Typical skill recommendation method applied to seat questions and answers and related equipment

Abstract

The invention belongs to the technical field of artificial intelligence, and relates to a verbal skill recommendation method and related equipment applied to seat questioning and answering, and the method comprises the following steps: carrying out structured analysis operation on preposed information according to an NLP semantic understanding technology to obtain key feature information; performing vector conversion operation on the key feature information to obtain a key feature vector; reading a system database, and obtaining effective business behavior data in the system database; performing fine adjustment on the GPT pre-training language model according to the effective business behavior data and the finance and taxation knowledge system to obtain a trained GPT language model; inputting the key feature information into a trained GPT language model for portrait analysis operation to obtain user portrait information; generating a recommended verbal skill corresponding to the user portrait according to a dialogue management technology; and outputting the recommended verbal skills. According to the method and the device, a more comprehensive and accurate customer portrait can be generated, and a recommendation verbal skill with higher adaptability is provided according to customer portrait analysis and agent exhibition data.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to a method for recommending speech techniques for agent question and answer and related equipment. Background Art

[0002] Customer service script recommendation means that during the service process provided by manual or electronic customer service, the recommendation system intelligently retrieves customer service response scripts from historical customer service scripts based on the context of the current conversation, allowing manual or electronic customer service to quickly select the recommended script to continue replying, thereby improving customer service response efficiency.

[0003] There is a customer service script recommendation solution, which is a personalized information recommendation system that recommends information and products of interest to users based on their information needs and interests. Most of the current information recommendation systems are based on generative large language models, which can generate human-like responses.

[0004] However, the applicant found that traditional customer service script recommendation solutions often have incomplete user portraits, resulting in poor quality of recommended scripts, which affects the user consultation experience. Summary of the invention

[0005] The purpose of the embodiments of the present application is to propose a method and related equipment for recommending speech techniques for agent question and answer, so as to solve the problem of poor quality of recommended speech techniques in traditional customer service speech recommendation schemes.

[0006] In order to solve the above technical problems, the present application embodiment provides a method for recommending speech techniques for agent question and answer, which adopts the following technical solution:

[0007] Receiving pre-information input by the agent terminal;

[0008] Performing structured analysis on the pre-information according to NLP semantic understanding technology to obtain key feature information;

[0009] Performing a vector conversion operation on the key feature information to obtain a key feature vector;

[0010] Reading a system database, and obtaining valid business behavior data in the system database;

[0011] Fine-tune the GPT pre-trained language model according to the effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model;

[0012] Inputting the key feature information into the trained GPT language model to perform a portrait analysis operation to obtain user portrait information;

[0013] Generate a recommendation script corresponding to the user profile based on the dialogue management technology;

[0014] Output the recommended words.

[0015] Furthermore, before the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology, the following steps are also included:

[0016] Performing sentiment analysis on the pre-processed information according to sentiment analysis technology to obtain the user's emotional state;

[0017] The step of generating a recommended speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps:

[0018] Generate recommendation words corresponding to the user portrait and the user's emotional state based on the dialogue management technology.

[0019] Furthermore, the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps:

[0020] In the process of speech text generation, the similarity between the real-time speech text content and the professional knowledge in the financial and tax knowledge graph is calculated in real time;

[0021] Correct the real-time speech text content whose similarity is lower than the preset similarity threshold;

[0022] When all the scripts are generated, you will get real-time recommended scripts.

[0023] Furthermore, the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps:

[0024] Generate an initial speech text corresponding to the user portrait according to the dialogue management technology;

[0025] The initial speech text is optimized according to the attention mechanism and the key feature information to obtain the recommended speech.

[0026] Furthermore, after the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology, the following steps are also included:

[0027] Calculate the user acceptance of the recommended words according to the sentiment analysis model;

[0028] If the user acceptance is lower than a preset acceptance threshold, a recommendation speech corresponding to the user portrait is regenerated according to the dialogue management technology until the user acceptance of the generated recommendation speech meets the acceptance threshold.

[0029] Furthermore, after the step of outputting the recommended words, the method further includes the following steps:

[0030] Receiving user evaluation data corresponding to the recommended speech;

[0031] The parameter weights of the trained GPT language model are updated according to the user evaluation data.

[0032] In order to solve the above technical problems, the embodiment of the present application further provides a speech recommendation device for agent question and answer, which adopts the following technical solution:

[0033] A pre-information acquisition module, used for receiving pre-information input by the proxy terminal;

[0034] A structured analysis module, used to perform a structured analysis operation on the pre-information according to NLP semantic understanding technology to obtain key feature information;

[0035] A vector conversion module, used to perform a vector conversion operation on the key feature information to obtain a key feature vector;

[0036] A behavior data acquisition module, used to read a system database and acquire valid business behavior data from the system database;

[0037] A model fine-tuning module is used to fine-tune the GPT pre-trained language model according to the effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model;

[0038] A portrait analysis module is used to input the key feature information into the trained GPT language model to perform a portrait analysis operation to obtain user portrait information;

[0039] A recommended speech acquisition module, used to generate a recommended speech corresponding to the user portrait based on the dialogue management technology;

[0040] The recommended words output module is used to output the recommended words.

[0041] Furthermore, the device further comprises: a sentiment analysis module, and the recommended speech acquisition module further comprises: a recommended speech acquisition submodule, wherein:

[0042] The sentiment analysis module is used to perform sentiment analysis on the pre-information according to sentiment analysis technology to obtain the user's emotional state;

[0043] The recommended speech acquisition submodule is used to generate recommended speech corresponding to the user portrait and the user's emotional state based on the dialogue management technology.

[0044] In order to solve the above technical problems, the embodiment of the present application further provides a computer device, which adopts the following technical solution:

[0045] It includes a memory and a processor, wherein the memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for recommending speech techniques applied to agent question and answer are implemented as described above.

[0046] In order to solve the above technical problems, the embodiment of the present application further provides a computer-readable storage medium, which adopts the following technical solution:

[0047] The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for recommending speech techniques for agent question and answer as described above.

[0048] The present application provides a method for recommending speech applied to agent question and answer, including: receiving pre-information input by an agent terminal; performing a structured analysis operation on the pre-information according to NLP semantic understanding technology to obtain key feature information; performing a vector conversion operation on the key feature information to obtain a key feature vector; reading a system database, and obtaining effective business behavior data in the system database; fine-tuning a GPT pre-trained language model according to the effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model; inputting the key feature information into the trained GPT language model for a portrait analysis operation to obtain user portrait information; generating a recommended speech corresponding to the user portrait according to a dialogue management technology; and outputting the recommended speech. Compared with the prior art, the present application can generate a more comprehensive and accurate customer portrait by performing semantic understanding and key feature extraction on the information provided by the agent. It is based on various data obtained from different channels and is trained by applying a machine learning model to help agents gain an in-depth understanding of potential customers and their needs; using an AI model to comprehensively analyze the data and mine hidden laws and patterns, providing agents with more reliable and targeted sales skills recommendations. This can greatly save time and energy, and enhance the sales team's real-time reference to make wise decisions; combined with the financial and tax knowledge system, it ensures that the generated recommendations have the perfection of retrieval resources inside and outside the industry, and has auxiliary functions when handling complex cases; based on customer portrait analysis and agent business development data, the system can provide specific guidance on communication tone, logic and progress. This can help agents develop personalized and effective marketing strategies and adjust them in real time to meet customer needs. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In order to more clearly illustrate the scheme in the present application, a brief introduction is given below to the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0050] Figure 1 is an exemplary system architecture diagram to which the present application may be applied;

[0051] Figure 2 It is a flowchart of the implementation of the method for recommending speech techniques for agent question and answer provided in an embodiment of the present application;

[0052] Figure 3 : is a structural diagram of a speech recommendation device for agent question and answer provided in an embodiment of the present application;

[0053] Figure 4 It is a structural diagram of an embodiment of a computer device according to the present application. DETAILED DESCRIPTION

[0054] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by technicians in the technical field of the present application; the terms used in the specification of the application herein are only for the purpose of describing specific embodiments and are not intended to limit the present application; the terms "including" and "having" and any variations thereof in the specification and claims of the present application and the above-mentioned drawings are intended to cover non-exclusive inclusions. The terms "first", "second", etc. in the specification and claims of the present application or the above-mentioned drawings are used to distinguish different objects, not to describe a specific order.

[0055] Reference to "embodiments" herein means that a particular feature, structure, or characteristic described in conjunction with the embodiments may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those skilled in the art that the embodiments described herein may be combined with other embodiments.

[0056] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the accompanying drawings.

[0057] like Figure 1As shown, the system architecture 100 may include a terminal device 101, a network 102 and a server 103. The terminal device 101 may be a laptop 1011, a tablet computer 1012 or a mobile phone 1013. The network 102 is used to provide a medium for a communication link between the terminal device 101 and the server 103. The network 102 may include various connection types, such as wired, wireless communication links or optical fiber cables, etc.

[0058] The user can use the terminal device 101 to interact with the server 103 through the network 102 to receive or send messages, etc. Various communication client applications can be installed on the terminal device 101, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social platform software, etc.

[0059] The terminal device 101 can be any electronic device with a display screen and supporting web browsing. In addition to the laptop computer 1011, tablet computer 1012 or mobile phone 1013, the terminal device 101 can also be an e-book reader, an MP3 player (Moving Picture Experts Group Audio Layer III), an MP4 (Moving Picture Experts Group Audio Layer IV), a laptop computer, a desktop computer, etc.

[0060] The server 103 may be a server that provides various services, such as a background server that provides support for a web page displayed on the terminal device 101 .

[0061] It should be noted that the speech recommendation method applied to the agent question and answer provided in the embodiment of the present application is generally executed by a server / terminal device, and accordingly, the speech recommendation device applied to the agent question and answer is generally set in the server / terminal device.

[0062] It should be understood that Figure 1 The number of terminal devices, networks and servers in the embodiment is only for illustration. Any number of terminal devices, networks and servers may be provided according to the implementation requirements.

[0063] Continue to refer Figure 2 , shows a flow chart of an embodiment of a method for recommending speech techniques for agent question and answer according to the present application. The method for recommending speech techniques for agent question and answer comprises: step S201, step S202, step S203, step S204, step S205, step S206, step S207 and step S208.

[0064] In step S201, the pre-information input by the agent terminal is received.

[0065] In an embodiment of the present application, when an agent encounters a high-net-worth potential client, he is preparing to do business or has already started doing business, but has encountered certain bottlenecks and conducts system consultation. The system will invite the agent to input prior information to understand the current situation of the client and the agent as much as possible.

[0066] In step S202, a structured analysis operation is performed on the pre-information according to the NLP semantic understanding technology to obtain key feature information.

[0067] In the embodiment of the present application, the information provided by the agent is multiple narrative content or chat records. The system first cleans and standardizes these unstructured data, including removing stop words, stemming, part-of-speech tagging, etc., and analyzes and extracts key features through NLP semantic understanding, including but not limited to:

[0068] a) Customer sources, customer relationships, customer characteristics, and customer information disclosure willingness;

[0069] b) Basic information of the customer, such as personal information, asset status, marriage and family, and corporate data;

[0070] c) Explicit customer needs, such as the agent’s own analysis and inference, and the customer’s direct explanation.

[0071] In step S203, a vector conversion operation is performed on the key feature information to obtain a key feature vector.

[0072] In the embodiments of the present application, topic modeling techniques such as LDA are applied to extract topic features, and TF-IDF or Word2Vec and other methods are used to convert text into numerical feature vectors.

[0073] In step S204, the system database is read to obtain valid business behavior data from the system database.

[0074] In step S205, the GPT pre-trained language model is fine-tuned according to the effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model.

[0075] In the embodiment of the present application, based on the behavioral data of high-end customer transactions obtained from different channels, including customer interaction records, sales records, marketing activity history, etc., combined with the financial and tax knowledge system to ensure the correctness of the model understanding, GPT deep learning technology is used to comprehensively understand these data. By mining the hidden rules and patterns in these data, an AI model with strong adaptability and high ability to accurately predict results is established. This model undertakes the main work of professional consulting analysis.

[0076] In the embodiments of the present application, the K-means clustering algorithm is used to group customer characteristics and construct customer segmentation; the association rule mining technology is used to discover the potential connections between customer characteristics, and the hidden rules and patterns are mined from the behavioral data to establish an AI model with strong adaptability and accurate prediction results.

[0077] In the embodiments of the present application, the innovation lies in combining the financial and taxation knowledge system and fine-tuning the GPT pre-trained language model to improve the model's ability to understand texts in specific fields, so that the model can more accurately understand and predict the financial behavior of high-net-worth customers.

[0078] In step S206, the key feature information is input into the trained GPT language model for portrait analysis to obtain user portrait information.

[0079] In an embodiment of the present application, the system will output a more comprehensive user portrait for the agent. This portrait is derived from the prediction and completion of existing information. The system will perform similar portrait indexing based on these features and infer other possible information about him.

[0080] In step S207, a recommended speech corresponding to the user portrait is generated based on the dialogue management technology.

[0081] In the embodiments of the present application, the innovation lies in the construction of a personalized recommendation system: based on portrait analysis, the model uses collaborative filtering such as matrix decomposition and neighborhood methods, and integrates reinforcement learning technology to dynamically adjust the recommendation strategy to adapt to changes in customer behavior.

[0082] In the embodiment of the present application, the business development data input by the agent is finally combined, and the dialogue management technology is used to automatically generate communication scripts and speech according to the user portrait. The agent is pushed the appropriate communication tone, communication logic and progress, so that the agent has a comprehensive marketing strategy and solution support.

[0083] In step S208, the recommended words are output.

[0084] In an embodiment of the present application, a method for recommending speech applied to agent question and answer is provided, including: receiving pre-information input by an agent terminal; performing a structured analysis operation on the pre-information according to NLP semantic understanding technology to obtain key feature information; performing a vector conversion operation on the key feature information to obtain a key feature vector; reading a system database, and obtaining effective business behavior data in the system database; fine-tuning the GPT pre-trained language model according to the effective business behavior data and the financial and tax knowledge system to obtain a trained GPT language model; inputting the key feature information into the trained GPT language model for a portrait analysis operation to obtain user portrait information; generating a recommended speech corresponding to the user portrait according to the dialogue management technology; and outputting the recommended speech. Compared with the prior art, the present application can generate a more comprehensive and accurate customer portrait by semantically understanding and extracting key features of the information provided by the agent. It is based on various data obtained from different channels and is trained by applying machine learning models to help agents gain an in-depth understanding of potential customers and their needs; using AI models to comprehensively analyze data and mine hidden laws and patterns, providing agents with more reliable and targeted sales skills recommendations. This can greatly save time and energy, and enhance the sales team's real-time reference to make wise decisions; combined with the financial and tax knowledge system, it ensures that the generated recommendations have the perfection of retrieval resources inside and outside the industry, and has auxiliary functions when handling complex cases; based on customer portrait analysis and agent business development data, the system can provide specific guidance on communication tone, logic and progress. This can help agents develop personalized and effective marketing strategies and adjust them in real time to meet customer needs.

[0085] In some optional implementations of the embodiments of the present application, before the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology, the following steps are also included:

[0086] Perform sentiment analysis on the previous information according to sentiment analysis technology to obtain the user's emotional state;

[0087] The above-mentioned step of generating a recommendation script corresponding to the user profile based on the dialogue management technology specifically includes the following steps:

[0088] Generate recommendation words corresponding to user portraits and user emotional states based on dialogue management technology.

[0089] In the embodiment of the present application, sentiment analysis technology can be used to identify the emotional state of customers, providing insights into the emotional dimension for subsequent analysis, so that the model can further understand the inherent needs and preferences of customers.

[0090] In the embodiment of the present application, by adding the feature of the user's emotional state, a recommendation script that further improves the user experience can be generated.

[0091] In some optional implementations of the embodiments of the present application, the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps:

[0092] In the process of speech text generation, the similarity between the real-time speech text content and the professional knowledge in the financial and tax knowledge graph is calculated in real time;

[0093] Correct the real-time speech text content whose similarity is lower than the preset similarity threshold;

[0094] When all the scripts are generated, you will get real-time recommended scripts.

[0095] In the embodiment of the present application, when the system starts to generate the speech text, it will synchronously perform a real-time calculation process. The core of this calculation is to evaluate the similarity between the content of the currently generated speech text and the professional knowledge stored in the financial and tax knowledge graph.

[0096] In the embodiment of the present application, the finance and taxation knowledge graph is a structured knowledge base that contains key information such as professional terms, rules, cases, etc. in the finance and taxation field. This information is organized into nodes and edges for efficient retrieval and matching.

[0097] In the embodiments of the present application, real-time calculation of similarity usually involves natural language processing (NLP) technology, such as semantic analysis, word vector representation, etc., to capture the deep semantic associations between texts.

[0098] In the embodiment of the present application, the system presets a similarity threshold, which represents the minimum acceptable similarity between the speech text and the professional knowledge.

[0099] In the real-time calculation process of the embodiment of the present application, if the similarity of the content of a certain speech text is lower than the preset threshold, it means that it may not accurately or fully reflect the professional knowledge in the field of finance and taxation. At this time, the system will trigger a correction operation to make necessary adjustments or optimizations to the speech text to ensure that it is more in line with the professional standards of finance and taxation. The correction operation may include replacing inaccurate terms, adding necessary explanations or cases, etc.

[0100] In the embodiments of the present application, not only the professionalism and accuracy of the speech text are improved, but also the timely updating and accuracy of knowledge are ensured through real-time calculation and correction mechanisms.

[0101] In some optional implementations of the embodiments of the present application, the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps:

[0102] Generate the initial speech text corresponding to the user portrait based on the dialogue management technology;

[0103] The initial speech text is optimized according to the attention mechanism and key feature information to obtain the recommended speech.

[0104] In the embodiment of the present application, first, the system needs to extract key feature information from the user's input and the context of the conversation, such as the specific products, service time, location, etc. mentioned by the user, where this information is an important basis for optimizing the speech text; then, the system will assign attention weights to each part of the initial speech text based on these key feature information, the context of the conversation, and the relevant information in the user portrait, where the higher the weight, the more important or relevant it is in the current conversation context; finally, the system will optimize the speech text based on the attention weights, including adjusting the order of sentences, replacing inaccurate words, adding necessary explanations or details, etc. The optimized speech text more accurately reflects the user's needs and context, and is also more in line with the user's preferences and expectations.

[0105] In the embodiment of the present application, a highly personalized and accurate recommendation script is finally generated. This recommendation script can not only better meet the needs of users, but also improve user satisfaction and loyalty.

[0106] In some optional implementations of the embodiments of the present application, after the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology, the following steps are also included:

[0107] Calculate user acceptance of recommended words based on sentiment analysis model;

[0108] If the user's acceptance is lower than the preset acceptance threshold, the recommended words corresponding to the user portrait are regenerated based on the dialogue management technology until the user acceptance of the generated recommended words meets the acceptance threshold.

[0109] In the embodiment of the present application, the sentiment analysis model refers to a model based on natural language processing and machine learning technology, which learns to identify emotional features in text by training a large amount of text data labeled with emotional tendencies (such as positive, negative, and neutral). In the dialogue system, this model will analyze the text content of the recommended words and determine the emotional response that it may cause in users.

[0110] In an embodiment of the present application, based on the results of the sentiment analysis model, the system can calculate a quantitative user acceptance index, which can be a probability value indicating the possibility that the user has a positive emotion toward the dialogue; or it can be a score reflecting the degree of match between the dialogue and the user's expectations or emotions.

[0111] In the embodiments of the present application, the sentiment analysis model and user acceptance evaluation can not only help the system generate recommendation words that are more in line with user expectations, but also continuously improve the user experience through iterative optimization.

[0112] In some optional implementations of the embodiments of the present application, after the step of outputting the recommended words, the following steps are also included:

[0113] Receiving user evaluation data corresponding to the recommended words;

[0114] The parameter weights of the trained GPT language model are updated according to the user evaluation data.

[0115] In the embodiments of the present application, user evaluation data refers to the data collected after the user interacts with the dialogue system, which reflects the user's views on the satisfaction, relevance, accuracy, etc. of the recommended words. This data can exist in various forms, such as:

[0116] (1) Explicit evaluation: Users directly rate the dialogue technology or give textual evaluations, such as “very satisfied”, “somewhat inaccurate”, etc.

[0117] (2) Implicit feedback: User behavior data, such as click-through rate, dwell time, whether to continue the conversation, etc., can also be used as a source of evaluation data.

[0118] In the embodiment of the present application, the system is pre-designed with a mechanism to collect these evaluation data to ensure the accuracy and reliability of the data.

[0119] In the embodiment of the present application, receiving user evaluation data and updating the parameter weights of the GPT language model are key steps to improve the performance of the dialogue system and optimize the recommendation techniques. Through this process, the system can continuously learn and adapt to user preferences and needs, thereby providing more accurate and personalized services.

[0120] The embodiments of the present application can acquire and process relevant data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0121] AI basic technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. AI software technologies mainly include computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.

[0122] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through computer-readable instructions, and the computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a non-volatile storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0123] It should be understood that, although the steps in the flowchart of the accompanying drawings are displayed in sequence as indicated by the arrows, these steps are not necessarily executed in sequence in the order indicated by the arrows. Unless otherwise specified herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least a part of the steps in the flowchart of the accompanying drawings may include multiple sub-steps or multiple stages, and these sub-steps or stages are not necessarily executed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be executed in turn or alternately with other steps or at least a part of the sub-steps or stages of other steps.

[0124] Further references Figure 3 , as a response to the above Figure 2 In order to realize the method shown in the figure, the present application provides an embodiment of a speech recommendation device for agent question and answer, and the device embodiment is similar to Figure 2 Corresponding to the method embodiment shown, the device can be specifically applied to various electronic devices.

[0125] like Figure 3 As shown, the speech technique recommendation device 200 applied to the agent question and answer of the embodiment of the present application includes:

[0126] The pre-information acquisition module 210 is used to receive the pre-information input by the proxy terminal;

[0127] The structured analysis module 220 is used to perform a structured analysis operation on the pre-information according to the NLP semantic understanding technology to obtain key feature information;

[0128] A vector conversion module 230 is used to perform a vector conversion operation on the key feature information to obtain a key feature vector;

[0129] The behavior data acquisition module 240 is used to read the system database and obtain valid business behavior data in the system database;

[0130] A model fine-tuning module 250 is used to fine-tune the GPT pre-trained language model according to effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model;

[0131] The portrait analysis module 260 is used to input key feature information into the trained GPT language model to perform a portrait analysis operation to obtain user portrait information;

[0132] A recommended speech acquisition module 270 is used to generate a recommended speech corresponding to the user profile according to the dialogue management technology;

[0133] The recommended speech output module 280 is used to output the recommended speech.

[0134] In an embodiment of the present application, a speech recommendation device 200 for agent question and answer is provided, including: a pre-information acquisition module 210, which is used to receive pre-information input by an agent terminal; a structured analysis module 220, which is used to perform a structured analysis operation on the pre-information according to NLP semantic understanding technology to obtain key feature information; a vector conversion module 230, which is used to perform a vector conversion operation on the key feature information to obtain a key feature vector; a behavior data acquisition module 240, which is used to read a system database and obtain valid business behavior data in the system database; a model fine-tuning module 250, which is used to fine-tune the GPT pre-trained language model according to the valid business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model; a portrait analysis module 260, which is used to input the key feature information into the trained GPT language model for portrait analysis operation to obtain user portrait information; a recommended speech acquisition module 270, which is used to generate a recommended speech corresponding to the user portrait according to the dialogue management technology; and a recommended speech output module 280, which is used to output the recommended speech. Compared with the existing technology, this application can generate a more comprehensive and accurate customer portrait by semantically understanding the information provided by the agent and extracting key features. It is based on various data obtained from different channels and uses machine learning models for training to help agents gain a deep understanding of potential customers and their needs; it uses AI models to comprehensively analyze data and mine hidden rules and patterns to provide agents with more reliable and targeted sales skills recommendations. This can greatly save time and energy, and enhance the sales team's real-time reference to make wise decisions; combined with the financial and tax knowledge system, it ensures that the generated recommendations have the perfection of retrieval resources inside and outside the industry, and have auxiliary functions when handling complex cases; based on customer portrait analysis and agent business development data, the system can provide specific guidance on communication tone, logic and progress. This can help agents develop personalized and effective marketing strategies and adjust them in real time to meet customer needs.

[0135] In some optional implementations of the embodiments of the present application, the above-mentioned speech recommendation device 200 applied to the agent question and answer further includes: a sentiment analysis module, and the recommended speech acquisition module further includes: a recommended speech acquisition submodule, wherein:

[0136] The sentiment analysis module is used to perform sentiment analysis on the pre-set information according to sentiment analysis technology to obtain the user's emotional state;

[0137] The recommended words acquisition submodule is used to generate recommended words corresponding to the user portrait and the user's emotional state based on the dialogue management technology.

[0138] To solve the above technical problems, the present application also provides a computer device. Figure 4 , Figure 4 This is a basic structural block diagram of a computer device according to an embodiment of the present application.

[0139] The computer device 300 includes a memory 310, a processor 320, and a network interface 330 that are interconnected and communicated through a system bus. It should be noted that the figure only shows a computer device 300 having components 310-330, but it should be understood that it is not required to implement all the components shown, and more or fewer components can be implemented instead. Among them, those skilled in the art can understand that the computer device here is a device that can automatically perform numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes but is not limited to microprocessors, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable gate arrays (Field-Programmable Gate Array, FPGA), digital processors (Digital Signal Processor, DSP), embedded devices, etc.

[0140] The computer device may be a computing device such as a desktop computer, a notebook, a PDA, a cloud server, etc. The computer device may interact with a user through a keyboard, a mouse, a remote controller, a touch pad, or a voice control device.

[0141] The memory 310 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, hard disk, multimedia card, card-type memory (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 310 can be an internal storage unit of the computer device 300, such as a hard disk or memory of the computer device 300. In other embodiments, the memory 310 can also be an external storage device of the computer device 300, such as a plug-in hard disk equipped on the computer device 300, a smart memory card (Smart Media Card, SMC), a secure digital (Secure Digital, SD) card, a flash card, etc. Of course, the memory 310 can also include both the internal storage unit of the computer device 300 and its external storage device. In the embodiment of the present application, the memory 310 is generally used to store the operating system and various application software installed on the computer device 300, such as computer-readable instructions for the speech recommendation method for agent question and answer, etc. In addition, the memory 310 can also be used to temporarily store various data that have been output or are to be output.

[0142] The processor 320 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip in some embodiments. The processor 320 is generally used to control the overall operation of the computer device 300. In the embodiment of the present application, the processor 320 is used to run the computer-readable instructions stored in the memory 310 or process data, such as running the computer-readable instructions of the method for recommending speech techniques for agent question and answer.

[0143] The network interface 330 may include a wireless network interface or a wired network interface. The network interface 330 is generally used to establish a communication connection between the computer device 300 and other electronic devices.

[0144] The computer equipment provided by this application can generate a more comprehensive and accurate customer portrait by semantically understanding the information provided by the agent and extracting key features. It is based on various data obtained from different channels and is trained using machine learning models to help agents gain an in-depth understanding of potential customers and their needs; it uses AI models to conduct comprehensive analysis of data and mine hidden rules and patterns to provide agents with more reliable and targeted sales skills recommendations. This can greatly save time and energy, and enhance the sales team's real-time reference to make wise decisions; combined with the financial and tax knowledge system, it ensures that the generated recommendations have the perfection of retrieval resources inside and outside the industry, and have auxiliary functions when handling complex cases; based on customer portrait analysis and agent business development data, the system can provide specific guidance on communication tone, logic, and progress. This can help agents develop personalized and effective marketing strategies, and adjust them in real time to meet customer needs.

[0145] The present application also provides another implementation, namely, providing a computer-readable storage medium, wherein the computer-readable storage medium stores computer-readable instructions, and the computer-readable instructions can be executed by at least one processor to enable the at least one processor to perform the steps of the method for recommending speech techniques applied to agent question and answer as described above.

[0146] The computer-readable storage medium provided by this application can generate a more comprehensive and accurate customer portrait by semantically understanding the information provided by the agent and extracting key features. It is based on various data obtained from different channels and is trained using machine learning models to help agents gain an in-depth understanding of potential customers and their needs; it uses AI models to comprehensively analyze data and mine hidden rules and patterns to provide agents with more reliable and targeted sales skills recommendations. This can greatly save time and energy, and enhance the sales team's real-time reference to make wise decisions; combined with the financial and tax knowledge system, it ensures that the generated recommendations have the perfection of retrieval resources inside and outside the industry, and have auxiliary functions when handling complex cases; based on customer portrait analysis and agent business development data, the system can provide specific guidance on communication tone, logic, and progress. This can help agents develop personalized and effective marketing strategies and adjust them in real time to meet customer needs.

[0147] Through the description of the above implementation methods, those skilled in the art can clearly understand that the above-mentioned embodiment methods can be implemented by means of software plus a necessary general hardware platform, and of course by hardware, but in many cases the former is a better implementation method. Based on such an understanding, the technical solution of the present application, or the part that contributes to the prior art, can be embodied in the form of a software product, which is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes a number of instructions for a terminal device (which can be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in each embodiment of the present application.

[0148] Obviously, the embodiments described above are only some embodiments of the present application, rather than all embodiments. The preferred embodiments of the present application are given in the accompanying drawings, but they do not limit the patent scope of the present application. The present application can be implemented in many different forms. On the contrary, the purpose of providing these embodiments is to make the understanding of the disclosure of the present application more thorough and comprehensive. Although the present application is described in detail with reference to the aforementioned embodiments, for those skilled in the art, it is still possible to modify the technical solutions recorded in the aforementioned specific implementation methods, or to perform equivalent replacement of some of the technical features therein. Any equivalent structure made using the contents of the specification and drawings of this application, directly or indirectly used in other related technical fields, is similarly within the scope of patent protection of this application.

Claims

1. A method for recommending speech techniques for agent question and answer, characterized in that: The steps include: Receiving pre-information input by the agent terminal; Performing structured analysis on the pre-information according to NLP semantic understanding technology to obtain key feature information; Performing a vector conversion operation on the key feature information to obtain a key feature vector; Reading a system database, and obtaining valid business behavior data in the system database; Fine-tune the GPT pre-trained language model according to the effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model; Inputting the key feature information into the trained GPT language model to perform a portrait analysis operation to obtain user portrait information; Generate a recommendation script corresponding to the user profile based on the dialogue management technology; Output the recommended words.

2. The method for recommending speech techniques for agent question and answer according to claim 1, characterized in that: Before the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology, the following steps are also included: Performing sentiment analysis on the pre-processed information according to sentiment analysis technology to obtain the user's emotional state; The step of generating a recommended speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps: Generate recommendation words corresponding to the user portrait and the user's emotional state based on the dialogue management technology.

3. The method for recommending speech techniques for agent question and answer according to claim 1, characterized in that: The step of generating a recommended speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps: In the process of speech text generation, the similarity between the real-time speech text content and the professional knowledge in the financial and tax knowledge graph is calculated in real time; Correct the real-time speech text content whose similarity is lower than the preset similarity threshold; When all the scripts are generated, you will get real-time recommended scripts.

4. The method for recommending speech techniques for agent question and answer according to claim 1, characterized in that: The step of generating a recommended speech corresponding to the user portrait according to the dialogue management technology specifically includes the following steps: Generate an initial speech text corresponding to the user portrait according to the dialogue management technology; The initial speech text is optimized according to the attention mechanism and the key feature information to obtain the recommended speech.

5. The method for recommending speech techniques for agent question and answer according to claim 1, characterized in that: After the step of generating a recommendation speech corresponding to the user portrait according to the dialogue management technology, the following steps are also included: Calculate the user acceptance of the recommended words according to the sentiment analysis model; If the user acceptance is lower than a preset acceptance threshold, a recommendation speech corresponding to the user portrait is regenerated according to the dialogue management technology until the user acceptance of the generated recommendation speech meets the acceptance threshold.

6. The method for recommending speech techniques for agent question and answer according to claim 1, characterized in that: After the step of outputting the recommended words, the method further includes the following steps: Receiving user evaluation data corresponding to the recommended speech; The parameter weights of the trained GPT language model are updated according to the user evaluation data.

7. A speech recommendation device for agent question and answer, characterized in that: include: A pre-information acquisition module, used for receiving pre-information input by the proxy terminal; A structured analysis module, used to perform a structured analysis operation on the pre-information according to NLP semantic understanding technology to obtain key feature information; A vector conversion module, used to perform a vector conversion operation on the key feature information to obtain a key feature vector; A behavior data acquisition module, used to read a system database and acquire valid business behavior data from the system database; A model fine-tuning module is used to fine-tune the GPT pre-trained language model according to the effective business behavior data and the financial and taxation knowledge system to obtain a trained GPT language model; A portrait analysis module, used to input the key feature information into the trained GPT language model to perform a portrait analysis operation to obtain user portrait information; A recommended speech acquisition module, used to generate a recommended speech corresponding to the user portrait based on the dialogue management technology; The recommended words output module is used to output the recommended words.

8. The speech recommendation device for agent question and answer according to claim 7, characterized in that: The device further includes: a sentiment analysis module, and the recommended speech acquisition module further includes: a recommended speech acquisition submodule, wherein: The sentiment analysis module is used to perform sentiment analysis on the pre-information according to sentiment analysis technology to obtain the user's emotional state; The recommended speech acquisition submodule is used to generate recommended speech corresponding to the user portrait and the user's emotional state based on the dialogue management technology.

9. A computer device comprising a memory and a processor, characterized in that: The memory stores computer-readable instructions, and when the processor executes the computer-readable instructions, the steps of the method for recommending speech techniques for agent question and answer as described in any one of claims 1 to 6 are implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the method for recommending speech techniques for agent question and answer as described in any one of claims 1 to 6.

Images