Response method, device, equipment, medium and program product

Abstract

The invention provides a response method which can be applied to the technical field of finance. The method comprises the steps that intention type recognition is conducted on a natural language query statement input by a user, intention types are obtained, and the intention types comprise banking business handling and other intentions; under the condition that the intention type is bank business handling, obtaining transaction historical data, account data and user behavior mode data of the user; generating an embedded vector according to the natural language query statement, the transaction historical data, the account data and the user behavior mode data; based on the embedded vector, a query intention of the natural language query statement is determined, and the query intention represents the type of the banking business; and responding to the natural language query statement based on the query intention. The invention further provides a response device, equipment, a storage medium and a program product.

Description

Technical Field

[0001] The present disclosure relates to the field of financial technology, and in particular to a response method, device, equipment, medium and program product. Background Art

[0002] With the rapid development of financial technology, the digital transformation of banking business has become a key factor in promoting the development of the industry. Especially in the globalized and highly interconnected market environment, the digitalization of banking business not only provides customers with more convenient and efficient services, but also faces unprecedented challenges, especially in terms of customer service quality and service efficiency. As an important part of financial technology innovation, the performance of the digital human customer service system directly affects the bank's brand image and customer satisfaction. Summary of the invention

[0003] In view of the above problems, the present disclosure provides a response method, apparatus, device, medium and program product.

[0004] According to a first aspect of the present disclosure, a response method is provided, the method comprising:

[0005] Performing intent type recognition on the natural language query sentence input by the user to obtain the intent type, wherein the intent type includes banking business processing and other intents;

[0006] In the case where the intention type is banking business handling, obtaining the transaction history data, account data and user behavior pattern data of the user;

[0007] generating an embedding vector according to the natural language query statement, the transaction history data, the account data, and the user behavior pattern data;

[0008] Determining the query intent of the natural language query sentence based on the embedding vector, wherein the query intent represents the type of the banking service;

[0009] Based on the query intent, a response is given to the natural language query statement.

[0010] According to an embodiment of the present disclosure, the intent type is identified for a natural language query sentence input by a user to obtain the intent type, including:

[0011] Performing syntactic and semantic analysis on the natural language query sentence to obtain an analysis result;

[0012] The intent type is identified based on the analysis result to obtain the intent type of the natural language query sentence.

[0013] According to an embodiment of the present disclosure, the performing of intent type identification according to the analysis result to obtain the intent type of the natural language query sentence includes:

[0014] Inputting the natural language query sentence into a BERT model to extract context-dependent features of the natural language query sentence;

[0015] Based on the context-dependent features and the analysis results, the intent type of the natural language query statement is identified.

[0016] According to an embodiment of the present disclosure, generating an embedding vector according to the natural language query statement, the transaction history data, the account data, and the user behavior pattern data includes:

[0017] Perform one-hot encoding on the account data to obtain an encoding result;

[0018] Normalizing the transaction history data to obtain a normalized result;

[0019] The natural language query statement, the normalization result, the encoding result and the user behavior pattern data are converted into an embedding vector.

[0020] According to an embodiment of the present disclosure, determining the query intent of the natural language query sentence based on the embedding vector and includes:

[0021] Perform deep learning analysis based on the embedding vector to obtain the probability of various types of banking services;

[0022] The probabilities of the various types of banking services are input into a support vector machine to obtain the query intent of the natural language query statement.

[0023] According to an embodiment of the present disclosure, the deep learning analysis based on the embedded vector to obtain the probability of various types of banking services includes:

[0024] The embedding vector is input into a deep neural network, and the probability of each type of banking business is obtained based on the intent type, and the deep neural network has multiple hidden layers.

[0025] According to an embodiment of the present disclosure, responding to the natural language query statement based on the query intent includes:

[0026] Using advanced natural language generation techniques, develop an initial response strategy based on the type of banking activity being discussed;

[0027] The initial response strategy is scored in real time by using a deep belief network to obtain a scoring result;

[0028] Based on the scoring result, optimizing the preliminary response strategy using an adaptive adjustment algorithm to obtain a customized response strategy;

[0029] Based on the customized response strategy, a response is made to the natural language query statement.

[0030] According to an embodiment of the present disclosure, the method further includes:

[0031] Collect the user's feedback data and response effect data for quantitative analysis to obtain quantitative results;

[0032] Performing dimensionality reduction processing on the quantization result to obtain a dimensionality reduction result;

[0033] Based on the dimensionality reduction result, a linear regression model is used to determine the relationship between the target performance indicator and the user satisfaction, and a regression result is obtained;

[0034] The optimization goal is to minimize the sum of squares of errors, and the target performance index is adjusted according to the regression results.

[0035] According to an embodiment of the present disclosure, the method further includes:

[0036] The query intent of the natural language query request is saved in the session state.

[0037] A second aspect of the present disclosure provides a response device, the device comprising:

[0038] An identification module, used to identify the intent type of the natural language query sentence input by the user, and obtain the intent type, wherein the intent type includes banking business processing and other intents;

[0039] An acquisition module, used for acquiring the transaction history data, account data and user behavior pattern data of the user when the intention type is banking business handling;

[0040] A generating module, configured to generate an embedding vector according to the natural language query statement, the transaction history data, the account data and the user behavior pattern data;

[0041] A determination module, configured to determine the query intent of the natural language query statement based on the embedding vector and the intent type, wherein the query intent represents the type of the banking service;

[0042] A response module is used to respond to the natural language query statement based on the query intent.

[0043] A third aspect of the present disclosure provides an electronic device, comprising: one or more processors; and a memory for storing one or more computer programs, wherein the one or more processors execute the one or more computer programs to implement the steps of the above method.

[0044] The fourth aspect of the present disclosure further provides a computer-readable storage medium having a computer program or instructions stored thereon, which implements the steps of the above method when the above computer program or instructions are executed by a processor.

[0045] The fifth aspect of the present disclosure further provides a computer program product, including a computer program or instructions, which implement the steps of the above method when the above computer program or instructions are executed by a processor. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The above contents and other purposes, features and advantages of the present disclosure will become more apparent through the following description of the embodiments of the present disclosure with reference to the accompanying drawings, in which:

[0047] Figure 1 The application scenario diagram of the answering method, apparatus, device, medium and program product according to the embodiment of the present disclosure is schematically shown;

[0048] Figure 2 A flowchart of a response method according to an embodiment of the present disclosure is schematically shown;

[0049] Figure 3 A flowchart of responding to a natural language query statement based on query intent according to an embodiment of the present disclosure is schematically shown;

[0050] Figure 4 A flowchart schematically shows another answering method according to an embodiment of the present disclosure;

[0051] Figure 5 The structure block diagram of the answering device according to the embodiment of the present disclosure is schematically shown;

[0052] Figure 6 A block diagram of an electronic device suitable for implementing the answering method according to an embodiment of the present disclosure is schematically shown. DETAILED DESCRIPTION

[0053] Hereinafter, embodiments of the present disclosure will be described with reference to the accompanying drawings. However, it should be understood that these descriptions are exemplary only and are not intended to limit the scope of the present disclosure. In the following detailed description, for ease of explanation, many specific details are set forth to provide a comprehensive understanding of the embodiments of the present disclosure. However, it is apparent that one or more embodiments may also be implemented without these specific details. In addition, in the following description, descriptions of known structures and technologies are omitted to avoid unnecessary confusion of the concepts of the present disclosure.

[0054] The terms used herein are only for describing specific embodiments and are not intended to limit the present disclosure. The terms "comprise", "include", etc. used herein indicate the existence of the features, steps, operations and / or components, but do not exclude the existence or addition of one or more other features, steps, operations or components.

[0055] All terms (including technical and scientific terms) used herein have the meanings commonly understood by those skilled in the art unless otherwise defined. It should be noted that the terms used herein should be interpreted as having a meaning consistent with the context of this specification and should not be interpreted in an idealized or overly rigid manner.

[0056] When using expressions such as "at least one of A, B, and C, etc.", they should generally be interpreted according to the meaning of the expression commonly understood by those skilled in the art (for example, "a system having at least one of A, B, and C" should include but is not limited to a system having A alone, B alone, C alone, A and B, A and C, B and C, and / or A, B, C, etc.).

[0057] It should be noted that the response method, device, equipment, medium and program product provided by the present disclosure can be used for the application of artificial intelligence in the field of financial technology, and can also be used in any field other than the field of financial technology. The present disclosure does not limit the application field of the provided response method, device, equipment, medium and program product.

[0058] In the technical solution of the present disclosure, the user information (including but not limited to user personal information, user image information, user device information, such as location information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved are all information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with relevant laws, regulations and standards, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0059] In the scenario of using personal information for automated decision-making, the methods, devices, and systems provided by the embodiments of the present disclosure provide users with corresponding operation portals for users to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered. The expression "automated decision-making" here refers to the activity of automatically analyzing and evaluating a person's behavioral habits, interests and hobbies, or economic, health, credit status, etc. through computer programs, and making decisions. The expression "expert decision-making" here refers to the activity of making decisions by people who specialize in a certain field, have specialized experience, knowledge and skills, and have reached a certain level of professionalism.

[0060] The embodiments of the present disclosure provide a response method, apparatus, device, medium and program product, including: identifying the intent type of the natural language query statement input by the user to obtain the intent type, the intent type includes banking business processing and other intents; when the intent type is banking business processing, obtaining the user's transaction history data, account data and user behavior pattern data; generating an embedding vector based on the natural language query statement, transaction history data, account data and user behavior pattern data; based on the embedding vector and the intent type, determining the query intent of the natural language query statement, the query intent characterizing the type of banking business; based on the query intent, responding to the natural language query statement. It not only improves the number and speed of processing queries, but also significantly improves the service quality and customer satisfaction, and effectively solves the problems of slow response speed and insufficient service personalization in traditional methods. At the same time, it reduces the operation error rate, improves the accuracy and security of services, and brings higher customer retention rates and better market competitiveness to banks.

[0061] Figure 1 The application scenario diagram of the response method, apparatus, device, medium and program product according to the embodiments of the present disclosure is schematically shown.

[0062] like Figure 1 As shown, the application scenario 100 according to this embodiment may include a terminal device 101, a network 102, and a server 103. 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.

[0063] 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 shopping applications, web browser applications, search applications, instant messaging tools, email clients, social platform software, etc. (only for example).

[0064] The terminal device 101 can be any terminal device capable of interacting with a user, including but not limited to smart phones, tablet computers, smart wearable devices, laptop computers, vehicle-mounted terminals, smart speakers, unmanned vehicles or robots, etc.

[0065] The server 103 may be a server that provides various services, such as a background management server (only as an example) that provides support for applications used by users using the terminal device 101. The background management server may analyze and process received data such as user requests, and feed back processing results (such as web pages, information, or data obtained or generated according to user requests) to the terminal device.

[0066] In order to enable the terminal device 101 and the server 103 to provide the user with intelligent question-answering services by cooperating with each other, a target application corresponding to the server 103 may be run in the terminal device 101. Through the target application, the terminal device 101 can realize data interaction with the server 103. Further, the target application may be an independently running application, a small program based on a platform program, a web page, a plug-in or a component, etc., which is not limited in this application. Further, the terminal device 101 and the server 103 may be connected by a wireless communication network to realize data interaction between each other. The wireless communication network may include any one or combination of 5G mobile communication network technology (5th-Generation, 5G) system, long term evolution (long term evolution, LTE) system, Global System for Mobile Communication (Global System for Mobile Communication, GSM), Bluetooth (Bluetooth, BT), Wireless Fidelity (Wireless Fidelity, Wi-Fi), Code Division Multiple Access (Code Division Multiple Access, CDMA) network, wideband code division multiple access (wideband code division multiple access, WCDMA) network, long range wireless communication (Long Range, Lora) technology or Zigbee protocol (Zigbee).

[0067] It should be noted that the answering method provided in the embodiment of the present disclosure can generally be executed by the server 103. Accordingly, the answering device provided in the embodiment of the present disclosure can generally be set in the server 103. The answering method provided in the embodiment of the present disclosure can also be executed by a server or server cluster that is different from the server 103 and can communicate with the terminal device 101 and / or the server 103. Accordingly, the answering device provided in the embodiment of the present disclosure can also be set in a server or server cluster that is different from the server 103 and can communicate with the terminal device 101 and / or the server 103.

[0068] 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 implementation requirements.

[0069] The following will be based on Figure 1 The scene described by Figure 2~Figure 4 The response method of the disclosed embodiment is described in detail.

[0070] Figure 2 The flowchart of the response method according to the embodiment of the present disclosure is schematically shown.

[0071] like Figure 2 As shown, the response method of this embodiment includes operations S210 to S230, and the response method can be executed by the server 103.

[0072] In operation S210, the natural language query sentence input by the user is identified as an intent type to obtain an intent type, where the intent type includes banking business processing and other intents.

[0073] In operation S220, when the intent type is banking service processing, the user's transaction history data, account data, and user behavior pattern data are obtained.

[0074] In operation S230, an embedding vector is generated according to the natural language query statement, the transaction history data, the account data, and the user behavior pattern data.

[0075] In operation S240, based on the embedding vector, the query intent of the natural language query sentence is determined, and the query intent represents the type of banking services.

[0076] Banking service types may include loan applications, financial product purchases, card applications, etc., and this disclosure does not impose specific restrictions on this.

[0077] In operation S250 , a response is made to the natural language query sentence based on the query intent.

[0078] In an embodiment of the present disclosure, before obtaining the transaction history data, account data, and user behavior pattern data of the user, the user's consent or authorization may be obtained. For example, before operation S220, a request to obtain the user's transaction history data, account data, and user behavior pattern data may be issued to the user. When the user agrees or authorizes that the user's transaction history data, account data, and user behavior pattern data may be obtained, operation S220 is performed.

[0079] In the embodiments of the present disclosure, a corresponding operation portal may be provided for the user to choose to agree or reject the automated decision result. That is, before using the user information for response processing, the user's instruction to agree or reject the processing input through the corresponding operation portal may be obtained. If the user agrees to process, the user information is processed for response. If the user rejects the processing, the expert decision process is entered.

[0080] In some embodiments of the present disclosure, intent type identification is performed on a natural language query statement input by a user, and obtaining the intent type includes: performing syntactic and semantic analysis on the natural language query statement to obtain analysis results; and performing intent type identification based on the analysis results to obtain the intent type of the natural language query statement.

[0081] In some embodiments, a natural language query statement input by a user may be received through a digital platform, and the natural language query statement text may be received through Protocol encryption transmission to ensure data security:

[0082]

[0083] in, Represents the encrypted natural language query text. Indicates the use A function that uses an encryption algorithm to encrypt natural language query statement text.

[0084] The received encrypted query is decrypted on the server side to restore the user's original natural language query text:

[0085]

[0086] in, Indicates the use The decryption algorithm is a function that decrypts the encrypted natural language query statement text.

[0087] Use natural language processing algorithms to preprocess natural language query text, standardize natural language query text, remove useless spaces and special characters, convert all characters to lowercase, and perform word segmentation :

[0088]

[0089] in, Represents the processed natural language query text.

[0090] By determining the keywords and phrases in the query and their relative weights and language structures, the processed natural language query text can be subjected to syntactic and semantic analysis using a language model to obtain analysis results. :

[0091]

[0092] in, Represents the language model function applied to the processed natural language query text.

[0093] Based on the analysis results, identify the type of user query intent: ,in, Represents a function that identifies the intent type of a user query.

[0094] In some embodiments of the present disclosure, intent type identification is performed based on analysis results to obtain the intent type of a natural language query statement, including: inputting the natural language query statement into a BERT model to extract context-dependent features of the natural language query statement; and identifying the intent type of the natural language query statement based on the context-dependent features and the analysis results.

[0095] Loading pre-trained Model, which is used to process and parse users’ natural language queries:

[0096]

[0097] in, Represents the set of pre-trained parameters of the model.

[0098] Preprocess and encode the user's natural language query text and convert it into a format acceptable to the model:

[0099]

[0100] in, is a word in the natural language query text, is the corresponding word vector;

[0101] Then use The model processes the encoded natural language query text and extracts the context-dependent features of the natural language query text:

[0102]

[0103] in, represents the obtained context-dependent features, is the input word vector, are model parameters.

[0104] Analyze the context-dependent features obtained to identify the intent type of the natural language query statement:

[0105]

[0106] in, Represents the intent extracted from the natural language query text. is a specially designed extraction function, yes The context-dependent features of the model output, syntactic and semantic analysis to obtain analysis results .

[0107] In some embodiments of the present disclosure, generating an embedding vector based on natural language query statements, transaction history data, account data and user behavior pattern data includes: performing one-hot encoding on the account data to obtain an encoding result; normalizing the transaction history data to obtain a normalized result; and converting the natural language query statement, the normalization result, the encoding result and the user behavior pattern data into an embedding vector.

[0108] In some embodiments, transaction history data T, account data A, and user behavior pattern data B may be collected from the user's digital interactions.

[0109] For account information Perform one-hot encoding to get the encoding result . Normalize transaction history data , and get the normalized result ,in, and They are The mean and standard deviation of .

[0110] The natural language query sentence T and the normalized result , encoding results and user behavior pattern data B are transformed through a joint embedding model to form a unified embedding vector ,in, , , , Respectively represent , , , The feature vector extracted from , , , is the corresponding weight coefficient.

[0111] In some embodiments of the present disclosure, determining the query intent of a natural language query statement based on an embedding vector includes: performing deep learning analysis based on the embedding vector to obtain the probabilities of various types of banking services; and inputting the probabilities of various types of banking services into a support vector machine to obtain the query intent of the natural language query statement.

[0112] Embedding vector Will be used for further data analysis and model training to support machine learning tasks: .in, is the loss function, It is the actual various banking operations. So and model parameters is the prediction function of the input. In some embodiments, the model parameters can be adjusted by gradient descent method Update to optimize the performance of the entire system: .in, is the learning rate, is the loss function with respect to the model parameters gradient.

[0113] In some embodiments of the present disclosure, deep learning analysis is performed based on the embedding vector to obtain the probability of each type of banking business, including: inputting the embedding vector into a deep neural network, obtaining the probability of each type of banking business based on the intent type, and the deep neural network has multiple hidden layers.

[0114] Embedding vector Perform deep learning analysis to identify user needs and intentions. The model can be a deep neural network with multiple hidden layers. :

[0115]

[0116] in, The index of the presentation layer, is the activation function, and It is The weights and biases of the layers, is the input layer.

[0117] From the final output layer of the deep neural network Extract features through support vector machine To determine the user's specific needs, that is, the type of banking services they want to conduct:

[0118]

[0119] in, is the predicted banking category label, The function is The decision function of .

[0120] according to Model Output Conduct decision analysis:

[0121]

[0122] in, is a given output The next prediction is category The probability of is the total number of categories;

[0123] In some optional embodiments, predictive banking is implemented , by evaluating multiple service execution paths and using a graph search algorithm to determine the optimal path:

[0124]

[0125] in, It is a diagram representing the service process. is the path cost function, is a heuristic function.

[0126] Thus, banking based on predictions output from deep neural networks and the corresponding probability , choose the most suitable banking service:

[0127]

[0128] in, is a set of possible service functions, It is a function Utility evaluation;

[0129] For selected services Optimize the parameters to maximize service delivery efficiency and customer satisfaction:

[0130]

[0131] in, represents the service parameter vector, represents the cost coefficient, and Coefficient matrices and vectors representing constraints;

[0132] Banking services tailored to the specific needs of users and service strategies Initialization service process:

[0133]

[0134] in, It is a conditional function based on user needs and selected service strategies, which determines the specific path of service execution;

[0135] Execute services and use transaction logic to ensure data security and operational reliability: ; Execute service operations ; , Check whether the operation results meet business rules and user expectations;

[0136] After the service is completed, the results are evaluated and the user satisfaction survey is conducted, and the feedback loop is used to optimize future services:

[0137] ;

[0138] in, The function evaluates service satisfaction based on user feedback, It is the feedback information provided by the user.

[0139] In some embodiments, Figure 1 The method shown also includes: saving the query intent of the natural language query request in the session state. ,in, Represents a function that updates the session state.

[0140] Figure 3 A flowchart for responding to a natural language query statement based on query intent according to an embodiment of the present disclosure is schematically shown.

[0141] like Figure 3 As shown, the embodiment of responding to a natural language query statement based on the query intent includes operations S310 to S330, and the response method can be executed by the server 103.

[0142] In operation S310, a preliminary response strategy is formulated according to the type of banking services using advanced natural language generation technology.

[0143] Generate responses to user queries, leveraging advanced natural language generation Technology, based on analyzed user needs and service strategies Customize your initial response strategy: ,in, for Model parameters.

[0144] In operation S320, the preliminary response strategy is scored in real time through a deep belief network to obtain a scoring result.

[0145] Through deep learning models such as deep belief networks Real-time scoring of generated preliminary response strategies ,in, yes Parameters of the model.

[0146] In operation S330 , based on the scoring result, the preliminary response strategy is optimized using an adaptive adjustment algorithm to obtain a customized response strategy.

[0147] Optimize the response by adaptively adjusting the algorithm based on the scoring results: ,in, Functions are dynamically adjusted based on user feedback Parameters of the model.

[0148] Multi-criteria decision analysis can be used To determine the final presentation format of the response, that is, a customized response strategy ,in, The function considers the advantages and disadvantages of various output methods and selects the optimal interactive format.

[0149] In operation S340 , a response is made to the natural language query statement based on the customized response strategy.

[0150] Figure 4 The flowchart of a response method according to another embodiment of the present disclosure is schematically shown.

[0151] like Figure 4 As shown, the responding method of this embodiment includes operations S410 to S430, and the responding method can be executed by the server 103.

[0152] In operation S410, user feedback data and response effect data are collected for quantitative analysis to obtain a quantitative result.

[0153] Collect user feedback data and system response effect data, conduct quantitative analysis, and use the collected data Represented as observation vectors, each vector contains performance indicators of multiple dimensions ;in, is the performance indicator vector for the nth interaction.

[0154] In operation S420, the quantization result is subjected to dimensionality reduction processing to obtain a dimensionality reduction result.

[0155] Apply principal component analysis to reduce the dimensionality of the collected data and identify the most important performance factors ,in, , ,and They are The left singular vectors, singular values, and right singular vectors are used to extract the main performance influencing factors.

[0156] In operation S430, based on the dimension reduction result, a linear regression model is used to determine the relationship between the target performance indicator and the user satisfaction, and a regression result is obtained.

[0157] based on Results: A linear regression model was used to determine the relationship between the target performance indicator and user satisfaction. .in, Indicates user satisfaction, is the main performance indicator, is the regression coefficient, is the error term;

[0158] In operation S440, the target performance index is adjusted according to the regression result with minimizing the sum of squares of errors as the optimization goal.

[0159] Adjust the service logic according to the results of the regression model. The optimization goal is to adjust the service parameters to minimize the error. The sum of squares: .

[0160] In some implementations, new service processes are implemented and their effectiveness is monitored, and data will continue to be collected and fed back to And regression model, a closed-loop control system is formed:

[0161] ;

[0162] in, It is a new data set collected under the new service process.

[0163] The following data table shows the target performance indicator improvement after the bank customer service center applies the response method, device, equipment, medium and program product provided by the embodiments of the present disclosure.

[0164] Table 1

[0165] Data Category Before deploying the system After deploying the system Number of queries processed per month 20,000 times 30,000 times Response time 2.5 minutes 30 seconds User satisfaction rating 3.2 points 4.5 points Automation processing rate 65% 90% Error rate 5% 1.5% Personalized service matching 70% 95%

[0166] As can be seen from Table 1 above, after deploying the response method, device, equipment, medium and program product provided by the present disclosure, the number of customer inquiries that the bank can handle per month has increased significantly, from 20,000 to 30,000. This change directly reflects the enhancement of the system's processing capabilities. The significant reduction in response time, from 2.5 minutes to just 30 seconds, has greatly improved the customer experience and reflects the system's significant advantage in improving response speed. User satisfaction has increased from 3.2 points to 4.5 points, indicating that customers' overall acceptance and satisfaction with the service have been substantially improved.

[0167] The increase in the automated processing rate, from 65% to 90%, indicates that the system can more effectively handle standardized and routine queries, reducing the need for manual intervention and improving overall operational efficiency. The decrease in the error rate, from 5% to 1.5%, further demonstrates the system's ability to improve processing accuracy. In addition, the improvement in service personalization matching, from 70% to 95%, indicates that the system better understands and meets the personalized needs of customers, reflecting the efficiency of the present invention in customer data analysis and application.

[0168] It can be seen from these data that the response method, device, equipment, medium and program product provided by the present disclosure not only improves the number and speed of processing inquiries, but also significantly improves service quality and customer satisfaction, and effectively solves the problems of slow response speed and insufficient service personalization in traditional systems. At the same time, the implementation of the system greatly reduces the operational error rate and improves the accuracy and security of services. These improvements have brought higher customer retention rates and better market competitiveness to banks.

[0169] Based on the above response method, the present disclosure also provides a response device. Figure 5 The device is described in detail.

[0170] Figure 5 The structural block diagram of the response device according to the embodiment of the present disclosure is schematically shown.

[0171] like Figure 5 As shown, the response device 500 of this embodiment includes an identification module 510 , an acquisition module 520 , a generation module 530 , a determination module 540 and a response module 550 .

[0172] The recognition module 510 is used to recognize the intent type of the natural language query sentence input by the user to obtain the intent type, and the intent type includes banking business processing and other intents. In one embodiment, the recognition module 510 can be used to perform the operation S210 described above, which will not be repeated here.

[0173] The acquisition module 520 is used to acquire the transaction history data, account data and user behavior pattern data of the user when the intention type is bank business handling. In one embodiment, the acquisition module 520 can be used to perform the operation S220 described above, which will not be repeated here.

[0174] The generation module 530 is used to generate an embedding vector according to the natural language query statement, the transaction history data, the account data and the user behavior pattern data. In one embodiment, the generation module 530 can be used to perform the operation S230 described above, which will not be repeated here.

[0175] The determination module 540 is used to determine the query intent of the natural language query sentence based on the embedding vector and the intent type, wherein the query intent represents the type of the banking service. In one embodiment, the generation module 540 can be used to perform the operation S240 described above, which will not be described in detail here.

[0176] The response module 550 is used to respond to the natural language query sentence based on the query intent. In one embodiment, the response module 550 can be used to perform the operation S250 described above, which will not be described in detail here.

[0177] According to an embodiment of the present disclosure, the intent type identification of a natural language query statement input by a user to obtain the intent type includes: performing syntactic and semantic analysis on the natural language query statement to obtain an analysis result; and identifying the intent type based on the analysis result to obtain the intent type of the natural language query statement.

[0178] According to an embodiment of the present disclosure, performing intent type identification based on the analysis result to obtain the intent type of the natural language query statement includes: inputting the natural language query statement into a BERT model to extract context-dependent features of the natural language query statement; and identifying the intent type of the natural language query statement based on the context-dependent features and the analysis result.

[0179] According to an embodiment of the present disclosure, generating an embedding vector based on the natural language query statement, the transaction history data, the account data and the user behavior pattern data includes: performing one-hot encoding on the account data to obtain an encoding result; normalizing the transaction history data to obtain a normalized result; and converting the natural language query statement, the normalization result, the encoding result and the user behavior pattern data into an embedding vector.

[0180] According to an embodiment of the present disclosure, determining the query intent of the natural language query statement based on the embedding vector and includes: performing deep learning analysis based on the embedding vector to obtain the probabilities of various types of banking services; inputting the probabilities of various types of banking services into a support vector machine to obtain the query intent of the natural language query statement.

[0181] According to an embodiment of the present disclosure, performing deep learning analysis based on the embedding vector to obtain the probability of each type of banking business includes: inputting the embedding vector into a deep neural network, obtaining the probability of each type of banking business based on the intent type, and the deep neural network has multiple hidden layers.

[0182] According to an embodiment of the present disclosure, responding to the natural language query statement based on the query intent includes: using advanced natural language generation technology to formulate a preliminary response strategy according to the type of banking business; scoring the preliminary response strategy in real time through a deep belief network to obtain a scoring result; based on the scoring result, optimizing the preliminary response strategy using an adaptive adjustment algorithm to obtain a customized response strategy; and responding to the natural language query statement based on the customized response strategy.

[0183] According to an embodiment of the present disclosure, the device 500 also includes: a collection module, which is used to collect the user's feedback data and response effect data for quantitative analysis to obtain a quantitative result; a dimension reduction module, which is used to perform dimensionality reduction processing on the quantitative result to obtain a dimensionality reduction result; a regression module, which is used to determine the relationship between the target performance indicator and the user satisfaction based on the dimensionality reduction result using a linear regression model to obtain a regression result; and an adjustment module, which is used to adjust the target performance indicator according to the regression result with the minimization of the sum of squares of errors as the optimization goal.

[0184] According to an embodiment of the present disclosure, the apparatus 500 further includes: a saving module, configured to save the query intent of the natural language query request in a session state.

[0185] According to an embodiment of the present disclosure, any multiple modules of the identification module 510, the acquisition module 520, the generation module 530, the determination module 540 and the response module 550 can be combined into one module for implementation, or any one of the modules can be split into multiple modules. Alternatively, at least part of the functions of one or more of these modules can be combined with at least part of the functions of other modules and implemented in one module. According to an embodiment of the present disclosure, at least one of the identification module 510, the acquisition module 520, the generation module 530, the determination module 540 and the response module 550 can be at least partially implemented as a hardware circuit, such as a field programmable gate array (FPGA), a programmable logic array (PLA), a system on a chip, a system on a substrate, a system on a package, an application specific integrated circuit (ASIC), or can be implemented by hardware or firmware such as any other reasonable way of integrating or packaging the circuit, or implemented in any one of the three implementation methods of software, hardware and firmware or in a suitable combination of any of them. Alternatively, at least one of the identification module 510 , the acquisition module 520 , the generation module 530 , the determination module 540 , and the response module 550 may be at least partially implemented as a computer program module, and when the computer program module is executed, a corresponding function may be performed.

[0186] Figure 6 A block diagram of an electronic device suitable for implementing the answering method according to an embodiment of the present disclosure is schematically shown.

[0187] like Figure 6 As shown, the electronic device 600 according to an embodiment of the present disclosure includes a processor 601, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 602 or a program loaded from a storage part 608 to a random access memory (RAM) 603. The processor 601 may include, for example, a general-purpose microprocessor (e.g., a CPU), an instruction set processor and / or a related chipset and / or a dedicated microprocessor (e.g., an application-specific integrated circuit (ASIC)), etc. The processor 601 may also include an onboard memory for caching purposes. The processor 601 may include a single processing unit or multiple processing units for performing different actions of the method flow according to an embodiment of the present disclosure.

[0188] In RAM 603, various programs and data required for the operation of electronic device 600 are stored. Processor 601, ROM 602 and RAM 603 are connected to each other via bus 604. Processor 601 performs various operations of the method flow according to the embodiment of the present disclosure by executing the program in ROM 602 and / or RAM 603. It should be noted that the program can also be stored in one or more memories other than ROM 602 and RAM 603. Processor 601 can also perform various operations of the method flow according to the embodiment of the present disclosure by executing the program stored in the one or more memories.

[0189] According to an embodiment of the present disclosure, the electronic device 600 may further include an input / output (I / O) interface 605, which is also connected to the bus 604. The electronic device 600 may further include one or more of the following components connected to the input / output (I / O) interface 605: an input portion 606 including a keyboard, a mouse, etc.; an output portion 607 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage portion 608 including a hard disk, etc.; and a communication portion 609 including a network interface card such as a LAN card, a modem, etc. The communication portion 609 performs communication processing via a network such as the Internet. A drive 610 is also connected to the input / output (I / O) interface 605 as needed. A removable medium 611, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 610 as needed, so that a computer program read therefrom is installed into the storage portion 608 as needed.

[0190] The present disclosure also provides a computer-readable storage medium, which may be included in the device / apparatus / system described in the above embodiments; or may exist independently without being assembled into the device / apparatus / system. The above computer-readable storage medium carries one or more programs, and when the above one or more programs are executed, the method according to the embodiment of the present disclosure is implemented.

[0191] According to an embodiment of the present disclosure, the computer-readable storage medium may be a non-volatile computer-readable storage medium, for example, it may include but is not limited to: a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program, which may be used by or in combination with an instruction execution system, an apparatus or a device. For example, according to an embodiment of the present disclosure, the computer-readable storage medium may include the ROM 602 and / or RAM 603 described above and / or one or more memories other than ROM 602 and RAM 603.

[0192] The embodiment of the present disclosure also includes a computer program product, which includes a computer program, and the computer program contains program code for executing the method shown in the flowchart. When the computer program product is run in a computer system, the program code is used to enable the computer system to implement the response method provided by the embodiment of the present disclosure.

[0193] The above functions defined in the system / device of the embodiment of the present disclosure are performed when the computer program is executed by the processor 601. According to the embodiment of the present disclosure, the system, device, module, unit, etc. described above can be implemented by a computer program module.

[0194] In one embodiment, the computer program may rely on tangible storage media such as optical storage devices, magnetic storage devices, etc. In another embodiment, the computer program may also be transmitted and distributed in the form of signals on a network medium, and downloaded and installed through the communication part 609, and / or installed from a removable medium 611. The program code contained in the computer program may be transmitted using any appropriate network medium, including but not limited to: wireless, wired, etc., or any suitable combination of the above.

[0195] In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 609, and / or installed from the removable medium 611. When the computer program is executed by the processor 601, the above functions defined in the system of the embodiment of the present disclosure are performed. According to the embodiment of the present disclosure, the system, device, apparatus, module, unit, etc. described above can be implemented by a computer program module.

[0196] According to an embodiment of the present disclosure, the program code for executing the computer program provided by the embodiment of the present disclosure can be written in any combination of one or more programming languages. Specifically, these computing programs can be implemented using high-level process and / or object-oriented programming languages, and / or assembly / machine languages. Programming languages ​​include, but are not limited to, Java, C++, python, "C" language or similar programming languages. The program code can be executed entirely on the user computing device, partially on the user device, partially on the remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device can be connected to the user computing device through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computing device (for example, using an Internet service provider to connect through the Internet).

[0197] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each box in the flow chart or block diagram can represent a module, a program segment, or a part of a code, and the above-mentioned module, program segment, or a part of a code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram or flow chart, and the combination of the boxes in the block diagram or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0198] It will be appreciated by those skilled in the art that the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways, even if such combinations or combinations are not explicitly described in the present disclosure. In particular, without departing from the spirit and teachings of the present disclosure, the features described in the various embodiments of the present disclosure may be combined and / or combined in a variety of ways. All of these combinations and / or combinations fall within the scope of the present disclosure.

[0199] The embodiments of the present disclosure are described above. However, these embodiments are only for illustrative purposes and are not intended to limit the scope of the present disclosure. Although the embodiments are described above, this does not mean that the measures in the various embodiments cannot be used in combination to advantage. Without departing from the scope of the present disclosure, those skilled in the art may make a variety of substitutions and modifications, which should all fall within the scope of the present disclosure.

Claims

1. A response method, characterized in that: The method comprises: Performing intent type recognition on the natural language query sentence input by the user to obtain the intent type, wherein the intent type includes banking business processing and other intents; In the case where the intention type is banking business handling, obtaining the transaction history data, account data and user behavior pattern data of the user; generating an embedding vector according to the natural language query statement, the transaction history data, the account data, and the user behavior pattern data; Determining the query intent of the natural language query sentence based on the embedding vector, wherein the query intent represents the type of the banking service; Based on the query intent, a response is given to the natural language query statement.

2. The method according to claim 1, characterized in that The intent type of the natural language query sentence input by the user is identified, and the intent type obtained includes: Performing syntactic and semantic analysis on the natural language query sentence to obtain an analysis result; The intent type is identified based on the analysis result to obtain the intent type of the natural language query sentence.

3. The method according to claim 2, characterized in that The intent type is identified according to the analysis result to obtain the intent type of the natural language query sentence, including: Inputting the natural language query sentence into a BERT model to extract context-dependent features of the natural language query sentence; Based on the context-dependent features and the analysis results, the intent type of the natural language query statement is identified.

4. The method according to claim 1, characterized in that: The generating an embedding vector according to the natural language query statement, the transaction history data, the account data and the user behavior pattern data comprises: Perform one-hot encoding on the account data to obtain an encoding result; Normalizing the transaction history data to obtain a normalized result; The natural language query statement, the normalization result, the encoding result and the user behavior pattern data are converted into an embedding vector.

5. The method according to claim 1, characterized in that The determining, based on the embedding vector, the query intent of the natural language query statement comprises: Perform deep learning analysis based on the embedding vector to obtain the probability of various types of banking services; The probabilities of the various types of banking services are input into a support vector machine to obtain the query intent of the natural language query statement.

6. The method according to claim 5, characterized in that The deep learning analysis based on the embedding vectors to obtain the probabilities of various banking services includes: The embedding vector is input into a deep neural network, and the probability of each type of banking business is obtained based on the intent type, and the deep neural network has multiple hidden layers.

7. The method according to claim 1, characterized in that The responding to the natural language query statement based on the query intent includes: Using advanced natural language generation techniques, develop an initial response strategy based on the type of banking activity being discussed; The initial response strategy is scored in real time by using a deep belief network to obtain a scoring result; Based on the scoring result, optimizing the preliminary response strategy using an adaptive adjustment algorithm to obtain a customized response strategy; Based on the customized response strategy, a response is made to the natural language query statement.

8. The method according to claim 1, characterized in that The method further comprises: Collect the user's feedback data and response effect data for quantitative analysis to obtain quantitative results; Performing dimensionality reduction processing on the quantization result to obtain a dimensionality reduction result; Based on the dimensionality reduction result, a linear regression model is used to determine the relationship between the target performance indicator and the user satisfaction, and a regression result is obtained; The optimization goal is to minimize the sum of squares of errors, and the target performance index is adjusted according to the regression results.

9. The method according to claim 1, characterized in that: The method further comprises: The query intent of the natural language query request is saved in the session state.

10. A response device, characterized in that: The device comprises: An identification module, used to identify the intent type of the natural language query sentence input by the user, and obtain the intent type, wherein the intent type includes banking business processing and other intents; An acquisition module, used for acquiring the transaction history data, account data and user behavior pattern data of the user when the intention type is banking business handling; A generating module, configured to generate an embedding vector according to the natural language query statement, the transaction history data, the account data and the user behavior pattern data; A determination module, configured to determine the query intent of the natural language query statement based on the embedding vector and the intent type, wherein the query intent represents the type of the banking service; A response module is used to respond to the natural language query statement based on the query intent.

11. An electronic device, comprising: one or more processors; a memory for storing one or more computer programs, It is characterized in that the one or more processors execute the one or more computer programs to implement the steps of the method according to any one of claims 1 to 9.

12. A computer-readable storage medium having a computer program or instruction stored thereon, characterized in that: When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

13. A computer program product comprising a computer program or instructions, characterized in that When the computer program or instruction is executed by a processor, the steps of the method according to any one of claims 1 to 9 are implemented.

Images