Intelligent customer service question and answer method, device and computer equipment based on micro-service architecture

Abstract

Embodiments of the present disclosure disclose a smart customer service question and answer method, device and computer equipment based on a micro-service architecture. A specific embodiment of the method includes: in response to receiving question information sent by a target terminal, determining the similarity between each question and answer information in a preset question and answer information library and the question information; in response to determining that the similarity is greater than or equal to a preset similarity, using a pre-trained intelligent question and answer information model to generate an intelligent question and answer output information set for the question information; according to the obtained model prediction probability group set, fusing each model question and answer result in the model question and answer result group set to generate a fused model question and answer result set and a corresponding fused prediction probability set; according to the fused prediction probability set and the fused model question and answer result set, generating an actual model question and answer result group; and according to the actual model question and answer result group, generating target smart question reply information. This embodiment improves the accuracy and efficiency of the reply to the question information.

Description

TECHNICAL FIELD

[0001] Embodiments of the present disclosure relate to the field of intelligent customer service question answering, and in particular to an intelligent customer service question answering method and device based on a micro-service architecture and a computer device. BACKGROUND

[0002] In the field of intelligent customer service question answering, the reply to a user's question is usually performed in the following manner: the questions of various types of users are classified through a preset intelligent question answering platform, and then it is determined whether to reply by a model or by a customer service. However, the above-mentioned manner of replying to a user's question usually has the following technical problem: low reply efficiency.

[0003] The above information disclosed in this BACKGROUND section is only for the purpose of enhancing the understanding of the background of the present inventive concepts, and therefore, it can contain information that does not form the prior art known to those of ordinary skill in the art. SUMMARY

[0004] The summary of the present disclosure is used to introduce the concepts in a simple form, which will be described in detail in the specific embodiments section. The summary of the present disclosure is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0005] Some embodiments of the present disclosure propose an intelligent customer service question answering method and device based on a micro-service architecture, a computer device and a computer readable medium to solve one or more of the technical problems mentioned in the above BACKGROUND section.

[0006] In a first aspect, some embodiments of the present disclosure provide an intelligent customer service question and answer method based on a micro-service architecture, which comprises: in response to detecting a network security detection instruction, collecting behavior data information of each user in a current network system to obtain a behavior data information set; inputting the behavior data information set into a pre-trained user operation behavior prediction model to obtain a user operation behavior prediction result set, wherein the behavior data information in the behavior data information set corresponds to the user operation behavior prediction result in the user operation behavior prediction result set; collecting network situation awareness information of the network system within a preset time period, wherein the network situation awareness information comprises threat intelligence information, network traffic data, security logs and alarm data; performing network data analysis on the network situation awareness information to obtain a network data analysis result; in response to determining that there is a user operation behavior prediction result in the user operation behavior prediction result set that meets a user behavior alarm condition, determining the behavior data information corresponding to the user operation behavior prediction result that meets the user behavior alarm condition as abnormal behavior data information to obtain an abnormal behavior data information group, and sending the abnormal behavior data information group to an associated abnormal user alarm terminal; and sending the network data analysis result to an associated network system maintenance terminal.

[0007] In a second aspect, some embodiments of the present disclosure provide an intelligent customer service question and answer device based on a micro-service architecture, which comprises: a first collection unit configured to, in response to detecting a network security detection instruction, collect behavior data information of each user in a current network system to obtain a behavior data information set; an input unit configured to input the behavior data information set into a pre-trained user operation behavior prediction model to obtain a user operation behavior prediction result set, wherein the behavior data information in the behavior data information set corresponds to the user operation behavior prediction result in the user operation behavior prediction result set; a second collection unit configured to collect network situation awareness information of the network system within a preset time period, wherein the network situation awareness information comprises threat intelligence information, network traffic data, security logs and alarm data; an analysis unit configured to perform network data analysis on the network situation awareness information to obtain a network data analysis result; a determination unit configured to, in response to determining that there is a user operation behavior prediction result in the user operation behavior prediction result set that meets a user behavior alarm condition, determine the behavior data information corresponding to the user operation behavior prediction result that meets the user behavior alarm condition as abnormal behavior data information to obtain an abnormal behavior data information group, and send the abnormal behavior data information group to an associated abnormal user alarm terminal; and a sending unit configured to send the network data analysis result to an associated network system maintenance terminal.

[0008] In a third aspect, some embodiments of the present disclosure provide a computer device, comprising: one or more processors; a storage device having one or more programs stored thereon, when the one or more programs are executed by the one or more processors, the one or more processors implement the method described in any implementation manner of the first aspect.

[0009] In a fourth aspect, some embodiments of the present disclosure provide a computer readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method described in any implementation manner of the first aspect.

[0010] The above various embodiments of the present disclosure have the following beneficial effects: through the intelligent customer service question and answer method based on the micro-service architecture of some embodiments of the present disclosure, the accuracy and efficiency of the reply to the question information are improved. Based on this, the intelligent customer service question and answer method based on the micro-service architecture of some embodiments of the present disclosure, first, in response to receiving the question information sent by the target terminal, determines the similarity between each question and answer information in the preset question and answer information library and the above question information. Thus, it can be determined whether the question raised by the user is a new question through the question and answer information in the question and answer information library. Secondly, in response to determining that the above similarity is greater than or equal to a preset similarity, using a pre-trained intelligent question and answer information model, an intelligent question and answer output information set is generated for the above question information, wherein each intelligent question and answer output information includes: a model question and answer result group and a model prediction probability group, and the intelligent question and answer information model includes: an intelligent question and answer information network group. Thus, after determining that the question raised is not a new question, the pre-trained intelligent question and answer information network group can be used to reply to the question information comprehensively. Then, according to the obtained model prediction probability group set, each model question and answer result in the model question and answer result group set is fused to generate a fusion model question and answer result set and a corresponding fusion prediction probability set. Then, according to the above fusion prediction probability set and the above fusion model question and answer result set, an actual model question and answer result group is generated; according to the above actual model question and answer result group, a target intelligent question and answer reply information is generated. Thus, by implementing the fusion of the model question and answer result and the fusion of the corresponding prediction probability, the output content can be unified to facilitate the subsequent accurate reply to the question. Finally, the above target intelligent question and answer reply information is converted into an intelligent question and answer reply voice, and the above target intelligent question and answer reply information and the above intelligent question and answer reply voice are sent to the target terminal. Thus, the accuracy and efficiency of the reply to the question information are improved. BRIEF DESCRIPTION OF DRAWINGS

[0011] The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by describing in detail some embodiments thereof with reference to the attached drawings. Throughout the drawings, the same or similar reference numerals refer to the same or similar elements. It should be understood that the drawings are schematic, and elements and elements are not necessarily drawn to scale.

[0012] Figure 1 is a flowchart of some embodiments of the method for intelligent customer service question answering based on micro-service architecture according to the present disclosure;

[0013] Figure 2 is a flowchart of some embodiments of the apparatus for intelligent customer service question answering based on micro-service architecture according to the present disclosure;

[0014] Figure 3 is a structural schematic diagram of a computer device suitable for implementing some embodiments of the present disclosure. DETAILED DESCRIPTION

[0015] Embodiments of the present disclosure will be described below in greater detail with reference to the accompanying drawings. Although some embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure can be implemented in various forms, and should not be interpreted as being limited to the embodiments set forth herein. On the contrary, these embodiments are provided so as to more thoroughly and completely understand the present disclosure. It should be understood that the drawings and embodiments of the present disclosure are only for exemplary purposes, and are not intended to limit the scope of protection of the present disclosure.

[0016] It should also be noted that, for ease of description, only the parts related to the present application are shown in the drawings. The embodiments in the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0017] It should be noted that the terms "first", "second", etc. mentioned in the present disclosure are only used to distinguish different devices, modules or units, and are not intended to limit the order or interdependence of the functions performed by these devices, modules or units.

[0018] It should be noted that the adjectives "one", "multiple" mentioned in the present disclosure are illustrative and not limiting, and those skilled in the art should understand that, unless otherwise explicitly stated in the context, it should be understood as "one or more".

[0019] The names of the messages or information exchanged between the devices in the embodiments of the present disclosure are only for illustrative purposes, and are not intended to limit the scope of these messages or information.

[0020] The present disclosure will be described in detail below with reference to the accompanying drawings and in conjunction with the embodiments.

[0021] Figure 1 is a flowchart of some embodiments of the method for intelligent customer service question answering based on micro-service architecture according to the present disclosure. The flowchart 100 of some embodiments of the method for intelligent customer service question answering based on micro-service architecture according to the present disclosure is shown. The method for intelligent customer service question answering based on micro-service architecture is applied to an intelligent customer service system based on micro-service architecture, and includes the following steps:

[0022] In step 101, in response to receiving the question information sent by the target terminal, the similarity between each question and answer information in the preset question and answer information library and the question information is determined.

[0023] In some embodiments, the execution subject (e.g., a computing device) of the intelligent customer service question and answer method based on the micro-service architecture can determine the similarity between each question and answer information in the preset question and answer information library and the question information sent by the target terminal in response to receiving the question information. The target terminal can refer to a user terminal or a user terminal at a certain seat in the agent terminal. The question and answer information library can be a pre-established question information library. The similarity between each question and answer information in the preset question and answer information library and the question information can be determined by a cosine similarity formula.

[0024] The intelligent customer service system based on the micro-service architecture includes a presentation layer, an application layer, a service layer, and a data layer. The presentation layer includes a user terminal and an information display terminal. The application layer includes a communication client, an agent terminal, a background management terminal, a visitor terminal, a soft phone, a traffic monitoring terminal, and a bill management terminal. The service layer includes a communication server, a background system management server, an agent management server, a visitor terminal server, a knowledge base operation management server, a background management new function server, an agent management new function server, an information release management server, an intelligent knowledge base operation management function server, and an online customer service outbound management function server. The data layer includes a relational structured database, an unstructured database, an in-memory database, a distributed file library, and a distributed log library. The intelligent customer service system based on the micro-service architecture further includes a resource layer.

[0025] The user terminal can refer to a PC terminal, and the information display terminal can refer to a WeChat public number (H5) page terminal.

[0026] The application layer splits the intelligent customer service platform into a communication client, an agent terminal, a background management terminal, a visitor terminal, a soft phone, a traffic monitoring terminal, and a bill management terminal.

[0027] The service layer completely componentizes the requirements of the intelligent customer service platform, decouples the communication server, the background system management server, and the agent management server, and independently develops, designs, runs, and maintains the communication server and the background system management server. The communication between the communication server and the background system management server is performed through restful API interfaces and websockets. The communication server, the background system management server, the agent management server, and the visitor terminal server are decoupled through a kafka message queue.

[0028] Data layer: design full-service data center, structured data, unstructured data, resource files, audio files, log information are stored and managed uniformly in relational structured database, unstructured database, in-memory database, distributed file library, distributed log library, etc.

[0029] Resource layer: follow the cloud design specification, improve the security of software running environment through cloud security suite, deploy different services to corresponding network environment according to the isolation of network.

[0030] Among them, the communication server mainly realizes communication with the communication client, and provides message sending, receiving, storage, etc., the communication client is integrated into the background system management application and the agent management application, and communicates with the communication server. The communication architecture adopts the classic inbox model, aligns the messages through the global seq, and simplifies the use of communication architecture. This service needs to be deployed in the Internet area, and the database (relational database, unstructured database) involved needs to be deployed to the management information area to improve the security of data.

[0031] Background system management server: including system setting, basic information management, knowledge base management, monitoring management, quality inspection management, statistical management and other modules, the function permission and data permission under different services are different, this server is mainly used for users under the administrator and team leader roles, and is deployed in the PaaS cloud of the management information area.

[0032] Agent management server: including agent workbench, agent monitoring management, agent knowledge base management, agent common language management and other modules. This server is used by users in the role of agent, and the agent can chat with the visitor through the temporary conference room. This server is deployed in the PaaS cloud of the Internet area.

[0033] Visitor server: including consulting business, topic navigation, common problems, service evaluation, message board and other modules. This server mainly supports visitor access and is deployed in the PaaS cloud of the Internet area.

[0034] Knowledge base operation and management server: including intelligent knowledge category management, source file management, standard knowledge management, knowledge distribution, extended knowledge management, test knowledge management, task type conversation management, knowledge confirmation, knowledge synchronization, knowledge discovery, label optimization and other modules, the function permission and data permission under different services are different, this service is mainly used for users in the roles of knowledge operation and maintenance personnel and knowledge base administrator.

[0035] Background management new function server: including intelligent interaction data analysis, knowledge asset statistics, team personalized settings, navigation settings and other modules, the function permissions and data permissions are different under different services, this server is mainly used for users under the roles of administrator and team leader.

[0036] Agent management new function server: including artificial agent navigation, message reading, agent emotion and sentiment recognition modules. This server is used by users in the role of an agent.

[0037] Information release management server: including template management, material management, draft management, classification management, content list management, release management, shield word management, user permission management, business handling and other modules, the function permissions and data permissions are different under different services, this server is mainly used for users under the roles of administrator and WeChat channel information operation personnel.

[0038] Intelligent knowledge base operation management function server: intelligent knowledge base operation management mainly adds telephone business channel knowledge operation function on the basis of phase II, including channel management, standard knowledge management, knowledge distribution management, knowledge processing management, knowledge confirmation management and other modules, this server is used by users under the roles of administrator and knowledge operation.

[0039] Online customer service outbound call management function server: including local outbound call of agent, remote outbound call of mobile phone and other modules, this server is used by users under the role of agent.

[0040] As a result, service efficiency is improved: the intelligent customer service system based on micro-service architecture can automatically handle routine problems, improving service efficiency and reducing operating costs. Customer satisfaction is improved: the intelligent customer service system improves user experience through personalized recommendations and guidance, thereby improving customer satisfaction. System stability is enhanced: the high availability and fault tolerance of micro-service architecture ensure stable operation of the system, reducing the impact of system failures on business. Promote technological innovation: micro-service architecture supports technology diversity, providing more technology choices for enterprises and promoting technological innovation and development.

[0041] Optionally, an initial intelligent question and answer information model is determined. The initial intelligent question and answer information model includes an initial intelligent question and answer information network group, and each initial intelligent question and answer information network corresponds to a question and answer reply mode. The question and answer reply mode can be divided into a single question and answer mode, a multiple question and answer mode, a single question and answer mode + answer analysis, a multiple question and answer mode + answer analysis, and the like. That is, there can be multiple question and answer modes for one question. The initial intelligent question and answer information network can be an untrained intelligent question and answer network model. For example, the initial intelligent question and answer information network can be a BP neural network, an MLP neural network, a recurrent neural network (RNN), and a long short-term memory network (LSTM).

[0042] Optionally, for each initial intelligent question and answer information network in the initial intelligent question and answer information network group, the following training steps are performed:

[0043] First, an intelligent question and answer information sample set corresponding to the initial intelligent question and answer information network is obtained. The question and answer reply mode corresponding to the intelligent question and answer information sample set is the same as the question and answer reply mode corresponding to the initial intelligent question and answer information network.

[0044] Second, a target intelligent question and answer information sample is selected from the intelligent question and answer information sample set. One intelligent question and answer information sample can be randomly selected from the intelligent question and answer information sample set as the target intelligent question and answer information sample.

[0045] Third, the question information included in the target intelligent question and answer information sample is input into the initial intelligent question and answer information network to obtain an initial intelligent reply result.

[0046] Fourth, a loss value between the initial intelligent reply result and a sample label corresponding to the target intelligent question and answer information sample is determined. The loss value between the initial intelligent reply result and the sample label corresponding to the target intelligent question and answer information sample can be determined by a pre-set loss function. For example, the loss function can be a cross-entropy loss function or a hinge loss function.

[0047] Fifth, in response to determining that the loss value is less than or equal to a pre-set loss value, the initial intelligent question and answer information network is determined as a trained intelligent question and answer information network.

[0048] Optionally, in response to determining that the loss value is greater than the pre-set loss value, the network parameters of the initial intelligent question and answer information network are adjusted, a target intelligent question and answer information sample is selected from the unselected intelligent question and answer information samples, and the initial intelligent question and answer information network after adjustment is trained again. The network parameters of the initial intelligent question and answer information network can be adjusted by a gradient descent method.

[0049] Optionally, the trained intelligent question and answer information networks are fused into an intelligent question and answer information model.

[0050] At step 102, in response to determining that the similarity is greater than or equal to the preset similarity, the pre-trained intelligent question and answer information model is used to generate a set of intelligent question and answer output information for the question information.

[0051] In some embodiments, the execution subject can generate a set of intelligent question and answer output information for the question information using a pre-trained intelligent question and answer information model in response to determining that the similarity is greater than or equal to the preset similarity. Each intelligent question and answer output information includes a model question and answer result group and a model prediction probability group, and the intelligent question and answer information model includes an intelligent question and answer information network group. For example, the intelligent question and answer information network group can include a trained BP neural network, MLP neural network, recurrent neural network (RNN), and long short-term memory network (LSTM). The intelligent question and answer output information can be the output result of the intelligent question and answer information network. Each intelligent question and answer output information includes a model question and answer result group and a model prediction probability group. The model question and answer result is the output question and answer result of the intelligent question and answer information network. The model question and answer result in the model question and answer result group has a one-to-one correspondence with the model prediction probability in the model prediction probability group. The model prediction probability can represent the accuracy of the model question and answer result. The higher the model prediction probability, the more accurate the corresponding generated model question and answer result. The question information can be input into each intelligent question and answer information network to obtain a set of intelligent question and answer output information. One intelligent question and answer output information corresponds to one intelligent question and answer information network. One intelligent question and answer information network outputs multiple model question and answer results.

[0052] At step 103, according to the obtained set of model prediction probability groups, each model question and answer result in the set of model question and answer result groups is fused to generate a set of fused model question and answer results and a corresponding set of fused prediction probabilities.

[0053] In some embodiments, the execution subject can generate a set of fused model question and answer results and a corresponding set of fused prediction probabilities by fusing each model question and answer result in the set of model question and answer result groups according to the obtained set of model prediction probability groups. The fused model question and answer result in the set of fused model question and answer results corresponds to the fused prediction probability in the set of fused prediction probabilities. The fused prediction probability can represent the accuracy of the corresponding fused model question and answer result.

[0054] For example, first, the execution subject can group the model question and answer results in the model question and answer result group set into similar result groups to obtain each question and answer result group. Then, the question and answer results in each question and answer result group are combined in terms of result semantics to generate a combined result as a fusion model question and answer result, and the average of each model prediction probability in the corresponding each model prediction probability group is determined as a fusion prediction probability to obtain a corresponding fusion prediction probability set.

[0055] Step 104, generating an actual model question and answer result group according to the fusion prediction probability set and the fusion model question and answer result set.

[0056] In some embodiments, the execution subject can generate an actual model question and answer result group according to the fusion prediction probability set and the fusion model question and answer result set. The fusion model question and answer result corresponding to the fusion prediction probability greater than or equal to the preset probability can be screened from the fusion model question and answer result set as an actual model question and answer result to obtain an actual model question and answer result group.

[0057] Step 105, generating target intelligent question and answer reply information according to the actual model question and answer result group.

[0058] In some embodiments, the execution subject can generate target intelligent question and answer reply information according to the actual model question and answer result group. For example, each actual model question and answer result in the actual model question and answer result group can be combined into target intelligent question and answer reply information. For another example, the same content in each actual model question and answer result can be taken as main intelligent question and answer reply information, and the different content in each actual model question and answer result can be taken as branch intelligent question and answer reply information. Then, the main intelligent question and answer reply information and the branch intelligent question and answer reply information can be combined into target intelligent question and answer reply information.

[0059] Step 106, converting the target intelligent question and answer reply information into intelligent question and answer reply voice, and sending the target intelligent question and answer reply information and the intelligent question and answer reply voice to the target terminal.

[0060] In some embodiments, the execution subject can convert the target intelligent question and answer reply information into intelligent question and answer reply voice, and send the target intelligent question and answer reply information and the intelligent question and answer reply voice to the target terminal. For example, the target intelligent question and answer reply information can be converted into intelligent question and answer reply voice through text-to-speech technology.

[0061] Optionally, in response to determining that the outbound call processing stage is not in the initial start stage, generating an outbound call prediction quantity and a to-be-processed outbound call quantity according to an outbound call index set corresponding to the current time.

[0062] In some embodiments, the execution subject can generate an outbound call prediction quantity and a to-be-processed outbound call quantity according to a set of outbound call indicators corresponding to a current time, in response to determining that the outbound call processing stage is not in an initial start stage. The set of outbound call indicators can include: an average idle time of an agent, a number of times of a call agent, a number of outbound call agents, an outbound call processing quantity, and an outbound call connection rate. The outbound call processing stage can be an operation stage in which an outbound call processing operation is performed in a preset time period. The initial start stage can be a stage representing the beginning of the execution of the outbound call processing. The outbound call processing operation can be an operation of making a call to a client at an agent end. The set of outbound call indicators can be a set of outbound call indicator contents generated before the current time. The outbound call prediction quantity can be a quantity of calls to be made in the preset time period. The to-be-processed outbound call quantity can be a number of outbound calls that have been made and are waiting to be connected at the current time. The average idle time of the agent can be an average idle time of the agent during the outbound call process. That is, the average idle time of the agent can be a time period during which the agent is idle and does not make a call during the outbound call process. First, the execution subject can obtain a historical outbound call quantity sequence. Then, the execution subject can generate the outbound call prediction quantity according to the historical outbound call quantity sequence by using a regression model. Next, the execution subject can determine the to-be-processed outbound call quantity as the indicator information of the outbound call that is currently being made but not connected. The outbound call processing quantity can be a processing quantity of event processing after a call. The outbound call connection rate can be an average connection rate during the outbound call process. The number of times of a call agent can be a number of agents in a call state. The number of outbound call agents can be a number of agents that support the outbound call operation in the preset time period.

[0063] Optionally, an actual outbound call quantity corresponding to the agent end is generated according to the outbound call prediction quantity and the to-be-processed outbound call quantity.

[0064] In some embodiments, the execution subject can generate an actual outbound call quantity corresponding to the agent end according to the outbound call prediction quantity and the to-be-processed outbound call quantity.

[0065] In practice, the execution subject can generate an actual outbound call quantity corresponding to the agent end by the following steps:

[0066] First, in response to determining that the number of outbound call agents is less than a preset number of agents and the number of outbound call agents is in a first number of agents interval, the number of outbound call agents is determined as the actual outbound call quantity.

[0067] In the second step, in response to determining that the number of outbound call agents is less than the preset number of agents and the number of outbound call agents is in the second number of agents interval, a first outbound call prediction quantity is generated according to a first preset proportion corresponding to the outbound call processing quantity and a second preset proportion corresponding to the outbound call connection rate. The values in the second number of agents interval are greater than the values in the first number of agents interval. The first preset proportion and the second preset proportion can be proportions set for the number of outbound call agents being in the second number of agents interval. For example, the outbound call processing quantity and the outbound call connection rate can be weighted and summed according to the first preset proportion corresponding to the outbound call processing quantity and the second preset proportion corresponding to the outbound call connection rate to generate the first outbound call prediction quantity.

[0068] In the third step, an actual outbound call quantity is generated according to the to-be-processed outbound call quantity and the first outbound call prediction quantity. For example, the sum of the to-be-processed outbound call quantity and the first outbound call prediction quantity can be determined as the actual outbound call quantity.

[0069] In the fourth step, in response to determining that the number of outbound call agents is less than the preset number of agents and the number of outbound call agents is in the third number of agents interval, an adjusted prediction outbound call quantity is generated according to a first proportion corresponding to the outbound call processing quantity. The values in the third number of agents interval are greater than the values in the second number of agents interval. The first proportion can be preset. The product of the outbound call processing quantity and the first proportion can be determined as the adjusted prediction outbound call quantity.

[0070] In the fifth step, an actual outbound call quantity is generated according to the to-be-processed outbound call quantity and the adjusted prediction outbound call quantity. The sum of the to-be-processed outbound call quantity and the adjusted prediction outbound call quantity can be determined as the actual outbound call quantity.

[0071] In the sixth step, in response to determining that the number of outbound call agents is less than the preset number of agents and the number of outbound call agents is in the fourth number of agents interval, an adjusted outbound call quantity is generated according to a second proportion corresponding to the outbound call processing quantity. The values in the fourth number of agents interval are greater than the values in the third number of agents interval. The second proportion is greater than the first proportion. For example, the product of the outbound call processing quantity and the second proportion can be determined as the adjusted outbound call quantity.

[0072] In the seventh step, an actual outbound call quantity is generated according to the to-be-processed outbound call quantity and the adjusted outbound call quantity. The sum of the to-be-processed outbound call quantity and the adjusted outbound call quantity can be determined as the actual outbound call quantity.

[0073] Optionally, according to the actual outbound call quantity, an outbound call operation corresponding to the outbound call processing stage is performed on the agent end.

[0074] In some embodiments, the execution subject described above can perform the call operation corresponding to the call processing stage on the agent side according to the actual call volume described above. For example, a telephone distribution list corresponding to the actual call volume can be determined. Then, each telephone in the telephone distribution list is dialed to perform the call operation of the call processing stage in the preset time period.

[0075] Therefore, by requiring the length of time when the agent is idle, the actual call volume in the preset time period can be generated through the call index set, and the effective control of the call processing in the preset time period is realized through the actual call volume, so as to improve the call efficiency of the agent.

[0076] Further referring to Figure 2 , as an implementation of the method shown in the above figures, the present disclosure provides some embodiments of an intelligent customer service question and answer device based on a micro-service architecture, which correspond to the method embodiments shown in Figure 1 , and the intelligent customer service question and answer device based on the micro-service architecture can be applied to various computer devices.

[0077] As shown in Figure 2 , the intelligent customer service question and answer device 200 based on the micro-service architecture of some embodiments includes a determination unit 201, a first generation unit 202, a fusion unit 203, a second generation unit 204, a third generation unit 205, and a conversion unit 206. The determination unit 201 is configured to determine the similarity between each question and answer information in a preset question and answer information library and the question information sent by a target terminal in response to receiving the question information. The first generation unit 202 is configured to generate a set of intelligent question and answer output information for the question information by using a pre-trained intelligent question and answer information model in response to determining that the similarity is greater than or equal to a preset similarity, wherein each intelligent question and answer output information includes a model question and answer result group and a model prediction probability group, and the intelligent question and answer information model includes an intelligent question and answer information network group. The fusion unit 203 is configured to fuse each model question and answer result in the model question and answer result group set according to the obtained model prediction probability group set to generate a fusion model question and answer result set and a corresponding fusion prediction probability set. The second generation unit 204 is configured to generate an actual model question and answer result group according to the fusion prediction probability set and the fusion model question and answer result set. The third generation unit 205 is configured to generate target intelligent question and answer reply information according to the actual model question and answer result group. The conversion unit 206 is configured to convert the target intelligent question and answer reply information into intelligent question and answer reply voice, and send the target intelligent question and answer reply information and the intelligent question and answer reply voice to the target terminal.

[0078] It is to be understood that the units described in the intelligent customer service question and answer device 200 based on the micro-service architecture correspond to the units described with reference to Figure 1 The individual steps in the described method correspond. Thus, the operations, features and resulting advantages described above for the method also apply to the intelligent customer service question and answer device 200 based on the micro-service architecture and the units contained therein, which are not described again here.

[0079] Reference is made below to Figure 3 which shows a structural diagram of a computer device (e.g., a computing device) 300 suitable for implementing some embodiments of the present disclosure. The computer device in some embodiments of the present disclosure can include, but is not limited to, a mobile terminal such as a mobile phone, a notebook computer, a digital broadcast receiver, a PDA (Personal Digital Assistant), a PAD (Tablet PC), a PMP (Portable Multimedia Player), a car terminal (e.g., a car navigation terminal), and the like, as well as a stationary terminal such as a digital TV, a desktop computer, and the like. Figure 3 The computer device shown is merely an example and should not impose any limitation on the functions and scope of use of embodiments of the present disclosure.

[0080] As shown in Figure 3 , the computer device 300 can include a processing device (e.g., a central processing unit, a graphics processing unit, etc.) 301 that can perform various appropriate actions and processes according to programs stored in a read-only memory (ROM) 302 or programs loaded from a storage device 308 into a random access memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the computer device 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other through a bus 304. An input / output (I / O) interface 305 is also connected to the bus 304.

[0081] In general, the following devices can be connected to the I / O interface 305: an input device 306 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, and the like; an output device 307 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, and the like; a storage device 308 including, for example, a magnetic tape, a hard disk, and the like; and a communication device 309. The communication device 309 can allow the computer device 300 to communicate wirelessly or wired with other devices to exchange data. Although Figure 3 The computer device 300 is shown with various devices, but it is to be understood that all of the devices shown are not required to be implemented or present. More or fewer devices can alternatively be implemented or present. Figure 3 Each block shown in the figure can represent a device or, as needed, a plurality of devices.

[0082] In particular, the processes described above with reference to the flowcharts can be implemented as a computer software program according to some embodiments of the present disclosure. For example, some embodiments of the present disclosure include a computer program product comprising a computer program carried on a computer readable medium, the computer program comprising program code for performing the methods illustrated by the flowcharts. In some such embodiments, the computer program can be downloaded and installed from a network via the communication device 309, or installed from the storage device 308, or installed from the ROM 302. When the computer program is executed by the processing device 301, the above-described functions defined in the methods of some embodiments of the present disclosure are performed.

[0083] It should be noted that the computer readable medium recorded with the program code according to some embodiments of the present disclosure can be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium may, for example, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or apparatus, or any combination of the above. More specific examples of the computer readable storage medium can include, but are not limited to, an electrical connection having one or more wires, a portable computer diskette, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In some embodiments of the present disclosure, the computer readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus or device. In some embodiments of the present disclosure, the computer readable signal medium can include a data signal carried in a baseband or as a part of a carrier wave, which carries computer readable program code. Such a propagated data signal can take on many forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. The computer readable signal medium can also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate or transport program for use by or in connection with an instruction execution system, apparatus or device. The program code contained on the computer readable medium can be transmitted by any suitable medium, including but not limited to a wire, cable, optical fiber, RF (radio frequency), or any suitable combination of the above.

[0084] In some embodiments, the client, server, can communicate using any known or later developed network protocols such as HTTP (HyperText Transfer Protocol), and can be interconnected with any form or medium of digital data communication (e.g., communication network). Examples of communication networks include local area networks ("LAN"), wide area networks ("WAN"), the Internet, and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any then known or later developed networks.

[0085] The computer readable medium described above can be included in the computer device described above; or can exist separately, without being assembled into the computer device. The computer readable medium described above carries one or more programs, when the one or more programs are executed by the computer device, cause the computer device to: in response to receiving question information sent by a target terminal, determine a similarity between each question and answer information in a preset question and answer information library and the question information; in response to determining that the similarity is greater than or equal to a preset similarity, generate a set of intelligent question and answer output information for the question information by using a pre-trained intelligent question and answer information model, wherein each intelligent question and answer output information includes a model question and answer result group and a model prediction probability group, and the intelligent question and answer information model includes an intelligent question and answer information network group; according to the obtained set of model prediction probability groups, fuse each model question and answer result in the set of model question and answer results to generate a set of fused model question and answer results and a corresponding set of fused prediction probabilities; according to the set of fused prediction probabilities and the set of fused model question and answer results, generate an actual model question and answer result group; according to the actual model question and answer result group, generate target intelligent question and answer reply information; convert the target intelligent question and answer reply information into intelligent question and answer reply voice, and send the target intelligent question and answer reply information and the intelligent question and answer reply voice to the target terminal.

[0086] Computer program code for carrying out operations of some embodiments of the present disclosure can be written in any combination of one or more programming languages, including an object oriented programming language such as Java, Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code can execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the latter scenario, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection can be made to an external computer (for example, through the Internet using an Internet Service Provider).

[0087] The flow diagrams and the block diagrams in the drawings are illustrations of architectures, functionalities, and operations of possible implementations of systems, methods, and computer program products according to various embodiments of present disclosure. In this regard, each block in the flow diagrams or block diagrams can represent a module, a procedure, or a part of code, which comprises one or more executable instructions for implementing the specified logical functions. It should also be noted that in some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently or the blocks may

[0088] The units described in some embodiments of the present disclosure can be implemented by software, or can be implemented by hardware. The described units can also be arranged in a processor, for example, can be described as: a processor comprising a determination unit, a first generation unit, a fusion unit, a second generation unit, a third generation unit and a conversion unit. Among them, the names of these units do not constitute a limitation to the units themselves in some cases, for example, the sending unit can also be described as: "a unit for converting the target intelligent question reply information into intelligent question reply voice, and sending the target intelligent question reply information and the intelligent question reply voice to the target terminal".

[0089] The functionality described herein above can be performed, at least in part, by one or more hardware logic components. For example, and without limitation, illustrative types of hardware logic components that can be used include Field-programmable Gate Arrays (FPGAs), Program- specific Integrated Circuits (ASICs), Program- specific Standard Products (ASSPs), System-on-a-chip systems (SOCs), Complex Programmable Logic Devices (CPLDs), etc.

[0090] The above description is merely exemplary of the disclosure and the application made use of the principles of the technology. It is to be understood that the application scope of the embodiments of the disclosure is not limited to the specific combinations of technical features described above, and should also cover other technical solutions formed by any combination of the above technical features or equivalent features thereof without departing from the inventive concept. For example, the technical solutions formed by replacing the above features with technical features having similar functions disclosed in the embodiments of the disclosure (but not limited to) with each other.

Claims

1. A microservice-based intelligent customer service question-answering method, applied to a microservice-based intelligent customer service system, comprising: In response to receiving a question from a target terminal, the similarity between each question and answer in a preset question and answer information database and the question is determined, wherein the target terminal refers to a user terminal at a certain seat in the agent terminal; In response to determining that the similarity is greater than or equal to a preset similarity, a set of intelligent question-answering output information is generated for the question information using a pre-trained intelligent question-answering information model. Each intelligent question-answering output information includes a model question-answering result group and a model prediction probability group. The intelligent question-answering information model includes an intelligent question-answering information network group. An intelligent question-answering information network will output multiple model question-answering results. There is a one-to-one correspondence between the model question-answering results in the model question-answering result group and the model prediction probabilities in the model prediction probability group. The question-and-answer results of each model in the model question-and-answer result set are grouped into similar result groups to obtain each question-and-answer result group; the question-and-answer results in each question-and-answer result group are semantically combined to generate the fusion model question-and-answer result set; the average value of the prediction probabilities of each model in each corresponding model prediction probability group is determined as the fusion prediction probability to obtain the corresponding fusion prediction probability set. The fusion model question-and-answer results in the fusion model question-and-answer result set correspond to the fusion prediction probabilities in the fusion prediction probability set. From the set of fusion model question and answer results, select the fusion model question and answer results whose corresponding fusion prediction probability is greater than or equal to the preset probability, and use them as the actual model question and answer results to obtain the actual model question and answer result group; Extract the common content from the question-and-answer results of each actual model as the main intelligent question-and-answer response information; extract the different content from the question-and-answer results of each actual model as the branch intelligent question-and-answer response information; merge the main intelligent question-and-answer response information and the branch intelligent question-and-answer response information into the target intelligent question-and-answer response information; The target intelligent question-and-answer response information is converted into intelligent question-and-answer response voice, and the target intelligent question-and-answer response information and the intelligent question-and-answer response voice are sent to the target terminal.

2. The method according to claim 1, wherein, The intelligent question-answering information model is obtained through the following steps: An initial intelligent question-answering information model is determined, wherein the initial intelligent question-answering information model includes: an initial intelligent question-answering information network group, each initial intelligent question-answering information network corresponding to a question-answering response method, and the question-answering response methods are divided into: single question answer, multiple question answers, single question answer and answer analysis, and multiple question answers and answer analysis methods; For each initial intelligent question-answering information network in the initial intelligent question-answering information network group, perform the following training steps: Obtain a smart question-answering information sample set corresponding to the initial smart question-answering information network, wherein the question-answering response method corresponding to the smart question-answering information sample set is the same as the question-answering response method corresponding to the initial smart question-answering information network; Select target intelligent question and answer information samples from the intelligent question and answer information sample set; The question information included in the target intelligent question-answering information sample is input into the initial intelligent question-answering information network to obtain the initial intelligent response result; Determine the loss value between the initial intelligent response result and the sample label corresponding to the target intelligent question-and-answer information sample; In response to determining that the loss value is less than or equal to a preset loss value, the initial intelligent question-answering information network is determined as a trained intelligent question-answering information network; The various trained intelligent question-answering information networks are merged into an intelligent question-answering information model.

3. The method according to claim 2, wherein, The intelligent customer service system based on a microservice architecture includes: a presentation layer, an application layer, a service layer, and a data layer. The presentation layer includes: a user terminal and an information display terminal. The application layer includes: a communication client, an agent terminal, a backend management terminal, a visitor terminal, a softphone, a call monitoring terminal, and a call detail record management terminal. The service layer includes: a communication server, a backend system management server, an agent management server, a visitor terminal server, a knowledge base operation management server, a backend management new function server, an agent management new function server, an information publishing management server, an intelligent knowledge base operation management function server, and an online customer service outbound call management function server. The data layer includes: a relational structured database, an unstructured database, an in-memory database, a distributed file library, and a distributed log library.

4. The method according to claim 3, wherein, The method further includes: In response to the determination that the outbound call processing stage is not in the initial start stage, an outbound call prediction volume and a pending outbound call volume are generated based on the outbound call indicator set corresponding to the current time. The outbound call indicator set includes: average agent idle time, number of agent calls, number of outbound agents, outbound call processing volume, and outbound call connection rate. The outbound call processing stage is the operation stage in which outbound call processing operations are performed within a preset time period. The initial start stage represents the stage where outbound call processing has just begun. The outbound call processing operation involves making calls to customers at the agent's end. The outbound call indicator set represents outbound calls performed before the current time. The outbound call metrics set generated by the operation includes: outbound call prediction volume (the number of calls to be made within a preset time period), outbound call pending volume (the number of outbound calls that have been dialed and are waiting to be connected at the current moment), average agent idle time (the time during which agents do not make calls and are in an idle state), outbound call processing volume (the amount of event processing performed after a call), outbound call connection rate (the average connection rate during the outbound call process), number of agents in a call (the number of agents currently in a call), and number of outbound call agents (the number of agents that support outbound call operations within a preset time period). Based on the predicted outbound call volume and the outbound call volume to be processed, the actual outbound call volume for the corresponding agent is generated; Based on the actual outbound call volume, the agent terminal performs the outbound call operation corresponding to the outbound call processing stage; The step of generating the actual outbound call volume for the corresponding agent based on the predicted outbound call volume and the outbound call volume to be processed includes: In response to determining that the number of outbound call seats is less than the preset number of seats and that the number of outbound call seats is within the first number of seats range, the number of outbound call seats is determined as the actual outbound call volume; In response to determining that the number of outbound call seats is less than the preset number of seats and the number of outbound call seats is within the second number of seats interval, the outbound call processing volume and the outbound call connection rate are weighted and summed according to the first preset ratio corresponding to the outbound call processing volume and the second preset ratio corresponding to the outbound call connection rate to generate a first outbound call prediction volume, wherein the value in the second number of seats interval is greater than the value in the first number of seats interval; The sum of the outbound call volume to be processed and the first outbound call prediction volume is determined as the actual outbound call volume; In response to determining that the number of outbound call seats is less than the preset number of seats and the number of outbound call seats is within a third number of seats interval, the product of the outbound call processing volume and the first ratio is determined as the adjusted predicted outbound call volume, wherein the value in the third number of seats interval is greater than the value in the second number of seats interval. The sum of the outbound call volume to be processed and the adjusted predicted outbound call volume is determined as the actual outbound call volume; In response to determining that the number of outbound call seats is less than the preset number of seats and the number of outbound call seats is within the fourth number of seats range, the product of the outbound call processing volume and the second ratio is determined as the adjusted outbound call volume, wherein the value in the fourth number of seats range is greater than the value in the third number of seats range, and the second ratio is greater than the first ratio. The sum of the outbound call volume to be processed and the adjusted outbound call volume is determined as the actual outbound call volume.

5. A microservice-based intelligent customer service question-and-answer device, applied to a microservice-based intelligent customer service system, comprising: The determining unit is configured to, in response to receiving a question message sent by a target terminal, determine the similarity between each question and answer message in a preset question and answer information database and the question message, wherein the target terminal refers to a user terminal at a certain seat in the agent terminal; The first generation unit is configured to, in response to determining that the similarity is greater than or equal to a preset similarity, generate an intelligent question-answering output information set for the question information using a pre-trained intelligent question-answering information model. Each intelligent question-answering output information set includes: a model question-answering result set and a model prediction probability set. The intelligent question-answering information model includes: an intelligent question-answering information network set. An intelligent question-answering information network will output multiple model question-answering results. There is a one-to-one correspondence between the model question-answering results in the model question-answering result set and the model prediction probabilities in the model prediction probability set. The fusion unit is configured to group the question-and-answer results of each model in the model question-and-answer result set into similar result groups to obtain each question-and-answer result group; to perform semantic combination of the question-and-answer results in each question-and-answer result group to generate a fusion model question-and-answer result set; and to determine the average value of the model prediction probabilities in each corresponding model prediction probability group as the fusion prediction probability to obtain the corresponding fusion prediction probability set. The fusion model question-and-answer results in the fusion model question-and-answer result set correspond to the fusion prediction probabilities in the fusion prediction probability set. The second generation unit is configured to filter out fusion model question-and-answer results with corresponding fusion prediction probabilities greater than or equal to preset probabilities from the fusion model question-and-answer result set, and use them as actual model question-and-answer results to obtain actual model question-and-answer result groups; The third generation unit is configured to extract the common content from the question-and-answer results of each actual model as the main intelligent question-and-answer response information; extract the different content from the question-and-answer results of each actual model as the branch intelligent question-and-answer response information; and merge the main intelligent question-and-answer response information and the branch intelligent question-and-answer response information into the target intelligent question-and-answer response information. The conversion unit is configured to convert the target intelligent question-and-answer response information into intelligent question-and-answer response speech, and to send the target intelligent question-and-answer response information and the intelligent question-and-answer response speech to the target terminal.

6. A computer device, comprising: One or more processors; A storage device on which one or more programs are stored; When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-4.

7. A computer-readable medium having a computer program stored thereon, wherein, When the program is executed by the processor, it implements the method as described in any one of claims 1-4.

Images