Response information generation method and device and related equipment

Abstract

The invention provides a response information generation method and device and related equipment, and relates to the technical field of artificial intelligence, and the method comprises the steps: obtaining a to-be-processed user question of a target user; performing statement analysis on the user question, and determining a service scene to which the user question belongs and a response template associated with the service scene; generating personalized response logic corresponding to the target user based on the historical session information of the target user and the response template; querying corresponding user service information from a pre-constructed user information base according to an information field required by the response template; and splicing the user question, the user service information, the personalized response logic and the historical session information into an input prompt, inputting the input prompt into a pre-trained response information generation model, and outputting response information for the user question. The accuracy of response information generation can be improved, and model illusion can be reduced.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of artificial intelligence, and in particular, to a response information generation method and device and related equipment. BACKGROUND

[0002] In the current customer service consultant consultation question and answer process, due to the randomness of user problems, the personalized difference of artificial sorting of excellent dialogue, and the single reply style. While using a large model to directly reply to dialogue hallucination rate is high, it is difficult to approach actual business data. In addition, due to the large number of consultation users, it is difficult for the consultant to associate related topics in historical replies when replying, resulting in deviation of product recommendations in dialogue from user needs, or repeated confirmation of business needs to the client in different conversations.

[0003] It should be noted that the information disclosed in the above background section is only used to strengthen the understanding of the background of the present disclosure, and therefore can include information that does not constitute prior art known to those of ordinary skill in the art. SUMMARY

[0004] The present disclosure provides a response information generation method, device and related equipment, which at least partially overcomes the problem of low accuracy of response information in related technologies.

[0005] Other characteristics and advantages of the present disclosure will become apparent from the following detailed description, or will be learned by practice of the present disclosure.

[0006] According to one aspect of the present disclosure, a response information generation method is provided, comprising: obtaining a user problem to be processed by a target user; performing sentence analysis on the user problem to determine a business scenario to which the user problem belongs and an answer template associated with the business scenario; generating a personalized answer logic corresponding to the target user based on historical conversation information of the target user and the answer template; querying corresponding user business information from a pre-constructed user information library according to the information field required by the answer template; concatenating the user problem, the user business information, the personalized answer logic, and the historical conversation information into an input prompt and inputting the input prompt into a pre-trained answer information generation model to output response information for the user problem.

[0007] In some exemplary embodiments of the present disclosure, based on the foregoing scheme, performing sentence analysis on the user problem to determine a business scenario to which the user problem belongs and an answer template associated with the business scenario comprises: determining a business scenario to which the user problem belongs based on the user problem through a pre-trained scene recognition model; and matching an answer template associated with the business scenario in a pre-set answer knowledge base according to the identified business scenario.

[0008] In some example embodiments of the present disclosure, based on the foregoing scheme, after obtaining the user question to be processed by the target user, the method further comprises: obtaining historical session information associated with the target user, wherein the historical session information comprises preference information and a historical topic chain of the target user.

[0009] In some example embodiments of the present disclosure, based on the foregoing scheme, obtaining the historical session information associated with the target user comprises: performing semantic feature extraction on historical interaction content of the target user to construct a historical topic chain.

[0010] In some example embodiments of the present disclosure, based on the foregoing scheme, performing feature extraction on the historical interaction content of the target user to construct a historical topic chain comprises: performing semantic feature extraction on the historical interaction content of the target user to determine historical interaction topics; and concatenating the historical interaction topics based on time sequence to construct a historical topic chain.

[0011] In some example embodiments of the present disclosure, based on the foregoing scheme, obtaining the historical session information associated with the target user comprises: inputting historical interaction content of the target user into a user portrait model to output preference information of the target user.

[0012] In some example embodiments of the present disclosure, based on the foregoing scheme, generating the personalized response logic corresponding to the target user based on the historical session information of the target user and the response template comprises: generating the personalized response logic corresponding to the target user according to the preference information of the target user and the response template.

[0013] In some example embodiments of the present disclosure, based on the foregoing scheme, concatenating the user question, the user business information, the personalized response logic, and the historical session information into an input prompt comprises: combining the user question, the user business information, the personalized response logic, and the historical topic chain into input information.

[0014] According to another aspect of the present disclosure, there is also provided an answer information generation apparatus, comprising: a user question acquisition module configured to acquire a user question to be processed by a target user; a user question analysis module configured to perform sentence analysis on the user question, determine a business scenario to which the user question belongs, and an answer template associated with the business scenario; a personalized answer logic generation module configured to generate a personalized answer logic corresponding to the target user based on historical conversation information of the target user and the answer template; a user business information query module configured to query corresponding user business information from a pre-constructed user information library according to information fields required by the answer template; and an answer information generation module configured to splice the user question, the user business information, the personalized answer logic, and the historical conversation information into an input prompt, input the input prompt into a pre-trained answer information generation model, and output answer information corresponding to the user question.

[0015] According to still another aspect of the present disclosure, there is also provided an electronic device, comprising: a processor; and a memory configured to store executable instructions of the processor; wherein the processor is configured to perform any of the above-mentioned answer information generation methods via execution of the executable instructions.

[0016] According to yet another aspect of the present disclosure, there is also provided a computer-readable storage medium having stored thereon a computer program, the computer program being executed by a processor to implement any of the above-mentioned answer information generation methods.

[0017] According to another aspect of the present disclosure, there is also provided a computer program product, comprising: a computer program or instructions, the computer program or instructions being executed by a processor to implement any of the above-mentioned answer information generation methods.

[0018] The response information generation method, device and related equipment provided in the embodiments of the present disclosure can improve the response accuracy and scene adaptability, avoid answering irrelevant questions, and ensure accurate understanding and classification of user intent. Moreover, the response template associated with the business scene is introduced into the generation process as a structured constraint, which can enhance the consistency and consistency of the reply, effectively prevent expression deviation, unauthorized commitment or information omission caused by free generation of large models. In addition, according to the historical conversation information of the target user, the language style, emotional tendency or historical interaction habit of the user can be understood, and the tone, degree of detail and question processing order of the reply can be automatically adjusted to improve the naturalness of interaction and user satisfaction. Further, by providing the historical conversation information of the user, the current question, the user business information and the personalized response logic as inputs to the response information generation model, the effect of realizing personalized and contextually coherent interaction can be achieved, which can continue the user conversation, avoid repeated explanation, improve user experience and conversation naturalness, and realize more intelligent, accurate and personalized customer service response.

[0019] It should be understood that the foregoing general description and the following detailed description are only exemplary and explanatory, and cannot limit the present disclosure. BRIEF DESCRIPTION OF DRAWINGS

[0020] The accompanying drawings, which are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present disclosure and, together with the specification, serve to explain the principles of the present disclosure. It is obvious that the drawings in the following description are only some embodiments of the present disclosure, and other drawings can be obtained by those skilled in the art without creative labor.

[0021] Figure 1 An exemplary application system architecture schematic diagram of a response information generation method in an embodiment of the present disclosure is shown. Figure 2 A schematic diagram of a response information generation method in an embodiment of the present disclosure is shown. Figure 3 A schematic diagram of a response information generation method flow in a package fee dispute scenario in an embodiment of the present disclosure is shown. Figure 4 A schematic diagram of a response information generation device in an embodiment of the present disclosure is shown. Figure 5 A detailed module schematic diagram of a response information generation device in an embodiment of the present disclosure is shown. Figure 6 An electronic device applying a response information generation method in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0022] Example implementations are now described in greater detail with reference to the figures. Like reference numerals can be used to refer to like elements throughout. The example implementations can be implemented in numerous ways, including as a process, an apparatus, a system, a method, or a computer readable medium such as a non-transitory computer readable medium containing instructions for implementation of at least one of these. In this specification, these implementations, or any variations of these implementations, can be referred to as "logic" or a "module" or a "circuit." This logic can comprise software or hardware, and can operate to implement one or more features described herein.

[0023] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more embodiments. In the following description, numerous specific details are provided to provide a thorough understanding of embodiments of the disclosure. One skilled in the relevant art will recognize, however, that the

[0024] The flow diagrams shown in the Figures should be understood to be illustrative only and not mandatory steps or sequences of steps. For example, some steps can be combined or omitted, and / or some steps can be performed at the same time or in a different order. Also, the flow diagrams can represent conceptual views of certain methods, but not other methods.

[0025] Figure 1 An exemplary application system architecture diagram to which the reply information generation method in embodiments of the present disclosure can be applied is shown. As shown, the system architecture can include a terminal device 101, a network 102, and a server 103. Figure 1

[0026] The network 102 is a medium for providing a communication link between the terminal device 101 and the server 103, and can be a wired network or a wireless network.

[0027] ​Optionally, the wireless or wired networks described above use standard communications technologies and / or protocols. The network typically provides Internet or Intranet connectivity, but can also be a combination of networks, including, but not limited to, a Local Area Network (LAN), a Metropolitan Area Network (MAN), a Wide Area Network (WAN), a mobile, wired or wireless network, a private network, or a virtual private network (VPN). In some embodiments, the data exchanged over the network is represented using technologies and / or formats including, but not limited to, Hypertext Markup Language (HTML), Extensible Markup Language (XML), etc. In addition, conventional encryption technologies such as the Secure Socket Layer (SSL), Transport Layer Security (TLS), Virtual Private Network (VPN), Internet Protocol Security (IPsec), etc. can be used to encrypt all or some links. In other embodiments, custom and / or proprietary data communication technologies can be used in place of or in combination with the above technologies.

[0028] The terminal device 101 can be various electronic devices, including but not limited to a smartphone, a tablet computer, a laptop computer, a desktop computer, a wearable device, an augmented reality device, a virtual reality device, etc.

[0029] Optionally, the clients of the application programs installed in different terminal devices 101 are the same, or the clients of the same type of application programs based on different operating systems. Based on the difference of terminal platforms, the specific forms of the clients of the application programs can also be different, for example, the application program client can be a mobile phone client, a PC client, etc.

[0030] The server 103 can be a server providing various services, for example, a background management server providing support for the device operated by the user using the terminal device 101. The background management server can analyze and process the received request data, etc., and feed back the processing result to the terminal device.

[0031] Optionally, the server can be a stand-alone physical server, or a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, CDN (Content Delivery Network), and basic cloud computing services such as big data and artificial intelligence platforms. The terminal can be a smart phone, a tablet computer, a notebook computer, a desktop computer, a smart speaker, a smart watch, and the like, but is not limited thereto. The terminal and the server can be connected directly or indirectly through wired or wireless communication, which is not limited in the present application.

[0032] Those skilled in the art can know that, Figure 1 The number of terminal devices, networks and servers in the above system architecture is only illustrative, and any number of terminal devices, networks and servers can be provided according to actual needs. The present disclosure embodiments are not limited thereto.

[0033] Under the above system architecture, the present disclosure embodiments provide a response information generation method, which can be executed by any electronic device with computing processing capability.

[0034] In some embodiments, the response information generation method provided in the present disclosure embodiments can be executed by the terminal device of the above system architecture; in other embodiments, the response information generation method provided in the present disclosure embodiments can be executed by the server in the above system architecture; in other embodiments, the response information generation method provided in the present disclosure embodiments can be realized by the terminal device and the server in the above system architecture through interaction.

[0035] Figure 2 A schematic diagram of a response information generation method in the present disclosure embodiments is shown, which includes the following steps: S202, obtaining a user question to be processed by a target user.

[0036] It should be noted that the target user in the present disclosure embodiments refers to a specific user individual who is currently interacting, needs to provide services or respond, and can be determined by a unique identifier such as a user account, a mobile phone number, a device ID, a session ID, etc. For example, a mobile phone user who is using an online customer service of an operator App; the user question in the present disclosure embodiments can be any question input by the target user that needs to be processed, and the target question can be in various forms such as text, voice-to-text, image description, etc. For example, the target user inputs “This month, how did it deduct 5 yuan in addition to the package” in the online customer service of an operator App.

[0037] S204, perform sentence analysis on the user question to determine a business scenario to which the user question belongs and a response template associated with the business scenario.

[0038] It should be noted that the sentence analysis in the embodiments of the present disclosure refers to parsing of a natural language sentence input by a user in terms of grammar, semantics and intent to extract structural features and core information thereof, and the purpose is to understand a user question input by a target user, which usually includes operations such as word segmentation processing, part-of-speech tagging, syntax dependency analysis, named entity recognition, intent recognition, etc. For example, the user question “This month, how did it deduct 5 yuan in addition to the package?” may be a sentence analysis result involving a fee dispute problem. The business scenario in the embodiments of the present disclosure refers to a specific service or business category to which a user question belongs, reflecting the real demand background of its consultation, complaint or handling. Common business scenarios include: package change consultation, bill fee dispute, traffic usage anomaly, service subscription application, network fault repair, etc. For example, according to the user question “This month, how did it deduct 5 yuan in addition to the package?”, the business scenario to which it belongs can be a package fee dispute. The response template in the embodiments of the present disclosure is a structured reply framework designed in advance for a specific business scenario, which can include fixed scripts, variable fields and logic instructions. For example, in the package fee dispute scenario, the response template can be as follows: Hello, regarding the

deduction item

deduction time

amount

reason description

processing action

[0039] S206, generate a personalized response logic corresponding to the target user based on the historical conversation information of the target user and the response template.

[0040] It should be noted that the historical conversation information in the embodiments of the present disclosure is the dialogue record and related derivative information generated by the target user in the past interaction process with the system, which can include the past N rounds of question and answer dialogue, user language style preference, occupation characteristics, etc. It can also include the historical topic chain related to business such as package handling, problem complaint, traffic package handling, etc. existing in the past N rounds of question and answer dialogue of the target user; the personalized response logic in the embodiments of the present disclosure is a dynamic generated adaptive reply strategy according to the individual characteristics (user preference information) of the target user and the structure requirements of the response template, which is used to guide the organization mode of the final script.

[0041] S208, query corresponding user business information from a pre-constructed user information library according to the information fields required by the response template.

[0042] It should be noted that the information field in the embodiment of the disclosure is a specific data item in the response template that needs to be dynamically filled, usually in the form of a placeholder, representing a certain type of business entity or user attribute, and these fields are the basis for driving data queries. For example, the template fields can include:

remaining traffic

package expiration date

whether to open international roaming

[0043] S210, splice the user question, user business information, personalized response logic and historical conversation information into an input prompt, input into a pre-trained response information generation model, and output response information for the user question.

[0044] It should be noted that the pre-trained response information generation model in the embodiment of the disclosure is a language model trained by large-scale dialogue data, which can generate natural language replies according to the input context, and is usually a large language model; the response information in the embodiment of the disclosure is the final natural language reply generated by the model, which aims to accurately and appropriately answer the user's questions. For example, according to the user question "How did this month's 5 yuan fee come out besides the package?", the response information generated by the response information generation model can be: "The 5 yuan fee you were charged this month is the 'call reminder' function fee. This service is opened by default, and we have noticed that you have also feedback this problem last month. Now we have unsubscribed it for you, and there will be no further deductions in the future."

[0045] The response information generation method provided in the embodiments of the present disclosure first acquires a user question to be processed by a target user; secondly, performs sentence analysis on the user question to determine a business scenario to which the user question belongs and a response template associated with the business scenario; then, generates personalized response logic corresponding to the target user based on historical conversation information of the target user and the response template; then, queries corresponding user business information from a pre-constructed user information library according to information fields required by the response template; finally, concatenates the user question, the user business information, the personalized response logic, and the historical conversation information into an input prompt, inputs the input prompt into a pre-trained response information generation model, and outputs response information for the user question. Compared with the problem of low accuracy of response information in the related art, the embodiments of the present disclosure improve the accuracy of response and the effect of scene adaptation by performing sentence analysis on the received user question, determining the business scenario to which the user question belongs, and matching the corresponding response template, which can avoid answering the wrong question, ensure accurate understanding and classification of user intent, and has the effect of enhancing the consistency and consistency of the reply structure. Deviation, over-commitment or information omission caused by free generation of a large model is effectively prevented. In addition, the language style, emotional tendency or historical interaction habit of the user can be understood according to the historical conversation information of the target user, and the tone, degree of detail and problem processing order of the reply are automatically adjusted to improve the naturalness of interaction and user satisfaction. Further, by providing the historical conversation information of the user, the current question, the user business information, and the personalized response logic to the response information generation model as inputs, the effect of realizing personalized and contextually coherent interaction is achieved, which can continue the user conversation, avoid repeated explanation, improve user experience and conversation naturalness, and realize more intelligent, accurate and personalized customer service response.

[0046] In some embodiments, the embodiments of the present disclosure perform sentence analysis on the user question to determine the business scenario to which the user question belongs and the response template associated with the business scenario, including: based on the user question, determining the business scenario to which the user question belongs through a pre-trained scene recognition model; and according to the identified business scenario, matching the response template associated with the business scenario in a pre-set response knowledge base. Specifically, based on the user question, the pre-trained scene recognition model is used to determine the business scenario to which the user question belongs, which can map the user's diversified natural language expression to a standardized business category, avoid service misjudgment caused by semantic ambiguity, ensure that different users obtain standardized dialogue responses in the same scenario, and improve service professionalism.

[0047] In some embodiments, after obtaining the user question to be processed by the target user, the response information generation method in the embodiments of the present disclosure further includes: obtaining historical conversation information associated with the target user, wherein the historical conversation information includes preference information and a historical topic chain of the target user. Specifically, after obtaining the user question, the historical conversation information associated with the user question is further obtained, which has the effect of enhancing the contextual understanding capability, can generate coherent and natural responses based on the past interaction background of the user, and avoids the fragmented feeling of starting from zero for each dialogue; by obtaining the historical conversation data containing the user preference information, the effect of realizing style adaptation is achieved, for example, refined language is output for users who prefer concise expression, and complete explanation is provided for users who need detailed explanation, thereby improving the information transmission efficiency; further, by establishing a dual-channel input mechanism of “current input + historical memory”, the effect of enhancing the controllability of the large model is achieved, which ensures that the reply content conforms to the historical behavior mode of the user while exerting the generation capability, thereby reducing the risk of generating illusion or inconsistent content.

[0048] In some embodiments, the embodiments of the present disclosure obtain the historical conversation information associated with the target user, including: performing semantic feature extraction on the historical interaction content of the target user, and constructing a historical topic chain. Specifically, the historical topic chain representing the evolution relationship of the interaction theme is constructed, which has the effect of realizing cross-session context memory, can identify the logical association between the current question and the past interaction, and improves the coherence and intelligence level of the dialogue; and by establishing a topic chain memory mechanism dedicated to the user, the effect of supporting long-term personalized service is achieved, for example, identifying high-frequency question types, predicting potential demands, and actively pushing solutions, realizing the evolution from passive response to active service.

[0049] In some embodiments, the embodiments of the present disclosure perform feature extraction on the historical interaction content of the target user to construct a historical topic chain, including: performing semantic feature extraction on the historical interaction content of the target user to determine historical interaction themes; and splicing the historical interaction themes based on time sequence to construct a historical topic chain. Specifically, by performing semantic feature extraction on the historical interaction content of the target user, the user's attention theme and intention evolution path can be accurately identified, and the computable and organized semantic information can be extracted from unstructured dialogue text to provide structured support for context understanding; further, by taking the topic chain as a dynamically updated context input, the effect of reducing the illusion and logical conflict of the large model is achieved, which provides real and traceable user behavior basis for the generative model, and improves the accuracy and consistency of the output content.

[0050] In some embodiments, the disclosure embodiment obtains historical conversation information associated with a target user, including: inputting historical interaction content of the target user into a user portrait model, and outputting preference information of the target user. Specifically, by inputting the historical interaction content of the target user into the user portrait model, the preference information of the target user is extracted, which can realize automatic identification of user individual characteristics, and can mine behavior patterns that can guide response generation from unstructured conversation data; moreover, by constructing a dynamic user portrait based on historical interaction content, the service adaptability can be improved, so that the system can generate customized replies according to the language style, information acceptance habit and emotional tendency of the user, avoid one-size-fits-all general expressions, ensure that the output conforms to individual characteristics while exerting the generation capability, improve information transmission efficiency, and significantly improve the interactive experience.

[0051] In some embodiments, the disclosure embodiment proposes a dynamic memory mechanism for different users, which can effectively solve the problem of redundancy of historical conversation information and sparseness of effective information. The disclosure embodiment can also use a large model to dynamically depict user language style preferences, professions and other characteristics; and construct a historical topic chain related to business such as summary generation package handling, problem complaint, and traffic package handling for the current conversation, to improve the accuracy of conversational replies.

[0052] In some embodiments, the disclosure embodiment generates personalized response logic corresponding to a target user based on historical conversation information of the target user and a response template, including: generating personalized response logic corresponding to the target user according to preference information of the target user and the response template. Specifically, by generating personalized response logic according to the preference information of the target user and the response template, formal, concise or colloquial reply strategies can be automatically generated for different users, improving the naturalness and satisfaction of interaction.

[0053] In some embodiments, the disclosure embodiment concatenates user questions, user business information, personalized response logic and historical conversation information as input prompts, including: combining user questions, user business information, personalized response logic and historical topic chains into input information. Specifically, the disclosure embodiment queries corresponding user business information in the user information library based on the business information field required by the response template, which has the effect of ensuring the accuracy and consistency of the response content; further, by combining user questions, user business information, personalized response logic and historical topic chains into input information, the large model is provided with comprehensive and structured generation basis, which significantly improves the relevance and completeness of the output content.

[0054] In some embodiments, the embodiments of the present disclosure are also directed to the characteristics that cannot be eliminated by large model hallucination. By artificially organizing the preliminary optimal response knowledge base, the business scenario recognition and small model recall sorting method is used to recommend the response templates related to the current session user question, and the large model is used to supplement the missing information in the template and output in combination with the user business information, which effectively reduces the hallucination in the large model response information generation process.

[0055] In some embodiments, as Figure 3 shown, the embodiments of the present disclosure take the out-of-package fee dispute scene as an example to show the overall flow of the response information generation method, which specifically includes: S302, receiving the user question "How did this month deduct 5 yuan in addition to the package?" input by the current session user through the front-end window.

[0056] S304, business scenario recognition. The scenario recognition model is constructed by small model training to identify the business scenario to which the user's current question belongs, for example, the example identifies the business scenario to which the user's question belongs as: out-of-package fee dispute.

[0057] S306, recommend the response template of the current business scenario. From the pre-set response knowledge base (for example, the excellent dialogue template set composed of artificial writing constitutes the response knowledge base), use bm25, reordering and other model matching to match all response templates suitable for the current business scenario, and in the example, the response templates for the project such as voice and traffic exceeding the package are matched.

[0058] S308, generate personalized response template generation and arrangement method. Based on the user's preference information and the business information fields required by the response template, generate the personalized response logic of the response template.

[0059] S310, user business information extraction. Through the field association of the response template and the user information table metadata, use the large model to extract the user business information associated with the template information from the user information table.

[0060] S312, response information prompt engineering construction and response information generation. Through the prompt engineering, the personalized response logic, user business information, user dynamic memory history topic chain and user input user question are integrated together and input into the pre-trained response information generation model to generate the response information for the user question.

[0061] S314, user dynamic memory generation. After the session ends, input the user question and the output response information of the current session into the user portrait large model, extract new user portrait information from the user portrait large model, and construct a new topic chain.

[0062] In some embodiments, the embodiments of the present disclosure generate dynamic memories of the user by constructing a topic chain, and improve the accuracy of the generation of response information and reduce the model hallucination by recommending a response template in the manner of retrieval-augmented generation (RAG). In addition, the response information generated by combining the user business information is more targeted to the current user in the aspect of the reply.

[0063] Based on the same inventive concept, the embodiments of the present disclosure also provide a response information generation device, as follows. Since the principles of the embodiments of the device solve the problem similar to the above-mentioned method embodiments, the implementation of the embodiments of the device can be referred to the implementation of the above-mentioned method embodiments, and the repeated parts will not be described here.

[0064] Figure 4 A schematic diagram of a response information generation device in an embodiment of the present disclosure is shown, and the device includes: A user question acquisition module 401 is configured to acquire a user question to be processed by a target user; A user question analysis module 402 is configured to perform sentence analysis on the user question, determine a business scenario to which the user question belongs, and a response template associated with the business scenario; A personalized response logic generation module 403 is configured to generate a personalized response logic corresponding to the target user based on historical conversation information of the target user and the response template; A user business information query module 404 is configured to query corresponding user business information from a pre-constructed user information library according to an information field required by the response template; A response information generation module 405 is configured to concatenate the user question, the user business information, the personalized response logic, and the historical conversation information into an input prompt, input the input prompt into a pre-trained response information generation model, and output response information for the user question.

[0065] The response information generation apparatus provided in the embodiments of the present disclosure includes: a user question acquisition module, configured to acquire a user question to be processed by a target user; a user question analysis module, configured to perform sentence analysis on the user question, determine a business scenario to which the user question belongs, and a response template associated with the business scenario; a personalized response logic generation module, configured to generate personalized response logic corresponding to the target user based on historical conversation information of the target user and the response template; a user business information query module, configured to query corresponding user business information from a pre-constructed user information library according to information fields required by the response template; and a response information generation module, configured to splice the user question, the user business information, the personalized response logic, and the historical conversation information into an input prompt, input the input prompt into a pre-trained response information generation model, and output response information for the user question. Compared with the related art, the embodiments of the present disclosure can improve the response accuracy and the scene adaptability by performing sentence analysis on the received user question, determining the business scenario to which the user question belongs, and matching the corresponding response template, can avoid answering the wrong question, and ensure accurate understanding and classification of the user intent. Moreover, the response template associated with the business scenario is introduced into the generation process as a structured constraint, which can enhance the consistency and consistency of the reply, effectively prevent the expression deviation, unauthorized commitment, or information omission caused by the free generation of the large model. In addition, the historical conversation information of the target user can be used to understand the language style, emotional tendency, or historical interaction habit of the user, automatically adjust the tone, degree of detail, and problem processing order of the reply, and improve the naturalness of the interaction and the user satisfaction. Furthermore, the historical conversation information of the user, the current question, the user business information, and the personalized response logic are collectively provided as input to the response information generation model, which can realize personalized and contextually coherent interaction, continue the user conversation, avoid repeated explanation, improve the user experience and the naturalness of the conversation, and realize more intelligent, accurate, and personalized customer service response.

[0066] In some embodiments, as shown in FIG. 1, Figure 5 The response information generation apparatus in the embodiments of the present disclosure further includes: a scene identification module, configured to determine the business scenario to which the user question belongs based on the user question by using a pre-trained scene identification model, which can include specific business scenarios such as mobile phone failure, broadband failure, package or traffic fee dispute, etc.; and a response template recommendation module, configured to match the response template associated with the business scenario from a pre-set response knowledge base according to the identified business scenario. Further, the recommended response template can include multiple templates, which can form a response template recommendation list.

[0067] In some embodiments, as shown in FIG. 1, Figure 5As shown, the response information generation apparatus in the embodiment of the present disclosure further includes: a user dynamic memory module, configured to acquire historical conversation information associated with the target user after acquiring the user question to be processed by the target user, wherein the historical conversation information includes preference information and a historical topic chain of the target user.

[0068] In some embodiments, the user dynamic memory module in the embodiment of the present disclosure is further configured to perform semantic feature extraction on the historical interaction content of the target user, and construct the historical topic chain.

[0069] In some embodiments, the user dynamic memory module in the embodiment of the present disclosure is further configured to perform semantic feature extraction on the historical interaction content of the target user, determine a historical interaction theme, splice the historical interaction theme based on a time sequence, and construct the historical topic chain.

[0070] In some embodiments, the user dynamic memory module in the embodiment of the present disclosure is further configured to input the historical interaction content of the target user into a user portrait model, and output the preference information of the target user.

[0071] In some embodiments, as shown, Figure 5 The personalized response logic generation module in the embodiment of the present disclosure is further configured to generate the personalized response logic corresponding to the target user according to the preference information of the target user and the response template.

[0072] In some embodiments, as shown, Figure 5 The response information generation module in the embodiment of the present disclosure is further configured to combine the user question, the user business information, the personalized response logic, and the historical topic chain into input information.

[0073] Those skilled in the art can understand that each aspect of the present disclosure can be implemented as a system, a method or a program product. Therefore, each aspect of the present disclosure can be embodied as a complete hardware embodiment, a complete software embodiment (including firmware, microcode, etc.), or an embodiment combining hardware and software aspects, which can be collectively referred to as "circuitry", "module" or "system".

[0074] Based on the same inventive concept, the present disclosure further provides an electronic device, which includes: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to execute the response information generation method of any one of the above aspects by executing the executable instructions. Since the principle of solving problems of the electronic device embodiment is similar to that of the above method embodiment, the implementation of the electronic device embodiment can be referred to the implementation of the above method embodiment, and the repeated parts will not be described here.

[0075] The following refers to Figure 6The electronic device 600 according to this implementation of the present disclosure is described. Figure 6 The electronic device 600 shown is merely one example. It should be appreciated that the functions and uses of the electronic device 600 are not limited to the examples described herein.

[0076] As shown in Figure 6 The electronic device 600 is in the form of a general computing device. The components of the electronic device 600 can include, but are not limited to, the at least one processing unit 601 described above, the at least one storage unit 602 described above, and a bus 603 that connects different system components, including the storage unit 602 and the processing unit 601.

[0077] The storage unit stores program code that can be executed by the processing unit 601, so that the processing unit 601 performs the steps described in the "Exemplary Method" section above according to various exemplary implementations of the present disclosure.

[0078] In some embodiments, when the electronic device is used to control, for example, the answer information generation method described above, the processing unit 601 can perform the following steps of the method embodiments described above: Obtaining a user question to be processed by a target user; performing sentence analysis on the user question to determine a business scenario to which the user question belongs and an answer template associated with the business scenario; generating personalized answer logic corresponding to the target user based on historical conversation information of the target user and the answer template; querying corresponding user business information from a pre-constructed user information library according to information fields required by the answer template; concatenating the user question, the user business information, the personalized answer logic, and the historical conversation information into an input prompt and inputting the input prompt into a pre-trained answer information generation model to output answer information for the user question.

[0079] The storage unit 602 can include a readable medium in the form of a volatile storage unit, such as a random access memory (RAM) 6021 and / or a cache memory 6022, and can further include a read-only memory (ROM) 6023.

[0080] The storage unit 602 can further include a program / utility 6024 having a set of program modules 6025, including but not limited to, an operating system, one or more application programs, other program modules, and program data, each of which can include an implementation of a network environment, or some combination thereof.

[0081] The bus 603 can represent one or more of several types of bus structures, including a storage unit bus or storage unit controller, a peripheral bus, a graphics acceleration port, a processing unit bus, or a local bus using any of a variety of bus architectures.

[0082] The electronic device 600 can also communicate with one or more external devices 604 such as a keyboard or pointing devices, a Bluetooth device, or a database, etc.; and / or one or more devices that enable a user to interact with the electronic device 600; and / or any devices (e.g., a router, a modem, a server, etc.) that enable the electronic device 600 to communicate with one or more other computing devices. Such communication can occur via an input / output (I / O) interface 605. Still yet, the electronic device 600 can communicate with one or more networks, such as a local area network (LAN), a general wide area network (WAN), and / or the Internet, etc., through a network adapter 606. As depicted, the network adapter 606 communicates with the other components of the electronic device 600 via the bus 603. It should be appreciated that the bus 603 can be one of any suitable type, and that the components of the electronic device 600 can be implemented using one or more of any suitable type of hardware and / or software, including but not limited to microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data archival storage systems, etc.

[0083] Through the description of the above embodiments, those skilled in the art can easily understand that the example embodiments described herein can be implemented by software, or by software in combination with necessary hardware. Therefore, the technical solutions according to the embodiments of the present disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash disk, a mobile hard disk, etc.) or a network, and includes a number of instructions to enable a computing device (which can be a personal computer, a server, a terminal device, or a network device, etc.) to perform the methods according to the embodiments of the present disclosure.

[0084] Based on the same inventive concept, the present disclosure also provides a computer readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the response information generation method of any one of the above. Since the problem solving principle of the computer readable storage medium embodiment is similar to that of the above method embodiment, the implementation of the computer readable storage medium embodiment can be referred to the implementation of the above method embodiment, and the repeated parts will not be described here.

[0085] More specific examples of the computer readable storage medium in the present disclosure can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0086] In this disclosure, a computer readable storage medium can include a data signal transporting or broadcasting computer readable program code embodied in the computer readable storage medium. The computer readable storage medium can also be any computer readable medium that can be linked to or transfeπed to by a computer. In some embodiments, the computer readable storage medium includes storage media such as a hard disk drive, a floppy diskette, a CD-ROM, a DVD, a Blu-ray Disc, a magnetic tape, a silicon memory, or any other computer readable storage medium which can be used to store and / or transfer data or computer readable program code.

[0087] Optionally, program code embodied on a computer readable storage medium can be transmitted by any data transmission techniques, including but not limited to radio frequency, wireless, wire line, optical cable, Rf, etc., or any suitable combination of the foregoing.

[0088] In some embodiments, program code can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C++, or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computing device, partly on the user's computing device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device or entirely on the remote computing device or server. In the latter scenario, the remote computing device can be connected to the user's computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computing device, such as through the Internet using an Internet Service Provider (ISP).

[0089] Based on the same inventive concept, the disclosure also provides a computer program product, including a computer program product, including: a computer program or instructions, the computer program or instructions are executed by a processor to implement the response information generation method of any one of the above method embodiments. Since the problem solving principle of the computer program product embodiment is similar to the above method embodiments, the implementation of the computer program product embodiment can be referred to the implementation of the above method embodiments, and the repeated parts will not be described here.

[0090] It should be noted that although several modules or units of the device for action execution are mentioned in the above detailed description, such division is not mandatory. In fact, according to the embodiments of the disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided into a plurality of modules or units.

[0091] Moreover, although individual steps of the methods in the present disclosure are described in a particular order in the drawings, this is not required or implied, nor is it necessary to perform all of the steps shown to achieve the desired result. Additionally or alternatively, certain steps can be omitted, combined into a single step, performed in a different order, broken down into multiple steps, and / or the like.

[0092] Those skilled in the art will readily understand that the example embodiments described herein can be implemented by software and / or by hardware coupled with software, as described above. Thus, the technical solutions according to the embodiments of the present disclosure can be embodied in the form of a software product. The software product can be stored in a non-volatile storage medium (which can be a CD-ROM, a USB flash disk, a mobile hard disk, or the like) or on a network, and includes a number of instructions to enable a computing device (which can be a personal computer, a server, a mobile terminal, or a network device, etc.) to perform the methods according to the embodiments of the present disclosure.

[0093] Other embodiments of the present disclosure will be apparent to those skilled in the art from consideration of the specification and practice of the application disclosed herein. The present disclosure is intended to cover any variations, uses, or adaptations of the present disclosure following the general principles thereof and including such departures from the present disclosure that come within known use or custom in the art to which the present disclosure pertains. The specification and examples are to be regarded as illustrative only, and the true scope and spirit of the present disclosure are indicated by the appended claims.

Claims

1. A method for generating response information, characterized in that, include: Obtain the user issues pending processing from the target user; The user question is analyzed to determine the business scenario to which the user question belongs, and the response template associated with the business scenario; Based on the target user's historical session information and the response template, a personalized response logic corresponding to the target user is generated; Based on the information fields required by the response template, query the corresponding user business information from the pre-built user information database; The user question, the user business information, the personalized response logic, and the historical session information are concatenated into an input prompt, which is then input into a pre-trained response information generation model to output response information for the user question.

2. The response information generation method according to claim 1, characterized in that, The user question is analyzed to determine the business scenario to which it belongs, and the response template associated with that business scenario, including: Based on the user's question, the business scenario to which the user's question belongs is determined through a pre-trained scenario recognition model; Based on the identified business scenario, a response template associated with the business scenario is matched from a pre-set response knowledge base.

3. The response information generation method according to claim 1, characterized in that, After obtaining the user issues to be processed by the target user, the method further includes: Obtain historical session information associated with the target user, wherein the historical session information includes the target user's preference information and historical topic chains.

4. The response information generation method according to claim 3, characterized in that, Obtaining historical session information associated with the target user includes: Semantic features are extracted from the target user's historical interaction content to construct a historical topic chain.

5. The response information generation method according to claim 4, characterized in that, Feature extraction is performed on the target user's historical interaction content to construct a historical topic chain, including: Semantic features are extracted from the target user's historical interaction content to determine the historical interaction topics; The historical interaction topics are pieced together in chronological order to construct a historical topic chain.

6. The response information generation method according to claim 3, characterized in that, Obtaining historical session information associated with the target user includes: The target user's historical interaction content is input into the user profile model, and the target user's preference information is output.

7. The response information generation method according to claim 3, characterized in that, Based on the target user's historical session information and the response template, a personalized response logic is generated for the target user, including: Based on the target user's preference information and the response template, a personalized response logic is generated for the target user.

8. The response information generation method according to claim 3, characterized in that, The input prompt is constructed by concatenating the user question, the user service information, the personalized response logic, and the historical session information, including: The user question, the user business information, the personalized response logic, and the historical topic chain are combined into input information.

9. A response information generation device, characterized in that, include: The user issue acquisition module is used to acquire user issues that need to be processed by the target user. The user question analysis module is used to perform sentence analysis on the user questions, determine the business scenario to which the user questions belong, and the response template associated with the business scenario; A personalized response logic generation module is used to generate personalized response logic corresponding to the target user based on the target user's historical session information and the response template. The user business information query module is used to query the corresponding user business information from a pre-built user information database based on the information fields required by the response template. The response information generation module is used to concatenate the user question, the user business information, the personalized response logic, and the historical session information into an input prompt, input it into a pre-trained response information generation model, and output response information for the user question.

10. An electronic device, characterized in that, include: processor; as well as Memory for storing the executable instructions of the processor; The processor is configured to execute the response information generation method according to any one of claims 1 to 8 by executing the executable instructions.

11. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the response information generation method according to any one of claims 1 to 8.

12. A computer program product, comprising: A computer program or instruction, characterized in that, when executed by a processor, the computer program or instruction implements the response information generation method according to any one of claims 1 to 8.

Images