Question reply method and related equipment
By generating initial, unique prompts and dynamically adjusting the toolset, the problem of existing question-and-answer methods being unable to adapt to changes in users was solved, thus achieving high-quality, personalized responses.
Patent Information
- Application Number
- CN202510993446.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-17
- Publication Date
- 2025-10-28
AI Technical Summary
Existing question response methods are unable to effectively adapt to changes in users' current situations, resulting in low response quality.
By generating initial personalized prompts for target users, the toolset is dynamically adjusted based on user behavior data to optimize the question response process.
It improved the quality of question responses, making them more relevant to users' actual needs, avoiding distractions caused by tool proliferation, and enhancing execution effectiveness.
Smart Images
Figure CN120849562A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of problem response, and more specifically, to problem response methods, problem response apparatus, problem response devices, computer-readable storage media, and computer program products containing instructions. Background Technology
[0002] As business needs evolve, users encounter more and more problems, necessitating problem responses. Current problem-solving methods match user questions with pre-defined rules, processes, or prompts, and then generate corresponding answers.
[0003] However, existing question response methods can only respond to questions based on pre-set rules, processes, or prompts, lacking a comprehensive understanding of users and failing to adapt well to current user changes. Therefore, the quality of question responses from existing question response methods is low. Summary of the Invention
[0004] This application provides a problem response method, a problem response apparatus, a problem response device, a computer-readable storage medium, and a computer program product containing instructions, for performing problem responses while improving the quality of problem responses.
[0005] In a first aspect, embodiments of this application provide a method for answering questions, including:
[0006] Based on the target user information obtained from the target tenant's target business system, an initial exclusive prompt word is generated for the target user, wherein the target user information includes user information from various data sources;
[0007] Based on the initial unique prompt words of the target user, the initial toolset of the target user is determined in the pre-registered toolset of the target tenant. The pre-registered toolset of the target tenant is determined in a preset toolset pool based on the registration request of the target tenant.
[0008] Based on the target user's behavioral data, dynamically adjust the target user's initial personalized prompts and the target user's initial toolset to obtain the target user's target personalized prompts and the target user's target toolset;
[0009] Invoke the target user's target tool to answer the target question.
[0010] Secondly, embodiments of this application provide a problem-answering device, including:
[0011] The generation unit is used to generate an initial exclusive prompt word for the target user based on the target user information obtained from the target user's target business system of the target tenant. The target user information includes user information from various data sources.
[0012] The determining unit is used to determine the initial toolset of the target user based on the initial exclusive prompt word of the target user in the pre-registered toolset of the target tenant. The pre-registered toolset of the target tenant is determined from a preset toolset pool based on the registration request of the target tenant.
[0013] An adjustment unit is used to dynamically adjust the initial personalized prompts and the initial toolset of the target user based on the target user's behavioral data, so as to obtain the target personalized prompts and the target toolset of the target user;
[0014] The invocation unit is used to invoke the target user's target tool to respond to the target question.
[0015] Thirdly, embodiments of this application provide a problem-responding device, including:
[0016] Central processing unit, memory, input / output interfaces, wired or wireless network interfaces, and power supply;
[0017] The memory is either a short-term storage memory or a persistent storage memory;
[0018] The central processing unit is configured to communicate with the memory and execute instructions in the memory to perform the aforementioned problem response method.
[0019] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a computer, cause the computer to perform the aforementioned problem-solving method.
[0020] Fifthly, embodiments of this application provide a computer program product containing instructions that, when run on a computer, cause the computer to execute the aforementioned problem-solving method.
[0021] As can be seen from the above technical solutions, the embodiments of this application have the following advantages: This application generates initial exclusive prompts for users, accurately depicts user characteristics, makes the response more in line with the actual needs of users, and can dynamically adjust the toolset according to the target user's behavioral data, ensuring that the target user can complete the work efficiently in different tasks, avoids distraction caused by the growth of tools, and improves the execution effect. Therefore, the question response method of this application has a high quality of question response.
[0022] Accordingly, the problem-solving apparatus, problem-solving device, computer-readable storage medium, and computer program product containing instructions provided in this application also have the aforementioned technical effects. Attached Figure Description
[0023] Figure 1 This is a schematic diagram of the architecture of a question response system disclosed in an embodiment of this application;
[0024] Figure 2 This is a flowchart illustrating a problem-solving method disclosed in an embodiment of this application;
[0025] Figure 3 This is a flowchart illustrating another question-answering method disclosed in an embodiment of this application;
[0026] Figure 4 This is a schematic diagram of the structure of a problem response device disclosed in an embodiment of this application;
[0027] Figure 5 This is a schematic diagram of the structure of a problem response device disclosed in an embodiment of this application. Detailed Implementation
[0028] This application provides a problem response method, a problem response apparatus, a problem response device, a computer-readable storage medium, and a computer program product containing instructions, for performing problem responses while improving the quality of problem responses.
[0029] Please see Figure 1 One architecture of the question response system in this application includes:
[0030] The system consists of a problem response device (e.g., a server in a SaaS (Software as a Service) architecture) 101 and a client (target tenant, e.g., a client in a SaaS architecture) 102. When responding to a problem, the problem response device 101 can connect to the client (target tenant) 102. The problem response device 101 can obtain target user information from the target business system of the client (target tenant) 102, generate an initial personalized prompt for the target user based on this information, determine the target user's initial toolset from the pre-registered toolset of the target tenant based on the initial personalized prompt, dynamically adjust the initial personalized prompt and initial toolset based on the target user's behavioral data, obtain the target user's target personalized prompt and target toolset, call the target user's target tools to respond to the target problem, and finally return the response information to the client (target tenant).
[0031] based on Figure 1 Please refer to the question and answer system shown. Figure 2 , Figure 2 This is a flowchart illustrating a problem-solving method disclosed in an embodiment of this application. The method includes:
[0032] 201. Based on the target user information obtained from the target tenant's target business system, generate initial exclusive prompts for the target user. The target user information includes user information from various data sources.
[0033] In one alternative implementation, the target business system can refer to enterprise office software and its various sub-business systems, such as chat systems, organizational structure management systems, knowledge base systems, and process management systems. Data sources include chat logs, organizational structure information, knowledge base content, and workflow data. User profile information represents basic user information (such as name, department, and position), work habits, and preferences. Initial personalized prompts represent personalized prompts generated for each user that reflect their individual characteristics and needs. These prompts guide the AI Agent (artificial intelligence agent, i.e., AI intelligent agent) to provide personalized services tailored to the user's characteristics during interactions.
[0034] 202. Based on the target user's initial exclusive prompt word, determine the target user's initial toolset from the target tenant's pre-registered toolset. The target tenant's pre-registered toolset is determined from the preset toolset pool based on the target tenant's registration request.
[0035] In one alternative implementation, the preset tool pool represents a collection of various available tools and functions provided for the enterprise AI Agent. The pre-registered toolset of the target tenant represents the set of tools that the target tenant selects and registers from the preset tool pool according to its own needs.
[0036] 203. Based on the target user's behavioral data, dynamically adjust the target user's initial exclusive prompts and the target user's initial toolset to obtain the target user's target exclusive prompts and the target user's target toolset.
[0037] In one optional implementation, the target user's behavioral data represents the behavioral records generated during the user's interaction with the AI Agent, such as dialogue content, tools used, and operation frequency. By analyzing user behavioral data, the system dynamically adjusts personalized prompts to dynamically adjust the priority and weight of tool calls, better adapting to the user's real-time needs and preferences. For example, for users who frequently need data analysis, data analysis tools are prioritized; for users who need document editing, document editing tools are prioritized, ensuring efficient tool usage.
[0038] 204. Use the target user's target tool to answer the target question.
[0039] In one alternative implementation, for example, the AI Agent in the enterprise office software generates initial prompts based on the information of newly hired employees, determines the toolset, and adjusts the prompts and toolset as the employee's usage behavior changes.
[0040] In this way, by generating initial personalized prompts for each user, user characteristics are accurately depicted, making responses more tailored to the user's actual needs. Furthermore, the toolset can be dynamically adjusted based on the target user's behavioral data, ensuring efficient work completion across different tasks, avoiding distractions caused by tool proliferation, and improving execution effectiveness. The AI Agent role in this application is customized to the user, enhancing the quality of responses. Therefore, the question-response method in this application yields high-quality responses. Secondly, the target users can represent different roles across various industries; therefore, the system design is versatile and flexible, quickly adapting to the needs of different industries and roles, and exhibiting strong scalability. Furthermore, personalized prompts can be dynamically adjusted based on real-time behavioral data of the target user, continuously evolving with user usage. The more interaction with the target user, the more practical the system becomes, creating a virtuous cycle and enhancing user engagement.
[0041] In one optional implementation, dynamically adjusting the target user's initial personalized prompts and initial toolset based on the target user's behavioral data to obtain the target user's target personalized prompts and target toolset includes: converting the target user's behavioral data into the target user's behavioral log and storing it in a buffer; if a preset comparison condition is met, comparing the target user's behavioral log in the buffer with the target user's initial personalized prompts, and dynamically adjusting the target user's initial personalized prompts and initial toolset to obtain the target user's target personalized prompts and target toolset.
[0042] Specifically, the target user's behavior logs represent a highly summarized and recorded information about user behavior data. By accumulating a certain amount of behavior logs in a buffer before comparison and adjustment, this buffering mechanism simulates the human cognitive process of quantitative change leading to qualitative change. It avoids the system burden caused by premature updates and also avoids frequent updates due to accidental user behavior, ensuring system stability and the effectiveness of updates.
[0043] In an optional implementation, the method further includes: if the current time reaches a preset time threshold, and / or the number of target user behavior logs in the buffer reaches a preset number threshold, then it is determined that the preset comparison condition is met.
[0044] Specifically, for example: the personalized prompts are only updated when the user's behavior logs have accumulated to a certain extent or after a period of time, to ensure the stability and efficiency of the system.
[0045] By setting clear time and quantity thresholds, the system can ensure timely responses to changes in user behavior patterns. This mechanism balances the timeliness and stability of system updates, preventing user experience from being negatively impacted by updates that are too frequent or too untimely.
[0046] In one optional implementation, the initial personalized prompt words for the target user are generated based on the target user information obtained from the target user's target business system of the target tenant. This includes: extracting corresponding information from the target user's target user information according to fixed fields and extended fields to summarize the initial prompt words for the target user; enriching the initial prompt words for the target user to obtain the enriched initial prompt words for the target user; and inserting the enriched initial prompt words for the target user into the dynamic part of the system prompt words to generate the initial personalized prompt words for the target user.
[0047] Specifically, fixed fields refer to standardized user information such as department and job title. Extended fields refer to supplementary user information such as years of service and personal interests. Enriching the initial prompts involves adding more relevant and detailed information after the initial prompts are generated, making the prompts more comprehensive and specific to enhance the AI Agent's understanding and responsiveness to user needs. It can be understood that system prompts include dynamic and static components. The dynamic component can be adjusted according to changes in user behavior, while the static component contains fixed rules and specifications (such as harmless defense, rule baselines, and enterprise standards).
[0048] In this way, by extracting fixed and extended fields and enriching the initial prompt words, more comprehensive and exclusive prompt words can be generated, which can more accurately describe user characteristics and provide a more solid foundation for subsequent personalized services.
[0049] In one optional implementation, there are multiple target users. Information is extracted from the target user profiles of each target user according to fixed and extended fields to generate initial prompts for each target user. This includes: distributing the target user profiles of each target user to multiple nodes; each node extracting corresponding field values from the target user profiles of each target user in its local data block according to fixed and extended fields to generate key-value pair intermediate results for each target user; obtaining the key-value pair intermediate results generated by each node for each target user; and then summarizing the key-value pair intermediate results corresponding to the same key for each target user to generate initial prompts for each target user.
[0050] By distributing user data across multiple nodes for parallel processing, computing resources can be fully utilized, accelerating data processing. By aggregating the intermediate key-value pairs generated by each node, initial prompts for each user can be efficiently created, thereby improving the overall system performance.
[0051] In an optional implementation, before generating the initial exclusive prompt word for the target user based on the target user information obtained from the target business system of the target tenant, the method further includes: after obtaining the initial user information of the target user from the target business system of the target tenant, determining the target parameters missing in the initial user information of the target user based on a preset parameter template, obtaining the supplementary information of the target parameters returned after actively raising the supplementary information question of the target parameters, and generating the target user information of the target user based on the initial user information of the target user and the supplementary information of the target parameters.
[0052] Specifically, the preset parameter template represents a template used to determine missing parameters in user profile information.
[0053] In this way, by determining the missing target parameters through preset parameter templates and obtaining supplementary information, user profiles can be improved, thereby enhancing the accuracy and completeness of the initial personalized prompts.
[0054] In one optional implementation, dynamically adjusting the target user's initial personalized prompts and initial toolset based on the target user's behavioral data to obtain the target user's target personalized prompts and target toolset includes: dynamically adjusting the target user's initial personalized prompts based on the target user's behavioral data to obtain the target user's target personalized prompts, acquiring the target user's long-term memory information, and dynamically adjusting the target user's initial toolset based on the target user's target personalized prompts and long-term memory information to obtain the target user's target toolset.
[0055] Specifically, the target user's long-term memory information representation utilizes long-term user behavior and preference information obtained through Mem0 technology. This application can be combined (fused) with other memory layer tools and RAG systems to achieve more powerful functions and richer application scenarios. For example, combined with Mem0 tools, its memory storage and management functions can be used to further enhance the AI Agent's ability to process user memories; combined with RAG systems, retrieved knowledge and information can be better utilized during the answer generation process.
[0056] In this way, by dynamically adjusting the initial toolset in conjunction with long-term memory information, user needs can be predicted more accurately, the toolset can be optimized, and user experience and system usability can be improved.
[0057] To further explain the problem-solving solutions provided in this application, please refer to [link / reference]. Figure 3 , Figure 3 This is a flowchart illustrating another problem-solving method disclosed in an embodiment of this application. Figure 3 It can be seen that, Figure 3 The question-and-answer method shown includes two phases: the "cold start (user modeling) phase" and the "daily conversation (human feedback learning) phase."
[0058] I. Cold Start (User Modeling) Phase (i.e., the phase of generating initial personalized prompts based on the target user's profile information):
[0059] During the cold start phase, the large model extracts user data and information from the business system based on fixed fields (such as basic user information, department, and position) and extended fields (such as supplementary information like user interests and work habits). For complex systems, a Map-Reduce approach is used to efficiently process large-scale data. In the Map phase, data is distributed across multiple computing nodes for local information extraction, generating key-value pair intermediate results. In the Reduce phase, intermediate results with the same key are aggregated, undergoing multiple rounds of extraction and aggregation to ultimately form an initial prompt for the user's profile.
[0060] Example: Suppose a user is a newly hired marketing employee. Their basic information includes name, start date, department (marketing), and position (specialist). Extended fields might include interests mentioned during the recruitment process (e.g., data analysis, social media marketing). The large model integrates these fixed and extended fields when extracting information. For complex enterprise office systems containing large amounts of employee data, a Map-Reduce framework is used. In the Map phase, each computing node processes its assigned employee data slice, extracting basic and extended information for each employee and converting it into key-value pairs (e.g., key "Marketing Specialist Interests," value "Data Analysis, Social Media Marketing"). In the Reduce phase, the system aggregates all key-value pairs with the same key, statistically analyzes common interests of marketing specialists, and combines this with other information. After multiple rounds of iterative optimization, it generates initial prompts for the new user, such as: "Newly hired marketing specialist, interested in data analysis and social media marketing."
[0061] II. Daily Conversation (Human Feedback Learning) Stage (i.e., the stage of dynamically adjusting personalized prompts based on the target user's behavioral data):
[0062] The original agent call chain refers to the path by which an agent invokes tools and performs operations when executing a task. It includes a series of steps and the order in which tools are invoked to complete the task. For example, planning a marketing campaign might involve the following steps: First, the agent invokes a market research tool to collect information about the target audience and competitors. Then, based on the research results, it invokes an event planning tool to determine the theme, content, and schedule of the event. Next, it invokes a budget allocation tool to allocate campaign funds appropriately. Afterward, it invokes a promotion tool to develop promotional strategies and channels. Finally, it invokes an effectiveness evaluation tool to track and evaluate the event's results. Throughout this process, the agent follows the pre-defined call chain to progressively complete the planning and execution of the marketing campaign. Tools / MPC (Multi-Tool / Multi-Capability) are a collection of various tools and capabilities that an agent can use, including document editing tools, data analysis tools, project management tools, email sending tools, etc. Identity change influences tendency, meaning that the agent's identity (reflected by a unique prompt) affects its preferences and tendencies when invoking tools. Specifically, the identity characteristics of an intelligent agent (such as the user's role, job content, etc.) determine which tools it tends to call when faced with different tasks, and how to combine these tools to complete the tasks.
[0063] Example: Suppose the agent's identity is a marketing manager (identified by a specific prompt), and their task is to plan a marketing campaign. The original agent's call chain might include market research, campaign planning, execution, and performance evaluation. Because the agent's identity is that of a marketing manager, market research tools (such as data analysis tools and competitor analysis tools) and campaign planning tools (such as creative generation tools and budget allocation tools) have higher weightings in the tools / MPC (Machine Programming Center) framework. The change in identity influences the agent's tendency to prioritize the use of these tools related to the marketing manager role. This process demonstrates the impact of the agent's identity characteristics on their tool-calling tendencies, and how the original agent's call chain organizes tool calls based on identity characteristics and task requirements.
[0064] For further details, please refer to Figure 4 One embodiment of the problem response device in this application includes:
[0065] The generation unit is used to generate an initial exclusive prompt word for the target user based on the target user information obtained from the target user's target business system of the target tenant. The target user information includes user information from various data sources.
[0066] The determining unit is used to determine the initial toolset of the target user based on the initial exclusive prompt word of the target user in the pre-registered toolset of the target tenant. The pre-registered toolset of the target tenant is determined from a preset toolset pool based on the registration request of the target tenant.
[0067] An adjustment unit is used to dynamically adjust the initial personalized prompts and the initial toolset of the target user based on the target user's behavioral data, so as to obtain the target personalized prompts and the target toolset of the target user;
[0068] The invocation unit is used to invoke the target user's target tool to respond to the target question.
[0069] In one alternative implementation, the adjustment unit can be used to:
[0070] The target user's behavior data is converted into the target user's behavior log and stored in a buffer. If a preset comparison condition is met, the target user's behavior log in the buffer is compared with the target user's initial exclusive prompt words. The target user's initial exclusive prompt words and the target user's initial toolset are dynamically adjusted to obtain the target user's target exclusive prompt words and the target user's target toolset.
[0071] In one alternative implementation, the determining unit may be used for:
[0072] If the current time reaches a preset time threshold, and / or the number of the target user's behavior logs in the buffer reaches a preset number threshold, then the preset comparison condition is determined to be met.
[0073] In one alternative implementation, the generating unit may be used for:
[0074] Information is extracted from the target user profile information of the target user according to the fixed fields and extended fields to summarize the initial prompt words for creating the target user. The initial prompt words for the target user are enriched to obtain the enriched initial prompt words for the target user. The enriched initial prompt words for the target user are inserted into the dynamic part of the system prompt words to generate the initial exclusive prompt words for the target user.
[0075] In one alternative implementation, the generating unit may be used for:
[0076] The target user information of each target user is distributed to multiple nodes. Each node is used to extract the corresponding field values from the target user information of each target user in the local data block according to the fixed field and the extended field to generate key-value pair intermediate results for each target user. After obtaining the key-value pair intermediate results of each target user generated by each node, the key-value pair intermediate results corresponding to the same key of each target user are summarized to generate the initial prompt words for each target user.
[0077] In one alternative implementation, the determining unit may be used for:
[0078] After obtaining the initial user information of the target user from the target business system of the target tenant, the target parameters missing in the initial user information of the target user are determined based on the preset parameter template.
[0079] The generation unit can be used for:
[0080] After obtaining the supplementary information of the target parameter returned after actively proposing the supplementary information question of the target parameter, the target user information is generated based on the initial user information of the target user and the supplementary information of the target parameter.
[0081] In one alternative implementation, the adjustment unit can be used to:
[0082] Based on the target user's behavioral data, the initial personalized prompt words of the target user are dynamically adjusted to obtain the target user's target personalized prompt words. The target user's long-term memory information is obtained. Based on the target user's target personalized prompt words and the target user's long-term memory information, the initial toolset of the target user is dynamically adjusted to obtain the target user's target toolset.
[0083] For further details, please refer to Figure 5 One embodiment of the problem response device in this application includes:
[0084] Central processing unit 501, memory 505, input / output interface 504, wired or wireless network interface 503, and power supply 502;
[0085] Memory 505 is either a short-term storage memory or a persistent storage memory;
[0086] The central processing unit 501 is configured to communicate with the memory 505 and execute instructions stored in the memory 505 to perform the aforementioned operations. Figure 2 The method in the illustrated embodiment.
[0087] Furthermore, embodiments of this application also provide a computer-readable storage medium, which includes instructions that, when executed on a computer, cause the computer to perform the aforementioned... Figure 2 The method in the illustrated embodiment.
[0088] Furthermore, embodiments of this application also provide a computer program product containing instructions, which, when run on a computer, causes the computer to perform the aforementioned... Figure 2 The method in the illustrated embodiment.
[0089] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0090] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0091] 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 units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units 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 an indirect coupling or communication connection between apparatuses or units through some interfaces, and may be electrical, mechanical, or other forms.
[0092] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0093] In addition, the functional units in the various embodiments of the present application may be integrated into a single processing unit, or each unit may exist physically separately, or two or more units may be integrated into a single unit. The aforementioned integrated units may be implemented in the form of hardware or software functional units.
[0094] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it 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 all or part 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, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
Claims
1. A method for answering questions, characterized in that, include: Based on the target user information obtained from the target tenant's target business system, an initial exclusive prompt word is generated for the target user, wherein the target user information includes user information from various data sources; Based on the initial unique prompt words of the target user, the initial toolset of the target user is determined in the pre-registered toolset of the target tenant. The pre-registered toolset of the target tenant is determined in a preset toolset pool based on the registration request of the target tenant. Based on the target user's behavioral data, dynamically adjust the target user's initial personalized prompts and the target user's initial toolset to obtain the target user's target personalized prompts and the target user's target toolset; Invoke the target user's target tool to answer the target question.
2. The method according to claim 1, characterized in that, The process of dynamically adjusting the initial personalized prompts and initial toolset of the target user based on the target user's behavioral data to obtain the target user's target personalized prompts and target toolset includes: The target user's behavioral data is converted into the target user's behavioral log and stored in a buffer; If the preset comparison conditions are met, the behavior log of the target user in the buffer is compared with the initial exclusive prompt words of the target user, and the initial exclusive prompt words and the initial toolset of the target user are dynamically adjusted to obtain the target exclusive prompt words and the target toolset of the target user.
3. The method according to claim 2, characterized in that, The method further includes: If the current time reaches a preset time threshold, and / or the number of the target user's behavior logs in the buffer reaches a preset number threshold, then the preset comparison condition is determined to be met.
4. The method according to claim 1, characterized in that, The step of generating initial personalized prompts for the target user based on the target user information obtained from the target tenant's target business system includes: Based on the fixed fields and extended fields, extract the corresponding information from the target user's profile information to summarize the initial prompt words for creating the target user; The initial prompt words for the target user are enriched to obtain the enriched initial prompt words for the target user; The enriched initial prompt word corresponding to the target user is inserted into the dynamic part of the system prompt word to generate the initial exclusive prompt word for the target user.
5. The method according to claim 4, characterized in that, The number of target users is multiple. The step of extracting corresponding information from the target user profile information of each target user according to fixed fields and extended fields to summarize the initial prompt words for creating the target users includes: The target user information of each target user is distributed to multiple nodes. Each node is used to extract the corresponding field values from the target user information of each target user in the local data block according to the fixed field and the extended field, so as to generate key-value pair intermediate results for each target user. After obtaining the intermediate results of key-value pairs of each target user generated by each node, the intermediate results of key-value pairs corresponding to the same key of each target user are summarized to generate the initial prompt words of each target user.
6. The method according to claim 1, characterized in that, Before generating the initial exclusive prompt word for the target user based on the target user information obtained from the target tenant's target business system, the method further includes: After obtaining the initial user information of the target user from the target business system of the target tenant, the target parameters missing in the initial user information of the target user are determined based on the preset parameter template. After obtaining the supplementary information of the target parameter returned after actively proposing the supplementary information question of the target parameter, the target user information is generated based on the initial user information of the target user and the supplementary information of the target parameter.
7. The method according to any one of claims 1 to 6, characterized in that, The process of dynamically adjusting the initial personalized prompts and initial toolset of the target user based on the target user's behavioral data to obtain the target user's target personalized prompts and target toolset includes: Based on the target user's behavioral data, the initial personalized prompt words for the target user are dynamically adjusted to obtain the target user's target personalized prompt words; Obtain the long-term memory information of the target user; Based on the target user's target-specific prompts and the target user's long-term memory information, the target user's initial toolset is dynamically adjusted to obtain the target user's target toolset.
8. A problem response device, characterized in that, include: The generation unit is used to generate an initial exclusive prompt word for the target user based on the target user information obtained from the target user's target business system of the target tenant. The target user information includes user information from various data sources. The determining unit is used to determine the initial toolset of the target user based on the initial exclusive prompt word of the target user in the pre-registered toolset of the target tenant. The pre-registered toolset of the target tenant is determined from a preset toolset pool based on the registration request of the target tenant. An adjustment unit is used to dynamically adjust the initial personalized prompts and the initial toolset of the target user based on the target user's behavioral data, so as to obtain the target personalized prompts and the target toolset of the target user; The invocation unit is used to invoke the target user's target tool to respond to the target question.
9. A problem response device, characterized in that, include: Central processing unit and memory; The memory is either a short-term storage memory or a persistent storage memory; The central processing unit is configured to communicate with the memory and execute instructions in the memory to perform the method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes instructions that, when executed on a computer, cause the computer to perform the method as described in any one of claims 1 to 7.