Dynamic planning intelligent dialogue method and device, electronic equipment and storage medium

Abstract

The invention provides an intelligent dialogue method and device for dynamic planning, electronic equipment and a storage medium. The method comprises the steps of obtaining current input of a user; determining a current dialogue target according to the current input based on a pre-constructed user portrait database; dynamically planning a multi-round dialogue action sequence according to the current dialogue target; and converting the multi-round dialogue action sequence into a text reply, and ending the current dialogue. According to the method, the defect that an existing intelligent dialogue system lacks coherence and personalized service ability in the dialogue process is overcome, the dialogue quality can be effectively improved, the final dialogue content can better fit the latest state of the user, powerful personalized service ability support is provided, and therefore the intelligent level and service quality of the system are improved.

Description

Technical Field

[0001] This invention relates to the field of intelligent dialogue technology, and in particular to a dynamic programming intelligent dialogue method, apparatus, electronic device, and storage medium. Background Technology

[0002] With the development of artificial intelligence and natural language processing technologies, intelligent dialogue systems have been widely used in various fields such as customer service, government affairs, education, and finance. However, most current dialogue systems still suffer from problems such as simplistic responses, limited understanding capabilities, and insufficient user modeling, making it difficult to meet the demand for personalized and continuous interactive services in complex scenarios.

[0003] Traditional intelligent dialogue systems primarily rely on preset rules or keyword-based response mechanisms, using fixed templates to respond by recognizing specific instructions or semantic tags in user input. While these methods are simple to implement and low in deployment costs, their interaction logic is rigid, lacking the ability to understand context and continuously track user behavior characteristics. Furthermore, existing systems often build user profiles based on single interactions, failing to effectively utilize historical behavioral data, resulting in a lack of coherence and personalized service capabilities in the dialogue process.

[0004] In terms of service strategy, traditional dialogue systems typically employ static flow control, failing to dynamically adjust dialogue paths and responses based on real-time user feedback. Even when some systems incorporate machine learning models for intent recognition or sentiment analysis, their model training data is often collected offline, lacking online learning and model update mechanisms. This makes it difficult to adapt to changing user behavior patterns, thus impacting the system's intelligence level and service quality.

[0005] Therefore, there is an urgent need to propose an intelligent dialogue system and method that can achieve deep contextual understanding, dynamic user profile modeling, and adaptive adjustment of dialogue strategies, so as to improve the intelligence, personalization, and user experience of human-computer interaction. Summary of the Invention

[0006] This invention provides a dynamic planning-based intelligent dialogue method, device, electronic device, and storage medium to overcome the shortcomings of existing intelligent dialogue systems in terms of lack of coherence and personalized service capabilities during dialogue, effectively improving dialogue quality and making the final dialogue content more relevant to the user's latest status.

[0007] On one hand, the present invention provides a dynamic planning intelligent dialogue method, comprising: acquiring the user's current input; determining the current dialogue target based on the current input according to a pre-built user profile database; dynamically planning a multi-turn dialogue action sequence according to the current dialogue target; converting the multi-turn dialogue action sequence into a text response, and ending the current dialogue.

[0008] Furthermore, determining the current dialogue target based on the current input, based on the pre-built user profile database, includes: acquiring the user profile database; acquiring the user's professional guidance needs and emotional needs based on the large language model, according to the current input and the user profile database; and determining the current dialogue target based on the professional guidance needs and the emotional needs.

[0009] Furthermore, the step of dynamically planning a multi-turn dialogue action sequence based on the current dialogue goal includes: determining the domain knowledge base corresponding to the current input; and dynamically planning the multi-turn dialogue action sequence based on the current dialogue goal and the domain knowledge base.

[0010] Furthermore, the step of dynamically planning the multi-turn dialogue action sequence based on the current dialogue target and the domain knowledge base includes: matching corresponding knowledge entities and semantic rules from the domain knowledge base based on the current dialogue target; constructing a candidate dialogue action set based on the knowledge entities and the semantic rules; and performing path search on the candidate dialogue action set using a preset dynamic programming algorithm to generate the optimal multi-turn dialogue action sequence.

[0011] Furthermore, the dynamic planning of the multi-turn dialogue action sequence further includes: real-time monitoring of user feedback and identifying the feedback type of the user feedback; if the user feedback type is positive, interacting with the user according to the multi-turn dialogue action sequence; if the user feedback type is negative, dynamically planning the multi-turn dialogue action sequence again according to the current dialogue goal.

[0012] Furthermore, it also includes: updating the user profile database in real time during the dialogue; wherein the user profile database includes the user's historical dialogue records, historical dialogue action sequences, personal information profiles, and preference information.

[0013] Secondly, the present invention also provides a dynamically planned intelligent dialogue device, comprising: a current input acquisition module for acquiring the user's current input; a current dialogue target determination module for determining the current dialogue target based on a pre-built user profile database and the current input; a dialogue action sequence planning module for dynamically planning a multi-turn dialogue action sequence based on the current dialogue target; and a text reply conversion module for converting the multi-turn dialogue action sequence into a text reply, thereby ending the current dialogue.

[0014] Thirdly, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the computer program, implements the intelligent dialogue method of dynamic programming as described above.

[0015] Fourthly, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the intelligent dialogue method of dynamic programming as described above.

[0016] Fifthly, the present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the intelligent dialogue method of dynamic programming as described above.

[0017] The dynamic programming-based intelligent dialogue method provided by this invention acquires the user's current input and, based on a pre-built user profile database, determines the current dialogue goal. Then, based on the current dialogue goal, it dynamically plans a multi-turn dialogue action sequence, converting this sequence into a text response, thus ending the current dialogue. This method overcomes the shortcomings of existing intelligent dialogue systems, which lack coherence and personalized service capabilities during dialogue. It effectively improves dialogue quality, ensuring the final dialogue content is more relevant to the user's current state and providing strong personalized service support, thereby enhancing the system's intelligence level and service quality. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0019] Figure 1 This is a flowchart illustrating the intelligent dialogue method based on dynamic programming provided in an embodiment of the present invention.

[0020] Figure 2 This is a schematic diagram of the structure of the dynamic planning intelligent dialogue device provided in an embodiment of the present invention.

[0021] Figure 3 This is a schematic diagram of the physical structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0023] It's important to note that traditional intelligent dialogue systems primarily rely on preset rules or keyword-based response mechanisms, using fixed templates to respond by recognizing specific instructions or semantic tags in user input. While these methods are simple to implement and low-cost to deploy, their interaction logic is rigid, lacking the ability to understand context and continuously track user behavior. Furthermore, existing systems often build user profiles based on single interactions, failing to effectively utilize historical behavioral data, resulting in a lack of coherence and personalized service capabilities in the dialogue process. In terms of service strategies, traditional dialogue systems typically employ static flow control, unable to dynamically adjust the dialogue path and response content based on real-time user feedback.

[0024] In view of this, the present invention proposes a dynamic programming-based intelligent dialogue method, specifically, Figure 1 A flowchart illustrating the intelligent dialogue method based on dynamic programming provided in an embodiment of the present invention is shown.

[0025] like Figure 1 As shown, the method includes: S110, obtaining the user's current input; S120, determining the current dialogue target based on the current input according to a pre-built user profile database; S130, dynamically planning a multi-turn dialogue action sequence according to the current dialogue target; S140, converting the multi-turn dialogue action sequence into a text response, and ending the current dialogue.

[0026] It should be noted that the dynamic programming intelligent dialogue method provided in this embodiment can be executed by any electronic device. The electronic device here can be not only smart home devices such as smart speakers and smart TVs, but also personal devices such as smartphones, tablets / computers, smartwatches, and smart bracelets, etc., without specific limitations.

[0027] For ease of understanding, the following will take the electronic device as the execution subject and provide a detailed description of steps S110-S140 and related steps.

[0028] S110, Get the user's current input.

[0029] It is easy to understand that electronic devices can receive user input in real time via voice, text, or images. That is to say, the received user input can be voice input, text input, or image input, without being specifically limited here.

[0030] The user input can be in the fields of psychological counseling, education, or medicine, and no specific restrictions are imposed here.

[0031] Based on receiving user input in step S110, step S120 is further executed.

[0032] S120, based on a pre-built user profile database, determine the current dialogue target according to the current input.

[0033] It is easy to understand that accurately identifying a user's dialogue goal is a key step in achieving personalized responses, contextual understanding, and service guidance. Traditional dialogue systems often rely solely on the current input text for intent recognition, easily overlooking the user's long-term interests, historical behavior, and preference characteristics.

[0034] In view of this, in order to improve the intelligence level of the dialogue system, this embodiment combines a pre-built user profile database with the user's current input content for joint analysis and intent reasoning, so as to more accurately determine the current dialogue target.

[0035] It should be noted that the user profile database in this embodiment may include static attribute information and / or dynamic information. Static attribute information includes, but is not limited to, user ID, name, age / date of birth, gender, region / geographical location, occupation, educational background, registration time, device type, and first language used. Dynamic information includes, but is not limited to, user intent information, preference evolution information, behavioral pattern information, and contextual information. Each user has a dedicated user profile database, which can serve as an important foundation / basis for realizing personalized user services.

[0036] Specifically, first, the current user's profile information (i.e., the user profile database corresponding to the current user) is loaded from the profile database based on the user ID. Then, semantic understanding and key information extraction are performed on the user's current input (dialogue content). Combined with the user profile database, the user's true intention is judged to determine the current dialogue target.

[0037] The current dialogue goal refers to the specific purpose or intention that a user hopes to achieve in this round of interaction when engaging in a dialogue with an intelligent system (such as a chatbot, voice assistant, or customer service robot). It is a semantic understanding and abstract expression of the real needs behind the user's current input. The current dialogue goal includes, but is not limited to, emotional guidance, question guidance, and behavioral incentives.

[0038] After determining the current dialogue target based on the current input according to the pre-built user profile database in step S120, step S130 is further executed.

[0039] S130, Based on the current dialogue objective, dynamically plan a multi-turn dialogue action sequence.

[0040] In human-computer interaction, a multi-turn dialogue action sequence refers to a series of logical actions and response strategies that the system should execute to effectively respond to the user's current dialogue objective. Multi-turn dialogue action sequences include, but are not limited to, question guidance, information confirmation, product recommendation, process navigation, emotional reassurance, and decision advancement.

[0041] Traditional dialogue systems often use fixed rules or single intent matching to determine the next action, lacking a comprehensive consideration of user history, preferences, and contextual state. Therefore, in this embodiment, when dynamically planning a multi-turn dialogue action sequence based on the current dialogue goal, factors such as user profile and contextual state can be considered to achieve a more natural and personalized interactive experience.

[0042] Specifically, firstly, based on the current dialogue goal, possible response action types (i.e. candidate actions) can be preliminarily determined. Then, the applicability of the action is judged by combining the user profile and historical context state. Based on semantic coherence and dialogue flow logic, multiple candidate actions are prioritized to construct and output a multi-turn dialogue action sequence that conforms to semantic logic.

[0043] It should be noted that the process of dynamically planning the multi-turn dialogue action sequence in this step can be implemented using various dynamic programming algorithms (such as rule-driven methods combined with reinforcement learning methods), and no specific limitations are made here.

[0044] After dynamically planning the multi-turn dialogue action sequence according to the current dialogue goal in step S130, step S140 is further executed.

[0045] S140, the multi-turn dialogue action sequence is converted into a text reply, and the current dialogue ends.

[0046] What is easy to understand is that after determining the sequence of actions in a multi-turn dialogue, the user is interacted with in the order of the sequence of actions, and the sequence of actions is converted into semantically consistent, stylistically consistent and easy-to-understand natural language text responses and feedback is given to the user, and the current dialogue ends.

[0047] Steps S110-S140 can be implemented through multiple network layers or through the cooperation of multiple agents; no specific limitations are made here.

[0048] For example, in a specific embodiment, steps S110-S140 are implemented through multiple network layers. Specifically, the target network layer obtains the user's current input and determines the current dialogue target based on the pre-built user profile database. The planning network layer dynamically plans a multi-turn dialogue action sequence based on the current dialogue target. The conversation network layer converts the multi-turn dialogue action sequence into natural language text responses. The types of network layers for the target network layer, planning network layer, and conversation network layer can be set according to actual needs and are not specifically limited here.

[0049] In another specific embodiment, steps S110-S140 can be implemented through the collaboration of multiple intelligent agents. Specifically, a first intelligent agent acquires the user's current input and determines the current dialogue target based on a pre-built user profile database; a second intelligent agent dynamically plans a multi-turn dialogue action sequence based on the current dialogue target; and a third intelligent agent converts the multi-turn dialogue action sequence into a natural language text response. The first, second, and third intelligent agents here can be any model capable of achieving the same function, such as a large language model, and are not specifically limited here.

[0050] In this embodiment, by acquiring the user's current input and based on a pre-built user profile database, the current dialogue goal is determined according to the current input. Then, based on the current dialogue goal, a multi-turn dialogue action sequence is dynamically planned, which is then converted into a text response, ending the current dialogue. This method overcomes the shortcomings of existing intelligent dialogue systems in lacking coherence and personalized service capabilities during dialogue, effectively improving dialogue quality. The final dialogue content is more closely aligned with the user's current state, providing strong support for personalized service capabilities, thereby enhancing the system's intelligence level and service quality.

[0051] Based on the above embodiments, the process of determining the current dialogue target will be described in detail below.

[0052] Based on a pre-built user profile database, the current dialogue goal is determined according to the current input, including: obtaining the user profile database; based on a large language model, obtaining the user's professional guidance needs and emotional needs according to the current input and the user profile database; and determining the current dialogue goal based on the professional guidance needs and emotional needs.

[0053] What is easy to understand is that after receiving the user's current input, the user ID is extracted, the user profile database interface is called, the user profile database is queried and obtained, and the latest profile data in the user profile database is extracted.

[0054] Then, the current input is concatenated with the latest profile data and fed into the large language model. The large language model performs semantic analysis on the current input, including named entity recognition (keyword extraction), intent recognition, and sentiment analysis. It also combines historical dialogue records, historical dialogue action sequences, and preference information from the latest profile data to comprehensively determine the user's current psychological / emotional state, thereby obtaining the user's professional guidance needs (corresponding to intent recognition) and emotional needs (corresponding to sentiment analysis and psychological / emotional state).

[0055] Finally, the current dialogue goal is determined based on the user's professional guidance needs and emotional needs. Specifically, corresponding weight values ​​can be assigned to professional guidance needs and emotional needs according to their urgency, thereby determining the current dialogue goal based on these two needs and their corresponding weight values.

[0056] For example, if a user's professional guidance needs are "solving code problems and debugging suggestions" and their emotional needs are "anxiety and helplessness", then the current dialogue goal is to "first soothe their emotions, and then guide them to provide code snippets for analysis".

[0057] For example, if a user's professional guidance needs are "learning methods and resource recommendations" and their emotional needs are "positive attitude and strong thirst for knowledge", then the current dialogue goal is to "provide structured suggestions and encourage consistent practice".

[0058] It is worth mentioning that this embodiment can more accurately determine the user's core needs by observing the content of the user's questions, tone, and emotional expression, thereby adjusting the current dialogue goals and achieving a warmer and more valuable dialogue.

[0059] In this embodiment, by acquiring a user profile database and based on a large language model, the system obtains the user's professional guidance needs and emotional needs according to the current input and the user profile database. Then, based on these needs, the current dialogue objective is determined, and the multi-turn dialogue action sequence is converted into a text response, ending the current dialogue. This method overcomes the shortcomings of existing intelligent dialogue systems in lacking coherence and personalized service capabilities during dialogue, effectively improving dialogue quality. The final dialogue content is more closely aligned with the user's current state, providing strong support for personalized service capabilities, thereby enhancing the system's intelligence level and service quality.

[0060] Based on the above embodiments, the following will further describe in detail the dynamic planning process for multi-turn dialogue action sequences.

[0061] Based on the current dialogue goal, dynamically plan a multi-turn dialogue action sequence, including: determining the domain knowledge base corresponding to the current input; and dynamically planning a multi-turn dialogue action sequence based on the current dialogue goal and the domain knowledge base.

[0062] The process is easy to understand. First, it determines the professional field of the current input, such as education, health, technology, or law, and then accesses the corresponding domain knowledge base, such as medical knowledge, programming skills, or psychological counseling methods. Then, based on the current dialogue goal (such as solving a problem, providing advice, or offering emotional support), it sets the purpose of each interaction step to generate a multi-turn dialogue action sequence.

[0063] Specifically, based on the previously determined current dialogue goal, corresponding knowledge entities and semantic rules are matched from the domain knowledge base, and a candidate dialogue action set is constructed based on the knowledge entities and semantic rules. Then, a preset dynamic programming algorithm (such as rule-driven algorithm combined with reinforcement learning algorithm) is used to perform path search on the candidate dialogue action set to generate the optimal multi-turn dialogue action sequence.

[0064] For example, in a specific implementation, the current input is "I feel increasingly reluctant to talk to people, am I depressed?", and the current dialogue goal is "to make a preliminary assessment of the emotional state, provide psychological support and coping suggestions, and guide the user to seek professional help." Therefore, the current input relates to the field of psychological counseling, and the corresponding domain knowledge base should at least include criteria for identifying depression, emotion regulation techniques, and key points of cognitive behavioral therapy.

[0065] Correspondingly, the dynamic programming multi-turn dialogue action sequence is as follows: "It sounds like your recent emotional state is indeed troubling you. Could you tell me when this feeling started?"; "Has this situation affected your work, life, or sleep?"; "Have you tried doing something relaxing, such as listening to music, taking a walk, or writing in a journal?"; "Sometimes social avoidance is a protective mechanism, but if it persists for a long time, I suggest talking to a professional therapist."; "If you'd like, I can help you find some psychological counseling resources, or we can practice some emotion regulation methods together."

[0066] It is worth mentioning that after dynamically planning the multi-turn dialogue action sequence, this embodiment will pay attention to the user's feedback in real time and dynamically adjust the multi-turn dialogue action sequence according to the user's feedback.

[0067] Specifically, the dynamic planning of multi-turn dialogue action sequences also includes: real-time monitoring of user feedback and identification of the feedback type; interacting with the user according to the multi-turn dialogue action sequence when the user feedback is positive; and dynamically planning the multi-turn dialogue action sequence again according to the current dialogue goal when the user feedback is negative.

[0068] Specifically, if the user's feedback is positive, it means that the user accepts the current response (the specific dialogue action in the current multi-turn dialogue action sequence) and is willing to proceed with the dialogue according to the current plan; conversely, if the user's feedback is negative, it means that the user does not agree with, is not satisfied with, or does not understand the current response (the specific dialogue action in the current multi-turn dialogue action sequence), and may need to change the way it is expressed or adjust the direction.

[0069] Therefore, if the user feedback is positive, continue interacting with the user based on the generated multi-turn dialogue action sequence; conversely, if the user feedback is negative, re-plan the multi-turn dialogue action sequence until all user feedback received is positive.

[0070] In this embodiment, based on the current dialogue objective, corresponding knowledge entities and semantic rules are matched from the domain knowledge base. A candidate dialogue action set is constructed based on these knowledge entities and semantic rules. A preset dynamic programming algorithm is then used to search for paths within the candidate dialogue action set, generating an optimal multi-turn dialogue action sequence. This sequence is adjusted in real-time based on user feedback to obtain a final multi-turn dialogue action sequence that matches negative user feedback. This sequence is then converted into a text response, ending the current dialogue. This method considers not only single-step decisions during the planning process but also constructs a long-term interaction strategy from a global perspective, adjusting it through real-time feedback. This achieves a dynamic programming effect that is significantly different from traditional fixed processes.

[0071] Based on the above embodiments, the following will further describe in detail the update process of the user profile database.

[0072] During the conversation, the user profile database is updated in real time; the user profile database includes the user's historical conversation records, historical conversation action sequences, personal information profiles, and preference information.

[0073] It is easy to understand that, in order to improve the intelligence level and service response capability of the system, a continuous dynamic update mechanism can be established. That is, whenever a new input is received from the user, dynamic key information is extracted from the new input, and the extracted new dynamic key information is compared and analyzed with the existing user profile content to identify the added or changed dynamic information. This automatically updates the user profile content, forming a continuously evolving user profile, and achieving the effect of real-time dynamic updating of the user profile database.

[0074] Dynamic key information includes intent information, preference evolution information, behavioral pattern information, and contextual information, which can be extracted by performing named entity recognition, intent recognition, and sentiment analysis on the input content. Dynamic information refers to some or all of the dynamic key information.

[0075] Intent information refers to capturing a user's current task intent through various questions, descriptions of needs, and expressions of goals raised during multiple rounds of dialogue, such as booking tickets, searching for information, and making consumption decisions. Preference evolution information refers to the preferences that users gradually reveal or change during the dialogue, including brand preferences, functional needs, price sensitivity, and inclinations towards specific options. Behavioral pattern information refers to usage habits and behavioral patterns mined from multiple rounds of user interaction, such as frequently used sentence structures, information retrieval habits, and time-specific contextual preferences. Contextual information includes the scene, emotional state, current dialogue topic, geographical location, and time frame of the dialogue, helping to determine the user's immediate needs and environmental influences.

[0076] Because the user profile database is updated in real time, each round of dialogue will serve as the basis for determining the next dialogue goal. The system will continuously adapt and optimize its response strategy, significantly improving overall dialogue efficiency and user experience.

[0077] In some other embodiments, the dynamic programming intelligent dialogue method provided in this embodiment can be applied to a variety of different scenarios, including but not limited to psychological counseling, intelligent customer service, educational consultation, and medical Q&A.

[0078] Corresponding to the dynamic programming-based intelligent dialogue method described in the above embodiments, the present invention also provides a dynamic programming-based intelligent dialogue device.

[0079] Specifically, Figure 2 A schematic diagram of the structure of the intelligent dialogue device based on dynamic programming provided in an embodiment of the present invention is shown.

[0080] like Figure 2As shown, the device includes: a current input acquisition module 210, used to acquire the user's current input; a current dialogue target determination module 220, used to determine the current dialogue target based on the current input according to a pre-built user profile database; a dialogue action sequence planning module 230, used to dynamically plan a multi-turn dialogue action sequence according to the current dialogue target; and a text reply conversion module 240, used to convert the multi-turn dialogue action sequence into a text reply, at which point the current dialogue ends.

[0081] In this embodiment, the current input is acquired by the current input acquisition module 210, and the current dialogue target determination module 220 determines the current dialogue target based on a pre-built user profile database. Then, the dialogue action sequence planning module 230 dynamically plans a multi-turn dialogue action sequence based on the current dialogue target. Finally, the text reply conversion module 240 converts the multi-turn dialogue action sequence into a text reply, ending the current dialogue. This device overcomes the shortcomings of existing intelligent dialogue systems in lacking coherence and personalized service capabilities during dialogue, effectively improving dialogue quality and ensuring that the final dialogue content is more relevant to the user's current state. It provides strong personalized service capabilities, thereby enhancing the system's intelligence level and service quality.

[0082] It should be noted that the dynamic programming intelligent dialogue device provided in the embodiments of the present invention can be referred to in correspondence with the dynamic programming intelligent dialogue method described in the above embodiments, and will not be repeated here.

[0083] Figure 3 An example is a schematic diagram of the physical structure of an electronic device, such as... Figure 3 As shown, the electronic device may include a processor 310, a communications interface 320, a memory 330, and a communication bus 340, wherein the processor 310, communications interface 320, and memory 330 communicate with each other via the communication bus 340. The processor 310 can invoke logical instructions in the memory 330 to execute a dynamic programming intelligent dialogue method. This method includes: acquiring the user's current input; determining the current dialogue target based on a pre-built user profile database and the current input; dynamically programming a multi-turn dialogue action sequence based on the current dialogue target; converting the multi-turn dialogue action sequence into a text response, and ending the current dialogue.

[0084] Furthermore, the logical instructions in the aforementioned memory 330 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, essentially, or the part that contributes to the prior art, or a 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 the present invention. 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.

[0085] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to execute the dynamic planning intelligent dialogue method provided by the above methods. The method includes: acquiring the user's current input; determining the current dialogue target based on the current input according to a pre-built user profile database; dynamically planning a multi-turn dialogue action sequence according to the current dialogue target; converting the multi-turn dialogue action sequence into a text response, and ending the current dialogue.

[0086] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements an intelligent dialogue method for dynamic planning provided by the methods described above. The method includes: acquiring the user's current input; determining the current dialogue target based on the current input according to a pre-built user profile database; dynamically planning a multi-turn dialogue action sequence according to the current dialogue target; converting the multi-turn dialogue action sequence into a text response, and ending the current dialogue.

[0087] The device embodiments described above are merely illustrative. 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 modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0088] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0089] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A dynamic programming-based intelligent dialogue method, characterized in that, include: Get the user's current input; Based on a pre-built user profile database, the current dialogue target is determined according to the current input; Based on the current dialogue objective, dynamically plan a multi-turn dialogue action sequence; The multi-turn dialogue action sequence is converted into a text response, and the current dialogue ends.

2. The intelligent dialogue method based on dynamic programming according to claim 1, characterized in that, The method of determining the current dialogue target based on the pre-built user profile database and the current input includes: Obtain the user profile database; Based on the large language model, and according to the current input and the user profile database, the user's professional guidance needs and emotional needs are obtained; Based on the stated professional guidance needs and the stated emotional needs, the current dialogue objective is determined.

3. The intelligent dialogue method based on dynamic programming according to claim 1, characterized in that, The step of dynamically planning a multi-turn dialogue action sequence based on the current dialogue goal includes: Determine the domain knowledge base corresponding to the current input; Based on the current dialogue objective and the domain knowledge base, the multi-turn dialogue action sequence is dynamically planned.

4. The intelligent dialogue method based on dynamic programming according to claim 3, characterized in that, The step of dynamically planning the multi-turn dialogue action sequence based on the current dialogue goal and the domain knowledge base includes: Based on the current dialogue objective, match the corresponding knowledge entities and semantic rules from the domain knowledge base; Based on the knowledge entities and the semantic rules, a set of candidate dialogue actions is constructed; A preset dynamic programming algorithm is used to perform path search on the candidate dialogue action set to generate the optimal multi-turn dialogue action sequence.

5. The intelligent dialogue method based on dynamic programming according to any one of claims 1-4, characterized in that, The dynamic programming multi-turn dialogue action sequence also includes: Monitor user feedback in real time and identify the type of feedback. If the user feedback is positive, then interact with the user according to the multi-turn dialogue action sequence; If the user feedback is negative, the multi-turn dialogue action sequence is dynamically replanned based on the current dialogue objective.

6. The intelligent dialogue method based on dynamic programming according to any one of claims 1-4, characterized in that, Also includes: During the dialogue, the user profile database is updated in real time; The user profile database includes users' historical dialogue records, historical dialogue action sequences, personal information profiles, and preference information.

7. A dynamic programming-based intelligent dialogue device, characterized in that, include: The current input acquisition module is used to acquire the user's current input. The current dialogue target determination module is used to determine the current dialogue target based on the current input, according to a pre-built user profile database. The dialogue action sequence planning module is used to dynamically plan a multi-turn dialogue action sequence based on the current dialogue goal; The text response conversion module is used to convert the multi-turn dialogue action sequence into a text response, at which point the current dialogue ends.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the intelligent dialogue method of dynamic programming as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the intelligent dialogue method of dynamic programming as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the intelligent dialogue method of dynamic programming as described in any one of claims 1 to 6.

Images