Data processing method and system and dialogue simulation system

Abstract

The embodiment of the invention provides a data processing method and system and a dialogue simulation system. The method comprises the steps of obtaining historical dialogue data, historical decision information corresponding to the historical dialogue data and an initial decision set; determining a target decision set according to the historical decision information and the initial decision set, and generating target prompt data according to the target decision set and the reference prompt word; according to the historical dialogue data and the target prompt data, target decision information is determined from a target decision set, and the target decision information comprises target emotion information and target action information; generating a target output result according to the target emotion information and the target action information; in addition, for the set initial decision set, different target decision sets can be determined under the condition of corresponding to different historical dialogue data and historical decision information, then target emotion information and target action information are determined from the target decision sets, and emotion circulation and behavior response in the dialogue process are achieved.

Description

Technical Field

[0001] The embodiments of this specification relate to the field of computer technology, and in particular, to a data processing method. One or more embodiments of this specification also relate to a data processing system, a dialogue simulation system, a computing device, an electronic device, a computer-readable storage medium, and a computer program product to address technical defects in the prior art. Background Art

[0002] Currently, the customer service training field generally adopts the method of building simulated characters based on language models to conduct practical exercises for customer service. During the dialogue interaction, the simulated characters rely on the fixed system prompt word templates of the language model for dialogue output, lacking the ability to dynamically change emotions and make independent behavioral decisions, resulting in insufficient realism.

[0003] In practice, prompt word engineering can create a configurable simulated character. However, once configured, the system prompt words for that character are fixed and do not change as the conversation progresses. For example, an angry character will remain angry throughout the simulated consumer conversation, lacking any thought or intention in its communication with customer service. In this case, because the character cannot effectively simulate the emotional changes during the conversation, the conversation appears stilted and unnatural. Templated output fails to achieve a truly personalized and natural conversation experience. Summary of the Invention

[0004] In view of this, embodiments of this specification provide a data processing method. One or more embodiments of this specification also relate to a data processing system, a conversation simulation system, a computing device, an electronic device, a computer-readable storage medium, and a computer program product to address technical deficiencies in the prior art.

[0005] According to a first aspect of an embodiment of this specification, there is provided a data processing method, including: Obtaining historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; Determining a target decision set based on the historical decision information and the initial decision set, and generating target prompt data based on the target decision set and reference prompt words, wherein the target decision set is a subset of the initial decision set; Determining target decision information from the target decision set according to the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information; A target output result is generated according to the target emotion information and the target action information.

[0006] According to a second aspect of the embodiments of this specification, a data processing system is provided, including a first intelligent decision-making unit and a second intelligent decision-making unit, wherein: The first intelligent decision-making unit is configured to obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; determine a target decision set based on the historical decision information and the initial decision set, and generate target prompt data based on the target decision set and a reference prompt word, wherein the target decision set is a subset of the initial decision set; determine target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information; The second intelligent decision-making unit is used to generate a target output result based on the target emotion information and the target action information.

[0007] According to a third aspect of the embodiments of this specification, there is provided a conversation simulation system, the conversation simulation system comprising a client and a simulation server; The client is configured to send a dialog generation request to the simulation server in response to an input operation on the dialog interaction interface; The simulation server is configured to, based on the conversation generation request, obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set using a first intelligent decision-making unit; determine a target decision set based on the historical decision information and the initial decision set; generate target prompt data based on the target decision set and a reference prompt word; determine target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision set is a subset of the initial decision set, and the target decision information includes target emotion information and target action information; generate a target output result based on the target emotion information and the target action information using a second intelligent decision-making unit, and return the target output result to the client; The client is further configured to receive and display the target output result on the dialogue interaction interface.

[0008] According to a fourth aspect of the embodiments of this specification, a computing device is provided, including: a memory and a processor, wherein the memory and the processor are connected via a bus; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor, the steps of the above-mentioned data processing method are implemented.

[0009] According to a fifth aspect of the embodiments of this specification, a computer-readable storage medium is provided, which stores a computer program / instruction, and the computer program / instruction implements the steps of the above-mentioned data processing method when executed by a processor.

[0010] According to a sixth aspect of the embodiments of this specification, a computer program product is provided, comprising a computer program / instruction, which implements the steps of the above-mentioned data processing method when executed by a processor.

[0011] An embodiment of the present specification provides a data processing method, which, when historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set are obtained, determines a subset of the initial decision set (target decision set) based on the historical decision information and the initial decision set, that is, different target decision sets can be determined from the initial decision set based on different historical decision information, and then generates target prompt data based on the target decision set and reference prompt words, thereby achieving different target prompt data based on different historical decision information, and determining target decision information from the target decision set based on the historical conversation data and the target prompt data, the target decision information including target emotion information and target action information, that is, based on different historical decision information. Historical decision information can determine target emotion information and target action information from different target decision sets, and generate target output results based on the target emotion information and target action information. The target decision set is determined based on the historical decision information and the initial decision set, and the target decision information including the target emotion information and target action information is determined from the target decision set. It is achieved that for the initially set initial decision set, different target decision sets can be determined corresponding to different historical dialogue data and historical decision information, and then the target emotion information and target action information are determined from the target decision set, realizing the emotional flow and behavioral response of the dialogue process, so that users can obtain a personalized and natural and smooth dialogue experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a scenario diagram of a data processing method provided by an embodiment of this specification; Figure 2 is a flow chart of a data processing method provided by one embodiment of this specification; Figure 3a This is a schematic diagram of a processing process of a data processing method applied to a simulated conversation scenario provided by an embodiment of this specification; Figure 3b This is a schematic diagram of a simulation effect corresponding to a data processing method provided in one embodiment of this specification; Figure 4This is a schematic diagram of the structure of a data processing system provided by one embodiment of this specification; Figure 5 This is a structural block diagram of a computing device provided by one embodiment of this specification. DETAILED DESCRIPTION

[0013] The following description sets forth many specific details to facilitate a thorough understanding of this specification. However, this specification can be implemented in many other ways than those described herein, and those skilled in the art can make similar generalizations without violating the scope of this specification. Therefore, this specification is not limited to the specific implementations disclosed below.

[0014] The terms used in one or more embodiments of this specification are for the purpose of describing specific embodiments only and are not intended to limit one or more embodiments of this specification. The singular forms "a," "the," and "the" used in one or more embodiments of this specification and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used in one or more embodiments of this specification refers to and includes any or all possible combinations of one or more associated listed items.

[0015] It should be understood that although the terms first, second, etc. may be used to describe various information in one or more embodiments of this specification, such information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of one or more embodiments of this specification, the first may also be referred to as the second, and similarly, the second may also be referred to as the first. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining".

[0016] In addition, it should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in one or more embodiments of this specification are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with the relevant laws, regulations and standards of relevant countries and regions, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0017] First, the terms involved in one or more embodiments of this specification are explained.

[0018] Prompt Engineering: Prompt engineering refers to the process of designing and optimizing input prompts when interacting with a language model to achieve better output. Because language models may respond differently to different input prompts, prompt engineering aims to guide the model to produce the desired output by selecting appropriate words, structures, and contextual information. This process is critical in improving model performance and controllability, especially when generating specific types of content or performing specific tasks.

[0019] Agents are autonomous intelligent entities built using language models or other large-scale AI models, capable of performing tasks and making decisions within a given environment. These agents typically combine natural language processing, machine learning, and reinforcement learning technologies, enabling them to not only understand and generate language but also continuously improve their performance in specific tasks through learning and adaptation. Agents are applicable in a variety of fields, including virtual assistants and automated customer service. A Multi-Agent System (MAS) is a system composed of multiple agents that interact to achieve their own goals or a common goal. Each agent can be considered an autonomous computing entity with perception, decision-making, and execution capabilities, capable of acting in its environment.

[0020] System prompts: System prompts are inputs used in dialogue generation systems to guide model behavior and set the tone of the conversation. System prompts typically contain context, instructions, or rules to help the model determine how to respond to user input. The specific content of system prompts varies by application and implementation, but they are typically not directly displayed to users in the user interface.

[0021] Traditional artificial human or character solutions rely primarily on predefined, fixed system prompt templates to generate conversations. For example, in previous iterations, a configurable simulated customer service character was implemented through the Prompt project. However, once configured, the system prompts were fixed and did not change as the conversation progressed. For example, an angry simulated character would remain angry and argumentative, lacking any deliberate thought or intention in their interactions with the customer service representative. Conversational simulations actually simulate possible customer inquiries or situations by setting a series of conditions and branching paths.

[0022] While this template-driven approach can achieve training for customer service personnel to a certain extent, it has significant limitations. First, the lack of dynamics in emotions and behaviors is a major problem. Traditional solutions are generally unable to effectively simulate the emotional changes of users during conversations, resulting in conversations that appear stiff and unnatural. Second, the responses of these systems are often based on pre-set rules rather than real-time analysis of user input, lacking flexibility and adaptability. This makes it easy for the system to misjudge or fail to provide reasonable responses when faced with complex or unexpected user input. In addition, templated output cannot achieve a truly personalized and natural conversation experience, affecting the effectiveness of customer service simulation exercises.

[0023] Crucially, traditional simulation solutions have limited scalability and struggle to adapt to evolving user needs and technological developments. As user expectations rise and AI advances, the shortcomings of traditional solutions in terms of humanized interaction and realism are becoming increasingly apparent, necessitating new solutions to overcome these bottlenecks.

[0024] In order to solve the above technical problems, a data processing method is provided in this specification. This specification also involves a data processing system, a dialogue simulation system, a computing device, an electronic device, a computer-readable storage medium, and a computer program product, which are described in detail one by one in the following embodiments.

[0025] See also Figure 1 , Figure 1 A schematic diagram of a scenario of a data processing method provided by an embodiment of this specification is shown.

[0026] Specifically, the data processing method is applied to a system consisting of an end-side device 102 and a server 104. Specifically, the end-side device 102 is used to send a dialogue generation request to the server 104. For example, the user input carried in the dialogue generation request is "Do you need to return the goods now?"; in actual applications, the user can input content in the end-side device 102 through text or voice. If voice is used, the end-side device 102 will also include a corresponding voice processing part, such as voice analysis, voice-to-text, voice synthesis and other modules, which are used to convert the content input by the user through voice into text. This manual does not limit this.

[0027] A first intelligent decision-making unit and a second intelligent decision-making unit are deployed in the server 104. The conversation data and corresponding decision information of historical conversation rounds are obtained from the storage module of the server 104. The currently received user input and the conversation data of the historical conversation rounds are determined as historical conversation data, and the decision information corresponding to the conversation data in the historical conversation rounds is determined as historical decision information. The first intelligent decision-making unit is used to obtain the historical conversation data, historical decision information, and an initial decision set. A target decision set is determined based on the historical decision information and the initial decision set. Target prompt data is generated based on the target decision set and a reference prompt word. Target decision information is determined from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision set is a subset of the initial decision set, and the target decision information includes target emotion information and target action information. The second intelligent decision unit is used to generate a target output result based on the target emotion information and the target action information, and the target output result is returned to the end-side device 102 so that the end-side device 102 receives and displays the target output result.

[0028] The end-side device 102 may include a browser, an application (APP), or a web application such as an H5 (Hypertext Markup Language 5) application, a lightweight application (also known as a mini-program, a type of lightweight application), or a cloud application. The end-side device may be developed based on a software development kit (SDK) for the corresponding service provided by the server, such as a real-time communication (RTC) SDK. The end-side device may be deployed in an electronic device and may rely on the device or certain apps in the device to operate. The electronic device may have a display and support information browsing, such as a personal mobile terminal such as a mobile phone, tablet computer, or personal computer. Various other types of applications may also be configured in the electronic device, such as human-computer interaction applications, model training applications, text processing applications, web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0029] Server 104 can be understood as a server that provides various services, including physical servers and cloud servers. For example, these servers provide communication services to multiple clients, servers that support backend training for models used by clients, and servers that process data sent by clients. It should be noted that server 104 can be implemented as a distributed server cluster consisting of multiple servers or as a single server. Server 104 can also be a server in a distributed system or a server integrated with blockchain. Server 104 can also be a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms, or it can be an intelligent cloud computing server or intelligent cloud host equipped with artificial intelligence technology.

[0030] It is worth noting that the data processing method provided in the embodiments of this specification can be executed by the server 104. In other embodiments of this specification, the first intelligent decision-making unit and the second intelligent decision-making unit can be deployed in the terminal device 102, so that the terminal device 102 can also have similar functions as the server 104, thereby executing the data processing method provided in the embodiments of this specification; in other embodiments, the data processing method provided in the embodiments of this specification can also be jointly executed by the terminal device 102 and the server 104.

[0031] In the data processing method provided by the embodiment of this specification, the terminal device is used to send a data processing request carrying current conversation data to the server, and the server uses the first intelligent decision unit to obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set, and determines a subset of the initial decision set (target decision set) based on the historical decision information and the initial decision set, that is, different target decision sets can be determined from the initial decision set based on different historical decision information, and then target prompt data is generated based on the target decision set and reference prompt words, so as to achieve different target prompt data based on different historical decision information, and determine target decision information from the target decision set based on the historical conversation data and the target prompt data, and the target decision information includes target emotion information and target prompt data. Target action information, that is, according to different historical decision information, target emotion information and target action information can be determined from different target decision sets, and the second intelligent decision unit is used to generate a target output result according to the target emotion information and the target action information, and the target decision set is determined according to the historical decision information and the initial decision set, and the target decision information including the target emotion information and the target action information is determined from the target decision set, so that the initial decision set that is initially set can also determine different target decision sets corresponding to different historical dialogue data and historical decision information, and then determine the target emotion information and target action information from the target decision set, so as to realize the emotional flow and behavioral response of the dialogue process, so that the user can obtain a personalized and natural and smooth dialogue experience.

[0032] See also Figure 2 , Figure 2 A flow chart of a data processing method provided according to an embodiment of the present specification is shown, which specifically includes the following steps.

[0033] Step 202: Acquire historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set.

[0034] Historical conversation data can be understood as the conversation records accumulated by participants during a conversation. This data includes not only the output of one party in the current conversation turn but also the conversation context from previous conversation turns, which is used to understand the interaction context and maintain conversation coherence. For example, a conversation record includes user input (questions, commands, or feedback) and system responses; a complete conversation turn is composed of user input and system responses.

[0035] The historical decision information corresponding to historical conversation data can be understood as the decision information corresponding to the conversation record during the conversation process. This decision information represents the emotion and action information selected by the system when responding to user input. Therefore, if each historical conversation turn includes a system response, each historical conversation turn corresponds to a decision information. In the case of this data processing method applied to customer service simulation training scenarios, user input is the input content of the customer service side, and the system response is the output content of the system side (also called the simulated human, simulated character, or simulated customer side).

[0036] The initial decision set can be understood as an initially set set of decision information that the system can choose when making a response corresponding to user input. The initial decision set includes an initial emotion set and a behavior set, and the initial emotion set and behavior set include initial optional and comprehensive emotion information and behavior information.

[0037] Specifically, an initial optional and comprehensive initial emotion set and behavior set are determined, and when a new round of dialogue is started, the user input of the current dialogue round is integrated with the dialogue records of the historical dialogue rounds to determine them as historical dialogue data, and the decision information corresponding to the historical dialogue rounds is determined as historical decision information.

[0038] Step 204: Determine a target decision set based on the historical decision information and the initial decision set, and generate target prompt data based on the target decision set and reference prompt words, wherein the target decision set is a subset of the initial decision set.

[0039] The target decision set can be understood as a set of optional decision information in the current dialogue round, and the decision information in the target decision set is determined from the initial decision set.

[0040] Reference prompts are conversational context information for the conversation process, used to guide the tone and direction of the conversation. Reference prompts include, but are not limited to, role settings for the simulated person (used to define the identity, position, or virtual role of the conversation participant, such as the role setting of "you are an online shopper"), role characteristics (used to describe details such as the character's personality, tone, and knowledge scope, such as the role characteristic of "you are now raising the issues and demands you have encountered regarding product quality to the platform customer service"), and task objectives (used to determine the output goals, such as the task objective of "determining the current emotional attitude and deciding the next action"), etc., which are not limited here.

[0041] In specific implementation, when the initial decision set contains multiple decision information, the currently optional decision information is determined from the initial decision set in each current dialogue round to form a target decision set, so as to guide the simulation person (constructed by the language model, which can be understood as the system side mentioned above) to make reasonable decision information selection based on the target decision set; the simulation person can dynamically select the optional decision information for the current dialogue round from the initially set initial decision set, thereby realizing the dynamic emotional flow of the simulation person.

[0042] The initial decision set contains comprehensive decision information. In order to ensure a reasonable flow of emotions, it is necessary to refer to historical decision information to determine the optional decision information for the current dialogue round from the initial decision set and obtain the target decision set. That is, when the target decision set is determined based on historical decision information, it ensures that the optional decision information for the current dialogue round and the decision information selected in the historical dialogue rounds maintain a moderate consistency, so as to avoid large emotional fluctuations and insufficient realism caused by excessive differences in decisions, thereby improving the interactive realism.

[0043] In one or more embodiments of this specification, the reference prompt word is determined by the initial prompt data that is initially set. The initial prompt data is a system prompt initially set for the simulated person. It is the initial input information that the simulated person relies on when starting to perform the dialogue generation task, helping the simulated person to reasonably determine decision information in the initial dialogue round. The specific implementation method is as follows: Before generating target prompt data according to the target decision set and the reference prompt words, the method further includes: The reference prompt word is obtained according to initial prompt data, wherein the initial prompt data consists of the reference prompt word and the initial decision set.

[0044] Among them, the initial prompt data can be understood as the initial input information that the system or model relies on when it starts to perform a task. In the customer service simulation practice scenario, the initial prompt data can be understood as the initial input information that the simulated person relies on when it starts to perform a dialogue generation task.

[0045] Specifically, a configurable simulation human is realized through initial prompt data, which includes conversation background information and an optional initial decision set, so that the simulation human can achieve highly controllable simulated interaction.

[0046] The conversation background information (i.e., reference prompt words) can be determined through the initial prompt data. By combining the conversation background information with the target decision set, the current system prompt corresponding to the simulator (i.e., target prompt data) is formed, so that the simulator can reasonably generate the output content corresponding to the current conversation round through the current system prompt. When the decision information in the target decision set can change dynamically, the target prompt data corresponding to the simulator also changes dynamically, thereby realizing the dynamic emotional flow of the simulator during the conversation.

[0047] It should be noted that in the initial dialogue round, the historical dialogue data does not contain the dialogue records of the historical dialogue round, so there is no system response, and there is no decision information selected by the system when making a response. At this time, the initial decision set is determined as the target decision set, that is, in the initial dialogue round, the simulator can determine the decision information corresponding to the initial dialogue round from the comprehensive decision information.

[0048] In one or more embodiments of this specification, the historical decision information includes historical emotion information, and the initial decision set includes an initial emotion set and an action set. In practice, when implementing dynamic emotion flow, the target emotion set is determined from the initial emotion set based on the historical emotion information. Specific implementation methods are as follows: Determining a target decision set based on the historical decision information and the initial decision set includes: Determining a target emotion set based on the historical emotion information and the initial emotion set; The target decision set is determined according to the target emotion set and the action set.

[0049] Historical emotion information can be understood as the emotion information selected by the system from historical conversation data. The initial emotion set can be understood as the set of emotion information initially set and available for selection when the system responds to user input, encompassing comprehensive emotion information. The target emotion set can be understood as the set of emotion information available for selection in the current conversation turn, with the emotion information in the target emotion set being determined from the initial emotion set. The action set can be understood as the set of behavioral information available for selection when the system responds to user input.

[0050] For example, the emotional information in the initial emotion set includes "confrontation", "neutrality", "cooperation", "non-cooperation", etc., and the emotional information in the target emotion set includes "resistance", "questioning", etc.

[0051] Specifically, when the initial decision set includes an initial emotion set and an action set, determining the target decision set from the initial decision set is specifically implemented by determining the target emotion set from the initial emotion set, thereby combining the target emotion set and the behavior set to form the target decision set.

[0052] In order to realize the emotional flow of the simulated person, the emotional information that the simulated person can choose in each dialogue round must be dynamically changed during the dialogue process, and the optional behavioral information can be comprehensive behavioral information. Therefore, the target emotional set is determined based on the initial emotional set.

[0053] In fact, in the above embodiment, when determining the optional decision information for the current dialogue round, it is necessary to refer to the historical decision information of the historical dialogue data. Specifically, when determining the optional emotional information for the current dialogue round, it is necessary to refer to the historical emotional information of the historical dialogue data. For example, when the initial emotion set includes emotions such as "confrontation", "resistance", "questioning", "neutrality", "cooperation", and "non-cooperation", if the emotion output by the simulated human in the previous dialogue round is "resistance", then in the current dialogue round, it is unlikely that the simulated human will directly convert from the "resistance" emotion of the previous dialogue round to the "cooperation" emotion. Therefore, in order to ensure a reasonable human-like emotional flow, it is necessary to refer to the historical emotional information to determine the target emotion set corresponding to the current dialogue round from the initial emotion set.

[0054] The data processing method provided in the embodiments of this specification, when realizing the emotional flow of a simulated person, determines the currently optional emotional information from the initial emotion set based on the historical emotional information in the historical decision information, forms a dynamically changing target emotion set, and combines the target emotion set and the behavior set, so that the simulated person can reasonably select emotional information and behavioral information from the target emotion set and the behavior set.

[0055] In one or more embodiments of this specification, when determining a target emotion set from an initial emotion set based on historical emotion information, emotion information in the initial emotion set that is associated with the historical emotion information is used as candidate emotion information in the target emotion set. Specific implementations are as follows: Determining a target emotion set according to the historical emotion information and the initial emotion set includes: Acquire an emotion state machine, wherein the emotion state machine is determined according to emotion information in the initial emotion set; According to the historical emotion information, candidate emotion information is determined from the emotion state machine, and the target emotion set is generated according to the candidate emotion information.

[0056] Among them, the emotional state machine can be understood as a state machine composed of comprehensive emotional information in the initial emotion set. The emotional state machine is composed of emotional nodes and edges. One emotional node corresponds to one emotional information, and the edge represents the connection relationship between different emotional information. There is an association relationship between two emotional information connected by an edge, and the edge can have a direction to indicate the flow direction of the emotional information, that is, it flows from one emotional state to another in a specified direction.

[0057] An emotional state machine is constructed using the emotional information in the initial emotional set. For example, the initial emotional set includes emotions such as "opposition", "resistance", "questioning", "incomprehension", "cooperation", and "non-cooperation". In the emotional state machine composed of the initial emotional set, "opposition", "resistance", "questioning", "incomprehension", "cooperation", and "non-cooperation" can be respectively used as emotional nodes of the emotional state machine, and the "resistance" emotional node points to the "questioning" emotional node and the "incomprehension" emotional node respectively. When the historical emotional information is "resistance", the determined candidate emotional information can be "questioning" and "incomprehension", so that the target emotional set is composed of "questioning" and "incomprehension", and the target emotional information will be determined from the candidate emotional information of the target emotional set later.

[0058] The data processing method provided in the embodiments of this specification can construct an emotional state machine through the emotional information in the initial emotional set. When the emotional state machine contains the association relationship between emotional information, candidate emotional information that is related to the historical emotional information and has coherence can be determined from the emotional state machine based on historical emotional information.

[0059] Step 206: Determine target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information.

[0060] Among them, the target decision information can be understood as the decision information selected by the system side in the current dialogue round, the target emotion information can be understood as the emotion information selected by the system side in the current dialogue round, and the target action information can be understood as the behavior information selected by the system side in the current dialogue round.

[0061] Specifically, when the target prompt data includes conversation context information, the target decision information is accurately determined from the target decision set in the target prompt data by referring to the historical conversation data and the conversation context information as reference data.

[0062] In one or more embodiments of this specification, the data processing method provided in the embodiments of this specification is applied to a data processing system (i.e., the simulated person in the above-mentioned embodiment), and the data processing system is a multi-agent system. Specifically, the data processing system is composed of a first intelligent decision-making unit and a second intelligent decision-making unit. The first intelligent decision-making unit is used to make emotional behavior decisions, i.e., to determine target decision information. The data processing system includes the first intelligent decision-making unit; Determining target decision information from the target decision set based on the historical conversation data and the target prompt data includes: In the first intelligent decision unit, target decision information is determined from the target decision set based on the historical dialogue data and the target prompt data.

[0063] Specifically, the first intelligent decision-making unit is an emotional behavior decision-making agent. An agent can be understood as an entity or program that can perceive the environment, understand input information, and make decisions and actions according to preset goals and rules. The agent mentioned in one or more embodiments of this specification can be understood as an AI agent (AI Agent), that is, an autonomous agent built using a language model or other large-scale artificial intelligence model. The target prompt data (i.e., system prompt) of the first intelligent decision-making unit includes conversation background information and a target decision set. The first intelligent decision-making unit selects target decision information from the target decision set by referring to historical conversation data and conversation background information.

[0064] In fact, when the target decision set includes a target emotion set and a behavior set, the target decision information includes target emotion information and target behavior information. The target emotion information is determined from the target emotion set, and the target behavior information is determined from the behavior set. The specific implementation is as follows: Determining target decision information from the target decision set based on the historical conversation data and the target prompt data includes: According to the historical dialogue data and the target prompt data, the target emotion information is determined from the target emotion set, and the target action information is determined from the action set.

[0065] Specifically, the task goal can be determined through the conversation background information in the target prompt data (selecting emotional information and behavioral information that are consistent with the current conversation). Based on historical conversation data, the user input intention can be analyzed, the conversation evolution path can be analyzed, etc. The range of optional emotional information and optional behavioral information can be limited based on the target emotion set and behavior set in the target prompt data.

[0066] When the target decision set includes the target emotion set and behavior information, the target emotion information is determined from the target emotion set, and the target behavior information is determined from the behavior information, that is, the emotion decision and the behavior decision are accurately made.

[0067] In fact, the system prompt template of the first intelligent decision-making unit can be "You are an online shopping consumer, you have encountered the problem of {problem type}, determine your current attitude from {attitude options} based on the chat record", where the slot part of the system prompt template is {problem type} (used to dynamically fill in the type of problem encountered, such as the problem type may include product quality, order progress, etc.), {attitude options} (used to dynamically fill in optional attitudes, such as attitudes may include questioning, neutrality, cooperation, non-cooperation, etc.).

[0068] In an embodiment of the present specification, the system prompt slot of the first intelligent decision-making unit may include dialogue background parameters, an optional set of emotions, and an optional set of behaviors. By obtaining the initial prompt data, the dialogue background information corresponding to the dialogue background parameters can be determined, thereby configuring and constructing a simulated person, and determining the behavior set in the initial prompt data as the optional set of behaviors; and when receiving historical dialogue data and historical decision information, the historical dialogue data and historical decision information are determined as user prompt data, and based on the historical decision information, the target emotion set can be determined from the initial emotion set of the initial prompt data, and the target emotion set is determined as the optional set of emotions, thereby generating the current system prompt (i.e., target prompt data) of the first intelligent decision-making unit; and the target emotion information and target behavior information are determined by combining the current system prompt and user prompt data dynamically obtained at runtime.

[0069] The data processing method provided in the embodiments of this specification uses historical conversation data and target prompt data to determine target emotion information from a target emotion set and target action information from an action set, thereby achieving human-like emotion flow and behavioral response.

[0070] In one or more embodiments of this specification, both emotion information and behavior information include behavior representations. When target emotion information and target behavior information are determined, a target emotion identifier corresponding to the target emotion information and a target behavior identifier corresponding to the target behavior information are output. Specific implementations are as follows: The target emotion information includes a target emotion identifier, and the target action information includes a target action identifier; After determining target decision information from the target decision set based on the historical conversation data and the target prompt data, the method further includes: The target emotion identifier in the target emotion information is determined and output, and the target action identifier in the target action information is determined and output.

[0071] It should be noted that the "confrontation", "resistance", "questioning" and so on mentioned in the above embodiments are all simplified expressions of emotional information, which are the emotion definition information in the emotional information. In fact, the complete emotional information is represented by the following format: ("emotion identification", "emotion definition information", "emotion description information"). For example, the first emotional information RESISTANCE is ("1", "confrontation", "You actively resist and refuse to accept the products or services provided. You have a strong tendency to complain or complain"), and the second emotional information DOUBT is ("2", "questioning", "You question the customer service's proposal, raise objections, and clearly refuse").

[0072] Similarly, behavior information is represented in the following format: ("behavior identifier", "behavior definition information", "behavior description information"). For example, the first behavior information EXPRESS_PROBLEM is ("A", "state the problem", "state the problem you encountered"), and the second behavior information DENY_PROBLEM is ("C", "deny the customer service's question", "deny the problem identified by the customer service").

[0073] When the target emotion information and the target behavior information are determined, the target emotion identifier in the target emotion information and the target behavior identifier in the target behavior information are determined. By outputting the target emotion identifier and the target behavior identifier, the output token of the intelligent agent can be greatly reduced to single digits, thereby reducing the response delay.

[0074] The data processing method provided in the embodiments of this specification, when the target emotion identifier and the target behavior identifier can be represented by a simple serial number, outputs the target emotion identifier and the target behavior identifier by designing a lightweight serialized decision output mechanism. Compared with the reasoning model that outputs thinking reasoning logic, it can greatly reduce the output token to single digits, control the response delay within 3 seconds, and improve data processing efficiency.

[0075] Step 208: Generate a target output result according to the target emotion information and the target action information.

[0076] Specifically, when the target emotion information and behavior information are determined, the target emotion information and target behavior information are used as input information of the dialogue generation model, thereby generating a target output result with the emotional color corresponding to the target emotion information and consistent with the target behavior information.

[0077] For example, the target behavior information is expressing a problem, and the target emotion information is anxiety. The target output result generated at this time can be "My express [order number 123456] has not arrived yet! What should I do if I need it tomorrow?"

[0078] In one or more embodiments of this specification, the data processing system includes a second intelligent decision-making unit; this second intelligent decision-making unit is a dialogue generation agent based on a dialogue generation model, and is used to generate a target output result based on the determined target emotion information and target behavior information. The target output result is the response data generated by the system side to the user input. The specific implementation is as follows: Generating a target output result according to the target emotion information and the target action information includes: In the second intelligent decision-making unit, a target output result is generated according to the target emotion information and the target action information.

[0079] Specifically, the target emotion information and the target behavior information are input into the second intelligent decision-making unit as system prompts of the second intelligent decision-making unit, so that the second intelligent decision-making unit generates a target output result based on the input target emotion information and target behavior information.

[0080] In fact, when the first intelligent decision-making unit outputs the target emotion identifier and the target behavior identifier, the target emotion description information corresponding to the target emotion identifier and the target behavior description information corresponding to the target behavior identifier are determined, and the target emotion description information and the target behavior description information are input into the second intelligent decision-making unit. The specific implementation method is as follows: The target emotion information includes target emotion description information, and the target action information includes target action description information; Before generating a target output result according to the target emotion information and the target action information, the method further includes: Determine the target emotion description information corresponding to the target emotion identifier, and determine the target action description information corresponding to the target action identifier; Generating a target output result according to the target emotion information and the target action information includes: The target output result is generated according to the target emotion description information and the target action description information.

[0081] Continuing with the above example, for example, the output of the first intelligent decision-making unit is (2, C), that is, the target emotion identifier is 2, and the target behavior identifier is C. At this time, according to the target emotion information, the target emotion description information corresponding to the target emotion identifier 2 is determined to be "You question the customer service's proposal, raise objections, and clearly refuse", and according to the target behavior information, the target behavior description information corresponding to the target behavior identifier C is determined to be "Neging the problem positioned by the customer service."

[0082] When specific emotion description information and behavior description information are input into the second intelligent decision-making unit, a target output result is generated. The target output result will be returned to the user's client as a system response to display the target output result on the client to complete the current dialogue round.

[0083] In the data processing method provided in the embodiments of this specification, a serialized decision output mechanism is designed to improve the response speed, so that the first intelligent decision unit outputs the target emotion identifier and the target behavior identifier. When the dialogue content is generated in the second intelligent decision unit, it is necessary to use the complete emotion description information and the behavior description information for accurate generation. Through the mapping relationship between the emotion / behavior identifier and the emotion / behavior description information, the realism of the dialogue simulation is improved on the basis of improving the data processing efficiency.

[0084] In one or more embodiments of this specification, the data processing method in the above embodiment is a specific implementation of generating a system response in the current conversation round. When a new conversation round is started, the previous conversation round is determined as a historical conversation round, so that the historical conversation data of the new current conversation round includes the user input and system response of the previous conversation round, and the historical decision information corresponding to the new current conversation round includes the target decision information determined in the previous conversation round. The specific implementation method is as follows: After generating a target output result according to the target emotion information and the target action information, the method further includes: In the case of obtaining target reply data corresponding to the target output result, updating and marking the historical conversation data, the target output result, and the target reply data as the historical conversation data; Update and mark the target decision information and the historical decision information as the historical decision information; Continue to execute the steps of obtaining the historical conversation data, the historical decision information corresponding to the historical conversation data, and the initial decision set.

[0085] In fact, when the above-mentioned target output result is returned to the client, the user can reply to the target output result, that is, the user can input target reply data to start a new round of dialogue, and the user input of the new round of dialogue is the target reply data of the above-mentioned target output result.

[0086] As the conversation progresses, historical conversation data and historical decision information are dynamically updated with each conversation turn. When target response data from the user is obtained, it indicates that the user has initiated a new conversation turn. Therefore, the target output from the previous conversation turn is also included in the historical conversation data, generating the historical conversation data corresponding to the current conversation turn. Specifically, the historical conversation data corresponding to the current conversation turn includes the user input from the previous conversation turn, the system response to the user input, and the user input from the current conversation turn (i.e., the target response data). The target decision information determined in the previous conversation turn is also included in the historical decision information, generating the historical decision information corresponding to the current conversation turn.

[0087] Thus, by continuing to execute the above steps of obtaining historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; determining a target decision set based on the historical decision information and the initial decision set, and generating target prompt data based on the target decision set and reference prompt words, wherein the target decision set is a subset of the initial decision set; determining target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information; generating a target output result based on the target emotion information and the target action information, the target decision information and target output result of the current conversation round are determined; and stopping until no user input is received.

[0088] The data processing method provided in the embodiments of this specification, when obtaining the target output result corresponding to the target response data, obtains historical dialogue data and historical decision information by updating, and realizes dialogue generation processing for the current dialogue round to ensure that the dialogue process can continue.

[0089] The data processing method provided in the embodiments of this specification constructs a two-layer intelligent agent collaborative architecture with emotional flow and behavioral decision-making by the first intelligent decision-making unit and the second intelligent decision-making unit, realizing the anthropomorphic interaction of artificial intelligence simulation characters, which can be extended to multiple simulation human application scenarios such as virtual assistants and virtual characters. It has generalization capabilities and, by designing a lightweight serialized decision output mechanism, greatly reduces the output token to single digits compared to the reasoning model that outputs thinking reasoning logic, and controls the response delay within 3 seconds, providing a new paradigm for building highly realistic, low-cost artificial intelligence simulation characters.

[0090] The following combined Figure 3a , taking the application of the data processing method provided in this specification in a simulation dialogue scenario as an example, the data processing method is further explained. Figure 3aA schematic diagram of the processing process of a data processing method applied to a simulated dialogue scenario provided by an embodiment of this specification is shown.

[0091] In response to the customer service input operation of the client chat interface, the chat messages between the dialogue participants are determined, the chat messages are stored in the storage module, the historical dialogue data is determined in the storage module, and the historical dialogue data is input into the emotional behavior decision-making agent (i.e., the first intelligent decision-making unit in the above embodiment).

[0092] The input parameters of the emotional behavior decision-making agent include system prompts and user prompt data. The slots of the system prompts contain conversation background parameters, an optional set of emotions corresponding to the current emotion, and an optional set of actions. The user prompt data contains historical conversation data and historical decision information (including historical emotion information).

[0093] Specifically, when it is determined that the initial prompt data corresponding to the emotional behavior decision-making agent contains dialogue background information, an initial emotion set, and a behavior set, the dialogue background information is filled into the dialogue background parameters, the behavior set is determined as an optional action set, and based on the historical emotion information in the user prompt data, the current optional emotion information is determined from the initial emotion set to form a target emotion set, and the target emotion set is determined as the optional emotion set corresponding to the current emotion.

[0094] In fact, when the emotion state machine is constructed based on the emotion information in the initial emotion set, the target emotion set is determined from the initial emotion set based on the emotion state machine and the historical emotion information; Figure 3a The emotional state machine in can determine the association between different emotional information. Therefore, based on the emotional state machine and the emotional information corresponding to the previous dialogue round, the emotional information associated with the emotional information corresponding to the previous dialogue round can be determined as the candidate emotional information in the target emotion set.

[0095] The Emotional Behavior Decision-Making Agent uses historical conversation data and context to determine target emotion information and target behavior information from the target emotion set and behavior set, respectively. It then outputs the target emotion identifier (i.e., emotion sequence number) from the target emotion information and the target behavior identifier (i.e., action sequence number) from the target behavior information. For example, the output of the Emotional Behavior Decision-Making Agent is 3, B. It should be noted that if the Emotional Behavior Decision-Making Agent requires fast output and low token output cost, the language model corresponding to the Emotional Behavior Decision-Making Agent can be larger and more intelligent than the language model corresponding to the Dialogue Generation Agent.

[0096] The emotion sequence number and the action sequence number are mapped into corresponding text definition descriptions to obtain the emotion definition description corresponding to the emotion sequence number (i.e. the emotion description information in the above embodiment) and the action definition description corresponding to the action sequence number (i.e. the behavior description information in the above embodiment).

[0097] The emotion definition description and the action definition description are input into the dialogue generation intelligent agent as system prompts. In fact, the dialogue generation intelligent agent can also obtain historical dialogue data from the storage module, and thus generate simulated dialogue content (i.e., target output result) that is contextually associated with the historical dialogue data, has the emotional color corresponding to the emotion definition description, and conforms to the action definition description based on the historical dialogue data, the emotion definition description, and the action definition description. The simulated dialogue content is returned to the client and displayed on the chat interface.

[0098] See also Figure 3b , Figure 3b A schematic diagram of a simulation effect corresponding to a data processing method provided in an embodiment of this specification is shown.

[0099] That is, in the existing method, when the emotion of the simulated person is initially set to a negative emotion through the Prompt project, the emotion output by the simulated person is a negative emotion and will not change dynamically; while in the embodiment of this specification, when the emotion of the simulated person is initially set to a negative emotion through the Prompt project, the emotion selected by the simulated person through the dynamic target emotion set can be dynamically changed. Furthermore, when the dialogue content is generated based on the emotion selected by the decision, the emotion displayed in the dialogue content also has a certain dynamic change space compared to the emotion selected by the decision. Therefore, for the simulated person whose emotion is initially set to a negative emotion, the emotion flow also becomes a dynamic range.

[0100] The data processing method provided in the embodiments of this specification realizes human-like emotional flow and behavioral response through a two-layer intelligent agent collaborative architecture. By realizing dynamic emotional changes and autonomous behavioral decision-making capabilities, the realism of the simulated person is improved and more natural dialogue content is output, thereby providing customer service with more effective simulation practice services and improving the user's dialogue experience.

[0101] The embodiment of this specification also provides a conversation simulation system, which includes a client and a simulation server; The client is configured to send a dialog generation request to the simulation server in response to an input operation on the dialog interaction interface; The simulation server is configured to, based on the conversation generation request, obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set using a first intelligent decision-making unit; determine a target decision set based on the historical decision information and the initial decision set; generate target prompt data based on the target decision set and a reference prompt word; determine target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision set is a subset of the initial decision set, and the target decision information includes target emotion information and target action information; generate a target output result based on the target emotion information and the target action information using a second intelligent decision-making unit, and return the target output result to the client; The client is further configured to receive and display the target output result on the dialogue interaction interface.

[0102] The dialogue interaction interface can be understood as the chat interface of the above embodiment, and the dialogue generation request is a request for the simulation server to generate the target output result of the simulation.

[0103] The specific implementation manner can be referred to the above embodiments and is not limited here.

[0104] Corresponding to the above method embodiment, this specification also provides a data processing system embodiment, Figure 4 FIG1 shows a schematic diagram of a data processing system provided by an embodiment of this specification. Figure 4 As shown, the system includes a first intelligent decision-making unit 402 and a second intelligent decision-making unit 404, wherein: The first intelligent decision-making unit 402 is configured to obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; determine a target decision set based on the historical decision information and the initial decision set, and generate target prompt data based on the target decision set and reference prompt words, wherein the target decision set is a subset of the initial decision set; and determine target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information; The second intelligent decision-making unit 404 is configured to generate a target output result according to the target emotion information and the target action information.

[0105] Optionally, the first intelligent decision-making unit 402 is further configured to: Determining a target emotion set based on the historical emotion information and the initial emotion set; The target decision set is determined according to the target emotion set and the action set.

[0106] Optionally, the first intelligent decision-making unit 402 is further configured to: Acquire an emotion state machine, wherein the emotion state machine is determined according to emotion information in the initial emotion set; According to the emotion state machine, candidate emotion information associated with the historical emotion information is determined from the initial emotion set, and the target emotion set is generated according to the candidate emotion information.

[0107] Optionally, the first intelligent decision-making unit 402 is further configured to: According to the historical dialogue data and the target prompt data, the target emotion information is determined from the target emotion set, and the target action information is determined from the action set.

[0108] Optionally, the first intelligent decision-making unit 402 is further configured to: The target emotion identifier corresponding to the target emotion information is determined and output, and the target action identifier corresponding to the target action information is determined and output.

[0109] The system further includes an information determination module configured to determine the target emotion description information corresponding to the target emotion identifier and determine the target action description information corresponding to the target action identifier.

[0110] Optionally, the second intelligent decision-making unit 404 is further configured to: The target output result is generated according to the target emotion description information and the target action description information.

[0111] The system further includes a prompt word acquisition module, configured to acquire the reference prompt word according to initial prompt data, wherein the initial prompt data is composed of the reference prompt word and the initial decision set.

[0112] The system also includes a storage module, which, when obtaining target response data corresponding to the target output result, updates and marks the historical conversation data, the target output result, and the target response data as the historical conversation data; and updates and marks the target decision information and the historical decision information as the historical decision information.

[0113] The data processing system provided in the embodiment of this specification, when obtaining historical dialogue data, historical decision information corresponding to the historical dialogue data, and an initial decision set, determines a subset of the initial decision set (target decision set) based on the historical decision information and the initial decision set, that is, different target decision sets can be determined from the initial decision set based on different historical decision information, and then target prompt data is generated based on the target decision set and reference prompt words, thereby achieving different target prompt data based on different historical decision information, and determining target decision information from the target decision set based on the historical dialogue data and the target prompt data, the target decision information including target emotion information and target action information, that is, different historical decision information can be determined from the initial decision set. The policy information can determine the target emotion information and target action information from different target decision sets, and generate the target output result based on the target emotion information and target action information. The target decision set is determined based on the historical decision information and the initial decision set, and the target decision information including the target emotion information and target action information is determined from the target decision set. It is achieved that for the initially set initial decision set, different target decision sets can be determined corresponding to different historical dialogue data and historical decision information, and then the target emotion information and target action information are determined from the target decision set, so as to realize the emotional flow and behavioral response of the dialogue process, so that users can obtain a personalized and natural and smooth dialogue experience.

[0114] The above is a schematic diagram of a data processing system according to this embodiment. It should be noted that the technical solution of the data processing system and the technical solution of the above-mentioned data processing method are based on the same concept. For details not described in detail in the technical solution of the data processing system, please refer to the description of the technical solution of the above-mentioned data processing method.

[0115] Figure 5 FIG. 5 shows a structural block diagram of a computing device 500 provided according to an embodiment of the present specification.

[0116] The computing device 500 includes: A memory 510 and a processor 520 , wherein the memory 510 and the processor 520 are connected via a bus 530 ; The memory 510 is used to store computer programs / instructions, and the processor 520 is used to execute the computer programs / instructions. When the computer programs / instructions are executed by the processor 520, the steps of the method are implemented.

[0117] Specifically, the components of the computing device 500 include but are not limited to a memory 510 and a processor 520. The processor 520 is connected to the memory 510 via a bus 530, and the database 550 is used to store data.

[0118] The computing device 500 also includes an access device 540 that enables the computing device 500 to communicate via one or more networks 560. Examples of such networks include a public switched telephone network (PSTN), a local area network (LAN), a wide area network (WAN), a personal area network (PAN), or a combination of communication networks such as the Internet. The access device 540 may include one or more of any type of network interface (e.g., a network interface card (NIC)) whether wired or wireless, such as an IEEE 802.11 wireless local area network (WLAN) wireless interface, a Worldwide Interoperability for Microwave Access (Wi-MAX) interface, an Ethernet interface, a universal serial bus (USB) interface, a cellular network interface, a Bluetooth interface, a near field communication (NFC) interface, and the like.

[0119] In one embodiment of the present specification, the above components of the computing device 500 and Figure 5 Other components not shown in the figure may also be connected to each other, for example, via a bus. Figure 5 The computing device structure block diagram shown is for illustrative purposes only and is not intended to limit the scope of this specification. Those skilled in the art may add or replace other components as needed.

[0120] Computing device 500 can be any type of stationary or mobile computing device, including a mobile computer or mobile computing device (e.g., a tablet computer, personal digital assistant, laptop computer, notebook computer, netbook computer, etc.), a mobile phone (e.g., a smartphone), a wearable computing device (e.g., a smartwatch, smart glasses, etc.), or other types of mobile devices, or a stationary computing device such as a desktop computer or personal computer (PC). Computing device 500 can also be a mobile or stationary server.

[0121] The above is a schematic scheme of a computing device of this embodiment. It should be noted that the technical scheme of the computing device and the technical scheme of the above-mentioned data processing method are of the same concept. For details not described in detail in the technical scheme of the computing device, please refer to the description of the technical scheme of the above-mentioned data processing method.

[0122] An embodiment of the present specification further provides a computer-readable storage medium storing a computer program / instruction, which implements the steps of the above-mentioned data processing method when executed by a processor.

[0123] The above is a schematic scheme of a computer-readable storage medium of this embodiment. It should be noted that the technical scheme of the storage medium and the technical scheme of the above-mentioned data processing method are based on the same concept. For details not described in detail in the technical scheme of the storage medium, please refer to the description of the technical scheme of the above-mentioned data processing method.

[0124] An embodiment of the present specification further provides a computer program product, comprising a computer program / instruction, which implements the steps of the above-mentioned data processing method when executed by a processor.

[0125] The above is a schematic solution of a computer program product of this embodiment. It should be noted that the technical solution of this computer program product and the technical solution of the above-mentioned data processing method are based on the same concept. For details not described in detail in the technical solution of the computer program product, please refer to the description of the technical solution of the above-mentioned data processing method.

[0126] The foregoing description of this specification describes specific embodiments. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be performed in an order different from that described in the embodiments and still achieve the desired results. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order shown or the sequential order to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0127] The computer program / instructions include computer program code, which may be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal, and software distribution medium. It should be noted that the content of the computer-readable medium may be appropriately increased or decreased based on the requirements of patent practice. For example, in some regions, according to patent practice, computer-readable media does not include electric carrier signals and telecommunication signals.

[0128] It should be noted that for the aforementioned method embodiments, for the sake of simplicity of description, they are all expressed as a series of action combinations, but those skilled in the art should be aware that the embodiments of this specification are not limited by the order of the actions described, because according to the embodiments of this specification, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by the embodiments of this specification.

[0129] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0130] The preferred embodiments disclosed above are intended only to help illustrate this specification. The optional embodiments do not exhaustively describe all details, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made based on the content of the embodiments of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the embodiments of this specification, so that those skilled in the art can better understand and utilize this specification. This specification is limited only by the claims and their full scope and equivalents.

Claims

1. A data processing method, comprising: Obtaining historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; Determining a target decision set based on the historical decision information and the initial decision set, and generating target prompt data based on the target decision set and reference prompt words, wherein the target decision set is a subset of the initial decision set; Determining target decision information from the target decision set according to the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information; A target output result is generated according to the target emotion information and the target action information.

2. The data processing method according to claim 1, wherein the historical decision information includes historical emotion information, and the initial decision set includes an initial emotion set and an action set; Determining a target decision set based on the historical decision information and the initial decision set includes: Determining a target emotion set based on the historical emotion information and the initial emotion set; The target decision set is determined according to the target emotion set and the action set.

3. The data processing method according to claim 2, wherein determining the target emotion set based on the historical emotion information and the initial emotion set comprises: Acquire an emotion state machine, wherein the emotion state machine is determined according to emotion information in the initial emotion set; According to the historical emotion information, candidate emotion information is determined from the emotion state machine, and the target emotion set is generated according to the candidate emotion information.

4. The data processing method according to claim 2, wherein determining target decision information from the target decision set based on the historical conversation data and the target prompt data comprises: According to the historical dialogue data and the target prompt data, the target emotion information is determined from the target emotion set, and the target action information is determined from the action set.

5. The data processing method according to any one of claims 1 to 4, wherein the target emotion information includes a target emotion identifier, and the target action information includes a target action identifier; After determining target decision information from the target decision set based on the historical conversation data and the target prompt data, the method further includes: The target emotion identifier in the target emotion information is determined and output, and the target action identifier in the target action information is determined and output.

6. The data processing method according to claim 4, wherein the target emotion information includes target emotion description information, and the target action information includes target action description information; Before generating a target output result according to the target emotion information and the target action information, the method further includes: Determine the target emotion description information corresponding to the target emotion identifier, and determine the target action description information corresponding to the target action identifier; Generating a target output result according to the target emotion information and the target action information includes: The target output result is generated according to the target emotion description information and the target action description information.

7. The data processing method according to claim 1, before generating target prompt data based on the target decision set and reference prompt words, further comprising: The reference prompt word is obtained according to initial prompt data, wherein the initial prompt data consists of the reference prompt word and the initial decision set.

8. The data processing method according to any one of claims 1 to 4, 6 to 7, applied to a data processing system, wherein the data processing system comprises a first intelligent decision-making unit; Determining target decision information from the target decision set based on the historical conversation data and the target prompt data includes: In the first intelligent decision unit, target decision information is determined from the target decision set based on the historical dialogue data and the target prompt data.

9. The data processing method according to any one of claims 1 to 4, 6 to 7, applied to a data processing system, wherein the data processing system comprises a second intelligent decision-making unit; Generating a target output result according to the target emotion information and the target action information includes: In the second intelligent decision-making unit, a target output result is generated according to the target emotion information and the target action information.

10. The data processing method according to any one of claims 1 to 4, 6 to 7, further comprising: after generating a target output result based on the target emotion information and the target action information; In the case of obtaining target reply data corresponding to the target output result, updating and marking the historical conversation data, the target output result, and the target reply data as the historical conversation data; Update and mark the target decision information and the historical decision information as the historical decision information; Continue to execute the steps of obtaining the historical conversation data, the historical decision information corresponding to the historical conversation data, and the initial decision set.

11. A data processing system comprising a first intelligent decision-making unit and a second intelligent decision-making unit, wherein: The first intelligent decision-making unit is configured to obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; Determining a target decision set based on the historical decision information and the initial decision set, and generating target prompt data based on the target decision set and a reference prompt word, wherein the target decision set is a subset of the initial decision set; determining target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein the target decision information includes target emotion information and target action information; The second intelligent decision-making unit is used to generate a target output result based on the target emotion information and the target action information.

12. A conversation simulation system, comprising a client and a simulation server; The client is configured to send a dialog generation request to the simulation server in response to an input operation on the dialog interaction interface; The simulation server is configured to, based on the conversation generation request, use a first intelligent decision unit to obtain historical conversation data, historical decision information corresponding to the historical conversation data, and an initial decision set; determine a target decision set based on the historical decision information and the initial decision set; generate target prompt data based on the target decision set and a reference prompt word; and determine target decision information from the target decision set based on the historical conversation data and the target prompt data, wherein: The target decision set is a subset of the initial decision set, and the target decision information includes target emotion information and target action information; generating a target output result based on the target emotion information and the target action information using a second intelligent decision-making unit, and returning the target output result to the client; The client is further configured to receive and display the target output result on the dialogue interaction interface.

13. A computing device comprising: memory and processor; The memory is used to store computer programs / instructions, and the processor is used to execute the computer programs / instructions. When the computer program / instructions are executed by the processor, the steps of the method according to any one of claims 1 to 10 are implemented.

14. A computer-readable storage medium storing a computer program / instruction, wherein the computer program / instruction, when executed by a processor, implements the steps of the method according to any one of claims 1 to 10.

15. A computer program product comprising a computer program / instructions, which implement the steps of the method according to any one of claims 1 to 10 when executed by a processor.

Images