problem-solving or task execution. LLM (Learning-Language Modeling) possesses powerful learning and planning capabilities, can handle more complex and abstract tasks, and demonstrates superior performance in natural language understanding, reasoning, and decision-making. In the context of LLM, an intelligent agent can possess the ability to autonomously understand, plan, make decisions, and execute complex tasks. The intelligent agent discussed below can be understood as an LLM-based intelligent agent.

[0076] In this application, the intelligent agent may also be referred to as an artificial intelligence agent (AI agent).

[0077] As shown in Figure 1, in an LLM-based agent, the LLM acts as the "brain" of the agent. Further expansion of the agent can include several key components: planning, memory, tool use, and action. These components are briefly introduced below.

[0078] Planning: Introducing a human-like problem-solving approach to the intelligent agent involves breaking down complex tasks into a series of simpler subtasks, which are then solved one by one. This method reduces the difficulty of solving the task all at once, helping to improve problem-solving efficiency and effectiveness, and enhancing the agent's adaptability to complex environments and operational reliability. To generate effective plans, the agent can simultaneously generate multiple candidate solutions and select the best one for execution. When dealing with complex tasks, the agent can also iteratively optimize based on real-time feedback from the environment, thereby solving problems involving complex reasoning more efficiently. Depending on the approach to the task, planning can include "one-time" decomposition of the task and generation steps (referred to as Plan A) and "iterative" decomposition of the task and generation steps (referred to as Plan B). Examples of Plan A include Chain of Thought (CoT) and Plan and Solve (PS), while examples of Plan B include Reasoning and Acting (ReAct) and Tree of Thought (ToT).

[0079] Memory: The human memory system is a complex and efficient information processing system. It can store new knowledge and review and use stored information when needed to assist in coping with the current environment and making decisions. Similarly, in an intelligent agent, the memory component constitutes the core storage unit, mainly used to store historical interaction records between the agent and the environment, and can be retrieved and used at any time. Memory can be further divided into short-term memory and long-term memory. Short-term memory corresponds to historical context, and LLM can memorize and learn within a certain context length. Long-term memory refers to the larger knowledge base that LLM can access. It is usually combined with retrieval-augmented generation (RAG) technology, using a vector database as a carrier, to obtain updated and expanded knowledge base from the vector database, supplementing the context information when LLM answers questions and improving the accuracy of the answers.

[0080] Tool Usage: For information missing in LLM, such as real-time weather, real-time search results, and mathematical reasoning capabilities, the agent obtains additional information by calling external application programming interfaces (APIs). Tools can be functions, API calls, or separately designed solutions, and may also include code interpreters.

[0081] Actions: These are the parts where the intelligent agent actually executes decisions or responses. Faced with different tasks, the intelligent agent can choose the actions to perform when making decisions, such as memory retrieval, reasoning, learning, and programming. Actions also include environmental awareness, which allows the intelligent agent to receive the results of external tool calls or real-time environmental information, thus responding more flexibly to different situations.

[0082] 2. Prompt engineering

[0083] Prompt engineering is used to design effective input prompts to guide the LLM in generating high-quality and relevant responses. These input prompts are called prompts, and the agent and LLM interact using them. Table 1 shows an example of the components of a prompt.

[0084] Table 1

[0085] 3. Call Assistant Service

[0086] Call assistant services are a brand-new business scenario under the new call network architecture, implemented based on call intelligence agents. As shown in Figure 2, call assistant services can include intelligent answering (referred to as "chatting on behalf"), intelligent assistance (referred to as "chatting assistance"), emotional companionship (referred to as "chatting companion"), and personal assistant / enabling third parties (referred to as "call entry point"). Each of these categories can further include multiple vertical business scenarios (referred to as "vertical domain"), such as celebrity callers, food ordering assistants, and mobile hall assistants.

[0087] Beyond intelligent voice assistants, other business scenarios involve human-computer interaction calls, such as users calling machines / call assistants (consumer-to-machine, C2M) or machines / call assistants calling users (machine-to-consumer, M2C). During the call, users interact with the intelligent agent through voice interaction. For example, in the food ordering assistant category, users provide feedback on their preferences, delivery addresses, and other needs to the intelligent agent through multiple rounds of interaction; in the celebrity caller category, users discuss the latest movies, music, and other topics with celebrities through multiple rounds of interaction; and in the hotel assistant category, users communicate information such as prices and reservation times with the intelligent agent through multiple rounds of interaction.

[0088] The call agent tries to remember as much as possible the content of multiple interactions with the user during the call, caching the user's questions, opinions and needs in the context memory component of the call so that it can better understand the user's intentions in the next round of interaction, give more accurate answers to the user's questions, ensure the logical coherence between past interactions and the next round of interaction, and bring a more efficient communication experience to the user.

[0089] Figure 3 is a schematic diagram of a network architecture for a call agent provided in an embodiment of this application. As shown in Figure 3, some network elements in this network architecture are as follows: AI agent management function (AAMF), SAAF, subscriber vector database function (SVDF), DCAS, media function (MF), data channel signaling function (DCSF), voice over long-term evolution application server (VoLTE AS), interrogating-call session control function (I-CSCF) in Internet Protocol Multimedia Subsystem (IMS), proxy-call session control function (P-CSCF) in IMS, serving-call session control function (S-CSCF) in IMS, IMS access media gateway (IMS-AGW), and LLM.

[0090] Among them, AAMF is responsible for allocating SAAF and SVDF instances to users when they power on or make business requests.

[0091] SAAF is the call agent, responsible for the logical functions of the user agent.

[0092] SVDF is responsible for storing the vectorized version of general user information, as well as storing personalized user data.

[0093] DCAS is used to provide logical orchestration for (data channel, DC) services.

[0094] MF is used to provide media capabilities for DC services such as rendering and voice translation.

[0095] DCSF is responsible for managing DC media channels and DC service triggers, and provides northbound interfaces for service calls.

[0096] VoLTE AS is responsible for handling call signaling and control, including call setup, maintenance, and termination.

[0097] Because LLMs are inherently stateless, in the new call network architecture, each time a user initiates a call with the call assistant, it is considered a completely new call. This requires the call agent to re-understand the user's needs. For example, as shown in Figure 4, a user elaborates on a problem in a call and has an in-depth discussion with the call assistant (SAAF). However, when the user initiates another call, the call assistant cannot access the previous call content, requiring the user to rephrase the problem, wasting the user's time and effort. Furthermore, this lack of memory for past call content fails to fully utilize the context of previous calls, affecting the accuracy and coherence of the call assistant's responses and reducing the user's communication experience.

[0098] In view of this, embodiments of this application provide a communication method in which DCAS is used to store the historical context of a user's past calls. SAAF can obtain the historical context of past calls from DCAS and send the call content of the current call and the historical context of past calls to LLM. In this way, based on the call content of the current call and the historical context of past calls, LLM can better understand the user's intent, improve the accuracy and coherence of the response, and bring an efficient communication experience to the user.

[0099] In the following embodiments, SAAF can be replaced by a multi-modal communication function (MCF), and DCAS can be replaced by a service application server.

[0100] Figure 5 is a schematic flowchart of a communication method 500 provided in an embodiment of this application. The steps of method 500 can be executed interactively by SAAF (or modules in SAAF, such as processors, chips, chip systems, circuits, etc.) and DCAS (or modules in DCAS, such as processors, chips, chip systems, circuits, etc.) and LLM. In addition, the processing performed by a single execution entity can also be divided into multiple execution entities, which can be logically and / or physically separated.

[0101] Method 500 includes, but is not limited to, S501 to S505. Optionally, method 500 also includes S506 and S507. The steps are described in detail below.

[0102] S501, SAAF obtains the text data of the nth round of interaction of the user's first call and the text data of the n-1 rounds of interaction before the nth round of interaction.

[0103] Here, the first call refers to the current call between the user and the call assistant, and the nth round of interaction refers to the current round of interaction between the user and the call assistant. A call may include one or more rounds of interaction.

[0104] Optionally, before the SAAF obtains the text data of the nth round of interaction of the user's first call, the SAAF can receive the audio data of the nth round of interaction of the user's first call from the MF. Afterwards, the SAAF can use automatic speech recognition (ASR) technology to convert the audio data of the nth round of interaction into text data, thereby obtaining the text data of the nth round of interaction.

[0105] Optionally, SAAF obtains the text data of the n-1 rounds of interaction prior to the nth round of interaction, including: SAAF obtaining the text data of the n-1 rounds of interaction prior to the nth round of interaction from its own memory.

[0106] S502, SAAF sends a first message to DCAS, the first message being used to request the user's historical call information. Accordingly, DCAS receives the first message.

[0107] In the application embodiment, the user's historical call information is stored in DCAS. To effectively utilize this historical call information and improve the accuracy and consistency of the call assistant's responses, SAAF can request the user's historical call information from DCAS after obtaining the text data of the nth round of interaction in the user's first call. Subsequently, SSAF can use the user's historical call information as part of the LLM input for the LLM to infer from the text data of the nth round of interaction. This historical call information includes the text data of at least one historical call prior to the first call, and each historical call includes one or more rounds of interaction.

[0108] S503, DCAS sends a second message to SAAF, which includes the user's first historical call information. SAAF receives the second message accordingly.

[0109] Based on the request from SAAF, DCAS determines the first historical call information from the user's historical call information and sends the user's first historical call information to SAAF.

[0110] S504, SAAF sends third information to LLM, which includes text data from the nth round of interaction, text data from the (n-1)th rounds of interaction preceding the nth round, and first historical call information. Accordingly, LLM receives the third information.

[0111] SAAF and LLM can interact based on cue words. For example, in the example of the components of a cue word above, the cue word can include short-term memory and long-term memory. SAAF can use the text data of the nth round of interaction and the text data of the n-1 rounds of interaction before the nth round of interaction as the content of the short-term memory of the cue word, and use the first historical call information as the content of the long-term memory of the cue word.

[0112] S505, the LLM sends a fourth message to the SAAF, which includes the inference result corresponding to the nth round of interaction. The SAAF receives the fourth message accordingly.

[0113] After receiving the third information, the LLM generates an inference result corresponding to the text information of the nth round of interaction of the first call based on the content included in the third information, and sends the inference result to the SAAF.

[0114] In the application embodiment, the user's historical call information is stored in DCAS. After obtaining the text data of the user's first call's nth round of interaction, SAAF can retrieve the user's first historical call information from DCAS and send the text data of the first call and the first historical call information to LLM. In this way, LLM can combine the first historical call information with the text information of the n-1 rounds of interaction prior to the nth round of interaction to better understand the user's nth round of interaction text information, making the reasoning results of the nth round of interaction text information more accurate, improving the accuracy and coherence of the response, and bringing an efficient communication experience to the user.

[0115] Understandably, optionally, after receiving the inference result corresponding to the nth round of interaction, SAAF uses text-to-speech (TTS) technology to convert the inference result into speech. Further, SAAF encapsulates the converted speech into real-time transport (RTP) data and sends it to MF, which then forwards it to the user.

[0116] Since DCAS may store historical call information for multiple users, the first piece of information may include the user's identifier, such as the user's trace identification. This user's identifier is used to determine the user's first historical call information from the historical call information of multiple users.

[0117] To enhance data security, DCAS can protect and process a user's historical call information before storing it. This protection process may include encryption, encoding, or obfuscation. Accordingly, when SAAF sends a first message to DCAS requesting access to the user's historical call information, the first message may include a first parameter. This first parameter is used to restore the user's historical call information, and the restoration process may include decryption, decoding, or deobfuscation.

[0118] It is understandable that if the protection process is encryption, the corresponding restoration process is decryption; if the protection process is encoding, the corresponding restoration process is decoding; and if the protection process is obfuscation, the corresponding restoration process is deobfuscation.

[0119] Since DCAS may store a large amount of historical call information from the user who initiated the first call, some of this historical call information may have little or no correlation with the text information of the nth round of interaction in that user's first call. Therefore, in order to save signaling overhead and obtain effective historical call information, there are two possible implementation methods.

[0120] In one possible implementation, the first information may include the range of the user's historical calls to be obtained, wherein the first historical call information is the user's historical call information within the range of historical calls.

[0121] For example, the range of historical calls is N historical calls prior to the first call, where N is a positive integer. SAAF requests to obtain the historical call information of the user's N historical calls prior to the first call. In other words, the first historical call information is the historical call information of the user within the N historical calls prior to the first call. For example, if N is 3, the first historical call information is the historical call information of the user's 3 historical calls. These 3 historical calls can be the 3 historical calls of the user that are closest to the first call among the user's multiple historical calls.

[0122] For example, the scope of the historical calls is the historical call information within a first time period. SAAF requests to obtain the user's historical call information within the first time period. That is, the first historical call information is the user's historical call information within the first time period. For example, if the first time period is 1 week, the first historical call information is the user's historical call information within the most recent week before the first call.

[0123] As shown in Figure 6A, during a call, the user inputs audio data for the nth round of interaction. The SAAF acquires this audio data and converts it into text data. Then, the SAAF requests the DCAS to retrieve information on the user's previous N historical calls or historical calls within a first time period. Based on this request, the DCAS returns this information to the SAAF. Further, the SAAF combines the text data from the nth round of interaction, the text data from the previous (n-1)th rounds of interaction, and the information from the previous N historical calls or historical calls within the first time period into a prompt word and sends it to the LLM. The LLM then uses this prompt word to infer the meaning of the text data from the nth round of interaction and outputs the inference result.

[0124] In another possible implementation, the first information includes the text data of the nth round of interaction of the user's first call, which is used to determine the first historical call information. In this implementation, SAAF sends the text information of the nth round of interaction of the user's first call to DCAS. Based on the text information of the nth round of interaction of the user's first call, DCAS obtains the first historical call information that is semantically closest to the text information of the nth round of interaction of the user's first call. This is beneficial for obtaining effective historical call information and saving signaling overhead.

[0125] DCAS can use semantic analysis techniques to determine the first historical call information that is semantically closest to the text information of the nth round of interaction. For example, semantic analysis techniques include natural language processing (NLP), vector matching, or best matching 25 (BM25) retrieval.

[0126] As shown in Figure 6B, during a call, the user inputs audio data for the nth round of interaction. The SAAF acquires this audio data and converts it into text data. Then, the SAAF requests the DCAS to retrieve the historical call information that semantically most closely matches the text data of the nth round of interaction. This request includes the text data of the nth round of interaction. Based on this request, the DCAS returns the historical call information that semantically most closely matches the text data of the nth round of interaction to the SAAF. Further, the SAAF combines the text data of the nth round of interaction, the text data of the previous n-1 rounds of interaction, and the historical call information that semantically most closely matches the text data of the nth round of interaction into a prompt word and sends it to the LLM. The LLM then uses this prompt word to infer the text data of the nth round of interaction and outputs the inference result.

[0127] It is understandable that the two implementation methods described above can be combined to determine the first historical call information. For example, the first historical call information can be the historical call information of N historical calls within a first duration. Another example is that the first historical call information can be the historical call information among the N historical calls within the first duration whose text data semantics are closest to the nth round of interaction. Yet another example is that the first historical call information can be the historical call information among the N historical calls prior to the first call whose text data semantics are closest to the nth round of interaction.

[0128] In another possible implementation, the first historical call information includes M historical contexts (or text information from M rounds of historical interactions) prior to the first call. These M historical contexts are determined from one or more historical calls prior to the first call. For example, if the first call is the user's Tth call and M = 10, and the user's (T-1)th call includes 12 contexts, then the first historical call information includes the 10 contexts closest to the Tth call from the 12 contexts of the user's (T-1)th call; if the user's (T-1)th call includes 6 contexts and the user's (T-2)th call includes 4 contexts, then the first historical call information includes the 6 contexts from the user's (T-1)th call and the 4 contexts from the user's (T-2)th call.

[0129] As an example, a user's interaction (or a question-and-answer session) can be viewed as a context, for example, {User: What is the price of XX car? Call Assistant: The current price of this car is XXX} This is a context.

[0130] Understandably, prior to S501, multiple network elements in the network architecture shown in Figure 3 interacted to complete the service setup and call establishment processes. For example, when a user initiates a new call using a terminal device, the terminal device interacts with DCAS to complete the process of subscribing to or setting up a call assistant service. When a user has a need, the user can use the terminal device to call the call assistant number. In response to the user's call, the VoLTE AS can interact with the DCSF to complete basic call negotiation, including call event notification, control, and user identification. Further, the VoLTE AS instructs the MF to create a media resource and sends the user's identification to the MF. After the media resource is created, the SAAF initiates the call assistant. During the call phase, the MF receives the user's audio data from the nth round of interaction through the DCSF. Further, the DCSF sends a request to the MF to copy the user's audio data from the nth round of interaction, which also instructs the copied audio data to be sent to the SAAF. Accordingly, based on this request, the MF conducts an audio data copy negotiation process with the SAAF, such as negotiating the port, uniform resource locator (URL), format (encoding method), and other information of the audio data. Furthermore, MF sends the user's nth round of interaction audio data to SAAF based on the negotiated content. SAAF can then convert the user's nth round of interaction audio data into text data that LLM can recognize.

[0131] Optionally, method 500 further includes S506: SAAF sends a fifth message to DCAS, the fifth message being used to request the storage of the text data of the user's first call. Accordingly, DCAS receives the fifth message and stores the text data of the user's first call based on the fifth message. The fifth message includes the text data of the user's first call. Optionally, method 500 further includes S507: DCAS sends a sixth message to SAAF, the sixth message being used to indicate that the text data of the user's first call has been stored.

[0132] After SAAF initiates call assistance, SAAF can request a memory segment for the first call to store the text data for each round of the first call. After the first call ends, SAAF can request DCAS to store the text data of the first call and clear the memory allocated by SAAF for the first call, releasing the memory resources occupied by the text data of the first call.

[0133] After the first call ends, the first call becomes a historical call for the user's subsequent calls. The text data of the first call can be used as historical call information for the user's subsequent calls. The SAAF can then obtain the text data of the first call to help the LLM better understand the text data of the user's subsequent calls and provide the user with more accurate and coherent responses.

[0134] In the subsequent calls of the user, in order for SAAF to efficiently obtain the text data of the user's first call, the fifth information may include the user's identifier. In this way, DCAS can associate the user with the text data of the first call. In the subsequent calls of the user, DCAS can determine the text data information of the first call associated with the user based on the user's identifier.

[0135] In one possible implementation, DCAS stores the user's first call text data in key-value pairs, using the user's identifier as the key and the text data of the first call as the value.

[0136] To enhance data security, the fifth message may include a second parameter used to protect the text data of the first call. Optionally, this protection process may include encryption, encoding, or obfuscation. After receiving the fifth message, DCAS can protect the text data of the first call based on the second parameter before storing it. Furthermore, during subsequent calls, SAAF can send a restoration parameter to DCAS to restore the text data of the first call. This restoration process may include decryption, decoding, or deobfuscation. It is understood that the technical means employed in the protection / restoration processes described in this application are possible examples, and this application does not limit the technical means adopted for the protection / restoration processes.

[0137] The second parameter can be an encryption parameter, an encoding parameter, or an obfuscation parameter.

[0138] Encryption makes text data difficult to crack even if it is leaked during storage or transmission. For example, DCAS can use symmetric or asymmetric encryption algorithms to encrypt the text data of the first call. Encoding converts text data into a specific encoding format, increasing the complexity and security of the text data. Obfuscation transforms text data in a specific way, making it difficult to directly recognize and understand.

[0139] Understandably, if the second parameter is an encryption parameter, DCAS encrypts the text data of the first call based on this encryption parameter when storing it. Correspondingly, when SAAF requests to retrieve the text data of the first call, it needs to send a decryption parameter corresponding to the encryption parameter to DCAS, and DCAS decrypts the text data of the first call based on this decryption parameter. If the second parameter is an encoding parameter, DCAS encodes the text data of the first call based on this encoding parameter when storing it. Correspondingly, when SAAF requests to retrieve the text data of the first call, it needs to send a decoding parameter corresponding to the encoding parameter to DCAS, and DCAS decodes the text data of the first call based on this decoding parameter. If the second parameter is an obfuscation parameter, DCAS obfuscates the text data of the first call based on this obfuscation parameter when storing it. Correspondingly, when SAAF requests to retrieve the text data of the first call, it needs to send a deobfuscation parameter corresponding to the obfuscation parameter to DCAS, and DCAS deobfuscates the text data of the first call based on this deobfuscation parameter.

[0140] It is understood that the various numerical designations used in the embodiments of this application are merely for descriptive convenience and are not intended to limit the scope of the embodiments of this application. The order of the process numbers does not imply the order of execution; the execution order of each process should be determined by its function and internal logic.

[0141] In the embodiments provided above, the methods provided by the embodiments of this application are described using SAAF, DCAS, and LLM as examples. In this application, each embodiment can be implemented independently or in combination based on certain inherent relationships; in each embodiment, different implementation methods can be implemented in combination or independently. To achieve the functions in the methods provided by the embodiments of this application above, the steps executed by SAAF, DCAS, or LLM can be implemented by SAAF, DCAS, or LLM itself, or by different functional entities constituting SAAF, DCAS, or LLM. To achieve the functions in the methods provided by the embodiments of this application above, SAAF, DCAS, or LLM can include hardware structures and / or software modules, implementing the above functions in the form of hardware structures, software modules, or hardware structures plus software modules. Whether a particular function is executed in the form of hardware structures, software modules, or hardware structures plus software modules depends on the specific application and design constraints of the technical solution.

[0142] The communication method according to the embodiments of this application has been described in detail above with reference to FIG5. The communication device according to the embodiments of this application will be described in detail below with reference to FIG7 and FIG8.

[0143] As shown in Figure 7, the communication device 700 includes a processing module 710 and a transceiver module 720. The transceiver module 720 can also be referred to as a communication interface or a communication module.

[0144] The device 700 can be used to perform the actions performed by SAAF or DCAS in the above method embodiments. Alternatively, the device 700 is a component (e.g., a chip) configured in SAAF or DCAS. The processing module 710 is used to perform processing-related operations of SAAF or DCAS in the above method embodiments. The transceiver module 720 is used to perform receiving and transmitting-related operations of SAAF or DCAS in the above method embodiments.

[0145] Optionally, the transceiver module 720 may include a sending module and a receiving module. The sending module is used to perform the sending operation in the above method embodiments. The receiving module is used to perform the receiving operation in the above method embodiments.

[0146] It should be noted that device 700 may include a transmitting module but not a receiving module. Alternatively, device 700 may include a receiving module but not a transmitting module. Specifically, it depends on whether the above-described scheme executed by device 700 includes both transmitting and receiving actions.

[0147] Optionally, the device 700 is used to perform the actions performed by SAAF or DCAS in the embodiment shown in FIG5 above. For details, please refer to the relevant description in the embodiment shown in FIG5 above, which will not be repeated here.

[0148] Optionally, the device 700 may further include a storage module, which can be used to store data and / or to store computer programs or instructions. The processing module 710 can read the computer programs / instructions and / or data in the storage module so that the device 700 can implement the above-described method embodiments.

[0149] When device 700 is used to implement the SAAF function in the method embodiment shown in FIG5, processing module 710 is used to: acquire text data of the nth round of interaction of the user's first call and text data of the n-1 rounds of interaction before the nth round of interaction; transceiver module 720 is used to: send first information to DCAS, the first information being used to request the acquisition of the user's historical call information; receive second information from DCAS, the second information including the user's first historical call information; send third information to LLM, the third information including text data of the nth round of interaction, text data of the n-1 rounds of interaction and the first historical call information; and receive fourth information from LLM, the fourth information including the inference result corresponding to the nth round of interaction.

[0150] Optionally, the first information includes the range of the user's historical calls to be obtained, and the first historical call information is the user's historical call information within the range of historical calls.

[0151] Optionally, the first information includes text data from the nth round of interaction, which is used to determine the first historical call information.

[0152] Optionally, the first information includes the user's identifier, which is used to determine the first historical call information from the historical call information of multiple users.

[0153] Optionally, the first information includes a first parameter, which is used to restore the user's historical call information.

[0154] Optionally, the restoration process includes decryption, decoding, or deobfuscation.

[0155] Optionally, the transceiver module 720 is configured to: send a fifth message to the DCAS, the fifth message being used to request the storage of the text data of the user's first call; and receive a sixth message from the DCAS, the sixth message being used to indicate that the text data of the first call has been stored.

[0156] Optionally, the fifth piece of information includes the user's identifier.

[0157] Optionally, the fifth information includes a second parameter, which is used to protect the text data of the first call.

[0158] Optionally, protection processing may include encryption, encoding, or obfuscation.

[0159] When the device 700 is used to implement the DCAS function in the method embodiment shown in FIG5, the transceiver module 720 is used to: receive first information from SAAF, the first information being used to request the acquisition of the user's historical call information; and send second information to SAAF, the second information including the user's first historical call information.

[0160] Optionally, the first information includes the range of the user's historical calls to be obtained, and the first historical call information is the user's historical call information within the range of historical calls.

[0161] Optionally, the first information includes text data of the nth round of interaction of the user's first call, which is used to determine the first historical call information.

[0162] Optionally, the first information includes the user's identifier, which is used to determine the first historical call information from the historical call information of multiple users.

[0163] Optionally, the first information includes a first parameter, which is used to restore the user's historical call information.

[0164] Optionally, the restoration process includes decryption, decoding, or deobfuscation.

[0165] Optionally, the transceiver module 720 is configured to: receive a fifth message from the SAAF, the fifth message being used to request the storage of the text data of the user's first call; the processing module 710 is configured to: store the text data of the first call based on the fifth message; the transceiver module 720 is further configured to: send a sixth message to the SAAF, the sixth message being used to indicate that the text data of the first call has been stored.

[0166] Optionally, the fifth piece of information includes the user's identifier. The processing module 710 is configured to: store the text data of the first call in key-value pairs, using the user's identifier as the key and the text data of the first call as the value.

[0167] Optionally, the fifth information includes a second parameter, which is used to protect the text data of the first call. The processing module 710 is used to: store the text data of the first call after protecting it based on the second parameter.

[0168] Optionally, protection processing may include encryption, encoding, or obfuscation.

[0169] It is understood that the module division in the above-described device is merely a logical functional division. Each function can correspond to a functional module, or two or more functions can be integrated into one functional module. In actual implementation, all or some modules can be integrated into a single physical entity, or they can be distributed across different physical entities. Furthermore, the aforementioned functional modules can be implemented in hardware, software, or a combination of both. Whether a function is executed in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0170] For a more detailed description of each step, please refer to the relevant descriptions in the method embodiments above, which will not be repeated here.

[0171] Figure 8 is a schematic block diagram of another communication device 800 provided in an embodiment of this application. As shown in Figure 8, the device 800 includes one or more processors 810 and an interface circuit 820. The one or more processors 810 and the interface circuit 820 are coupled to each other. It is understood that the interface circuit 820 can be a transceiver or an input / output interface. Optionally, the device 800 may also include a memory 830 for storing instructions executed by the processor 810, or for storing input data required by the processor 810 to execute instructions, or for storing data generated after the processor 810 executes instructions. Sometimes, the interface circuit 820 can also be understood as part of the one or more processors 810, in which case the device 800 includes the one or more processors 810.

[0172] Alternatively, in one design, processor 810 may include a computer program (also referred to as code or instructions) that can be run on processor 810 to cause device 800 to perform the methods executed by SAAF or DCAS in the above method embodiments. In yet another possible design, device 800 includes circuitry (not shown in FIG8) for implementing the functionality of SAAF or DCAS in the above method embodiments.

[0173] The one or more processors 810 and memory 830 can be configured separately or integrated, and this application does not limit this.

[0174] When device 800 is used to implement the method shown in FIG. 5, the one or more processors 810 are used to implement the functions of the processing module 710, and the interface circuit 820 is used to implement the functions of the transceiver module 720. Whether the communication interface 820 is used for sending or receiving depends on whether the device 800 is used to perform a sending or receiving action in the scheme it executes.

[0175] When the aforementioned device 800 is a chip applied to SAAF, the SAAF chip implements the functions of SAAF in the above method embodiments. The SAAF chip receives information from DCAS, which can be understood as the information being first received by other modules (such as RF modules or antennas) in the SAAF, and then sent to the SAAF chip by these modules. The SAAF chip sends information to DCAS, which can be understood as the information being first sent to other modules (such as RF modules or antennas) in the SAAF, and then sent to DCAS by these modules.

[0176] When the aforementioned device 800 is a chip applied to DCAS, the DCAS chip implements the functions of DCAS in the above method embodiments. The DCAS chip receives information from SAAF, which can be understood as the information being first received by other modules (such as RF modules or antennas) in the DCAS, and then sent to the DCAS chip by these modules. The DCAS chip sends information to SAAF, which can be understood as the information being first sent to other modules (such as RF modules or antennas) in the DCAS, and then sent to SAAF by these modules.

[0177] This application also provides a computer-readable storage medium for storing a computer program that, when run on a computer, causes the computer to perform the methods described in the above embodiments. Alternatively, the computer program includes instructions for implementing the methods described in the above embodiments.

[0178] This application also provides a computer program product, including: a computer program or instructions that, when run on a computer, cause the computer to perform the methods described above.

[0179] This application also provides an apparatus, which can be a chip, including at least one processor for supporting the implementation of the methods in the above embodiments, such as receiving or processing data involved in the methods in the above embodiments.

[0180] This application also provides a communication system including a SAAF and a DCAS. The SAAF can execute the methods described in the above embodiments, and the DCAS can execute the methods described in the above embodiments. Optionally, the communication system further includes an LLM. Optionally, the communication system further includes one or more of the following: a terminal device, a DCSF, an MF, and a VoLTE AS.

[0181] It should be understood that, in the embodiments of this application, the processor can be a central processing unit, or it can be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or any conventional processor.

[0182] In implementation, each step of the above method can be completed by integrated logic circuits in the processor's hardware or by instructions in software. The steps of the method disclosed in the embodiments of this application can be directly manifested as execution by a hardware processor, or as a combination of hardware and software modules within the processor. The software modules can reside in random access memory, flash memory, read-only memory, programmable read-only memory, electrically erasable programmable memory, registers, or other mature storage media in the art. This storage medium is located in memory, and the processor executes the instructions in the memory, combining them with its hardware to complete the steps of the above method. To avoid repetition, detailed descriptions are omitted here.

[0183] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0184] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and modules described above can be referred to the corresponding processes in the foregoing method embodiments, and will not be repeated here.

[0185] In the several embodiments provided in this application, it should be understood that the disclosed systems, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple modules or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between apparatuses or modules may be electrical, mechanical, or other forms.

[0186] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical modules; that is, they may be located in one place or distributed across multiple network modules. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0187] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0188] If the aforementioned functions are implemented as software functional modules and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, real-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0189] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A communication method, characterized in that, include: Obtain the text data of the nth round of interaction of the user's first call and the text data of the n-1 rounds of interaction before the nth round of interaction; the first call refers to the current call between the user and the call assistant, the first call includes multiple rounds of interaction, and the nth round of interaction refers to the current round of interaction between the user and the call assistant; Send first information to the Data Channel Application Server (DCAS), the first information being used to request the acquisition of the user's historical call information; each historical call includes one or more rounds of interaction; the first information includes text data of the nth round of interaction, the text data of the nth round of interaction being used to determine the first historical call information; Receive second information from the DCAS, the second information including the user's first historical call information; the first historical call information is the historical call information whose semantics are closest to the text data of the nth round of interaction; Send third information to the Large Language Model (LLM), the third information including the text data of the nth round of interaction, the text data of the (n-1)th round of interaction, and the first historical call information; Receive fourth information from the LLM, the fourth information including the reasoning result corresponding to the nth round of interaction.

2. The method as described in claim 1, characterized in that, The first information includes the range of the user's historical calls to be obtained, and the first historical call information is the user's historical call information within the range of the historical calls.

3. The method as described in claim 1 or 2, characterized in that, The first historical call information is the historical call information whose text data semantics are closest to the nth round of interaction among N historical calls within a first time period, or the first historical call information is the historical call information whose text data semantics are closest to the nth round of interaction among N historical calls before the first call.

4. The method according to any one of claims 1 to 3, characterized in that, The first information includes the user's identifier, which is used to determine the first historical call information from the historical call information of multiple users.

5. The method according to any one of claims 1 to 4, characterized in that, The first information includes a first parameter, which is used to restore the user's historical call information.

6. The method as described in claim 5, characterized in that, The restoration process includes decryption, decoding, or deobfuscation.

7. The method according to any one of claims 1 to 6, characterized in that, The method further includes: Send a fifth message to the DCAS, the fifth message being used to request the storage of the text data of the user's first call; Receive a sixth message from the DCAS, the sixth message indicating that the text data of the first call has been stored.

8. The method as described in claim 7, characterized in that, The fifth piece of information includes the user's identifier.

9. The method as described in claim 7 or 8, characterized in that, The fifth piece of information includes a second parameter, which is used to protect the text data of the first call.

10. The method as described in claim 9, characterized in that, The protection process includes encryption, encoding, or obfuscation.

11. A communication method, characterized in that, include: Receive first information from the user's intelligent agent function SAAF, the first information being used to request the acquisition of the user's historical call information; each historical call includes one or more rounds of interaction; the first information includes text data of the nth round of interaction of the user's first call, the text data of the nth round of interaction being used to determine the first historical call information; Send a second message to the SAAF, the second message including the user's first historical call information, the first historical call information being the historical call information semantically closest to the text data of the nth round of interaction, so that the SAAF sends a third message to the Large Language Model (LLM), the third message including the text data of the nth round of interaction of the first call, the text data of the n-1th round of interaction of the first call, and the first historical call information, the first call including multiple rounds of interaction, the nth round of interaction referring to the current round of interaction between the user and the call assistant.

12. The method as described in claim 11, characterized in that, The first historical call information is the historical call information whose text data semantics are closest to the nth round of interaction among N historical calls within a first time period, or the first historical call information is the historical call information whose text data semantics are closest to the nth round of interaction among N historical calls before the first call.

13. The method as described in claim 11 or 12, characterized in that, The first information includes text data of the nth round of interaction of the user's first call, and the text data of the nth round of interaction is used to determine the first historical call information.

14. The method according to any one of claims 11 to 13, characterized in that, The first information includes the user's identifier, which is used to determine the first historical call information from the historical call information of multiple users.

15. The method according to any one of claims 11 to 14, characterized in that, The first information includes a first parameter, which is used to restore the user's historical call information.

16. The method as described in claim 15, characterized in that, The restoration process includes decryption, decoding, or deobfuscation.

17. The method according to any one of claims 11 to 16, characterized in that, The method further includes: Receive a fifth message from the SAAF, the fifth message being used to request the storage of the text data of the user's first call; Based on the fifth piece of information, store the text data of the first call; A sixth message is sent to the SAAF, the sixth message indicating that the text data of the first call has been stored.

18. The method as described in claim 17, characterized in that, The fifth piece of information includes the user's identifier; The step of storing the text data of the first call based on the fifth information includes: The text data of the first call is stored in key-value pairs, using the user's identifier as the key and the text data of the first call as the value.

19. The method as described in claim 17 or 18, characterized in that, The fifth piece of information includes a second parameter, which is used to protect the text data of the first call. The step of storing the text data of the first call based on the fifth information includes: The text data of the first call is protected and then stored based on the second parameter.

20. The method as described in claim 19, characterized in that, The protection process includes encryption, encoding, or obfuscation.

21. A communication device, characterized in that, It includes modules for implementing the method as described in any one of claims 1 to 10, or modules for implementing the method as described in any one of claims 11 to 20.

22. A communication device, characterized in that, The device includes a processor coupled to a memory for storing programs or instructions that, when executed by the processor, cause the method as claimed in any one of claims 1 to 10 to be performed, or cause the method as claimed in any one of claims 11 to 20 to be performed.

23. A computer-readable storage medium, characterized in that, Used to store computer programs that, when run on a computer, cause the method as described in any one of claims 1 to 10 to be performed, or cause the method as described in any one of claims 11 to 20 to be performed.

24. A computer program product, characterized in that, include: A computer program or instruction that, when executed, causes the method as claimed in any one of claims 1 to 10 to be performed, or causes the method as claimed in any one of claims 11 to 20 to be performed.

Images