AI customer service interaction control method and system, terminal and storage medium
Through real -time analysis of calling voice and work order status, it is decided to turn into artificial customer service, solve the problem of poor existing AI intelligent customer service processing effects, and improve customer experience and service efficiency.
Patent Information
- Application Number
- CN202411503304.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-25
- Publication Date
- 2025-01-28
Smart Images
Figure CN119383264A_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the technical field of AI intelligent customer service, and in particular to an AI customer service interaction control method, system, terminal and storage medium. Background Art
[0002] With the continuous advancement of science and technology, artificial intelligence technology has been widely used in various industries. More and more companies have enabled AI intelligent customer service to answer customers' questions when they have consultation needs.
[0003] In related technologies, when a customer asks a question, AI intelligent customer service automatically extracts keywords from the content sent or expressed by the customer, matches the corresponding reply words in the local knowledge base based on the keywords, and then replies to the customer based on the matched reply words. When the number of times that the AI intelligent customer feedback content does not accurately reply to the customer exceeds the number threshold, it will automatically switch to manual customer service.
[0004] In the above process, the way AI intelligent customer service handles customer needs is relatively simple, which makes it difficult to meet the needs of multiple scenarios and the processing effect is also poor. Summary of the invention
[0005] In order to enable customers to have a better experience when consulting, this application proposes an AI customer service interaction control method.
[0006] In a first aspect of the present application, an AI customer service interaction control method is provided, the method comprising: Get real-time call audio; Recognize the call voice, generate voice text, and obtain sound features; Analyzing the emotional state of the customer based on the voice text and the sound characteristics; Analyzing the processing progress according to the voice text and the obtained work order status, where the work order is a document for recording, tracking and processing customer needs; Determining whether to transfer to manual customer service according to the emotional state and the processing progress; Call human customer service when you need to be transferred to human customer service.
[0007] By adopting the above technical solution, after receiving the customer service call, the real-time call voice is obtained, the call voice is recognized, converted into voice text, and the sound features are identified. The customer's emotional state is analyzed based on the voice text and sound features, and the processing progress is analyzed based on the voice text and work order status. Determine whether it is necessary to switch from AI customer service to manual customer service based on the emotional state and processing progress. By analyzing the customer's emotional state and current processing progress, and switching to manual customer service when appropriate, it can meet the needs of customers in more scenarios, solve customer needs more efficiently, and increase customer experience.
[0008] In a possible implementation, the voice features include volume, speech speed, and pitch, and analyzing the customer's emotional state based on the voice text and the voice features includes: Extracting keywords from the speech text; Quantifying the keywords, the volume, the speech speed and the pitch to obtain a multidimensional matrix; The multidimensional matrix is input into a trained neural network to obtain the emotional state.
[0009] By adopting the above technical solution, the emotional state can be analyzed more accurately.
[0010] In a possible implementation, the keyword, the volume, the speech speed and the pitch are digitized to obtain a multidimensional matrix: Perform dimensionality reduction processing on the multidimensional matrix.
[0011] By adopting the above technical solution, the occupied computing resources are reduced.
[0012] In a possible implementation, the neural network training method includes: Acquire a number of sound slice samples and a classification label of each of the sound slice samples; Inputting the plurality of sound slice samples into the neural network to obtain an output result corresponding to each of the sound slice samples; Calculating a loss function according to the deviation between the classification label and the output result, and adjusting the network parameters of the neural network according to the loss function; The above steps are repeatedly performed until the recognition rate of the neural network for the emotional state reaches a preset value.
[0013] In a possible implementation, the analyzing the processing progress according to the voice text and the acquired work order status includes: Extracting key sentences from the speech text; When the customer repeats the same question twice or rejects the proposed solution multiple times according to the key statement analysis, the processing progress is abnormal; When the work order status is abnormal, the processing progress is abnormal.
[0014] In a possible implementation, before obtaining the real-time call voice, the following steps are performed: When answering a customer's call, obtain the customer's contact number; Matching the communication number with a preset number library to obtain a matching result; Obtaining a customer profile according to the matching result; A customer service method is assigned based on the customer profile.
[0015] By adopting the above technical solution, AI customer service or manual customer service can be assigned to customers according to their customer profiles, so that services can be provided to each customer more reasonably and their needs can be met.
[0016] In a possible implementation: after switching to manual customer service, the voice text is analyzed to determine the customer's problem and the corresponding solution; The customer problems and the corresponding solutions are stored in a processing library.
[0017] By adopting the above technical solutions, AI customer service can automatically learn the way human customer service handles problems and the corresponding solutions, thereby improving the ability of AI customer service to handle problems.
[0018] In a second aspect of the present application, an AI customer service interaction control system is provided, the system comprising: The acquisition module is used to obtain the real-time call voice; A recognition module, used to recognize the call voice, generate voice text, and obtain sound features; A sentiment analysis module, used to analyze the emotional state of the customer based on the voice text and the sound features; A progress analysis module, used to analyze the processing progress according to the voice text and the acquired work order status, wherein the work order is a document for recording, tracking and processing customer needs; a determination module, used to determine whether to transfer to manual customer service according to the emotional state and the processing progress; and The switching module is used to call human customer service when a call to human customer service is required.
[0019] In a third aspect of the present application, there is provided an intelligent terminal comprising a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute the above-mentioned AI customer service interaction control method.
[0020] In a fourth aspect of the present application, a computer-readable storage medium is provided, storing a computer program that can be loaded by a processor and execute any of the above-mentioned AI customer service interaction control methods.
[0021] In summary, the present application includes at least one of the following beneficial technical effects: In this application, after receiving the customer service call from the customer, the real-time call voice is obtained, the call voice is recognized, converted into voice text, and the sound features are identified. The customer's emotional state is analyzed based on the voice text and sound features, and the processing progress is analyzed based on the voice text and work order status. Determine whether it is necessary to switch from AI customer service to manual customer service based on the emotional state and processing progress. By analyzing the customer's emotional state and the current processing progress, and switching to manual customer service when appropriate, it can meet the needs of customers in more scenarios, solve customer needs more efficiently, and increase the customer experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] Figure 1 It is a flowchart of an AI customer service interaction control method according to one embodiment of the present application.
[0023] Figure 2 It is a system schematic diagram of an AI customer service interactive control system of one embodiment of the present application.
[0024] Figure 3 It is a schematic diagram of the structure of a smart terminal according to one embodiment of the present application.
[0025] In the figure, 21, acquisition module; 22, recognition module; 23, sentiment analysis module; 24, progress analysis module; 25, determination module; 26, switching module; 301, CPU; 302, ROM; 303, RAM; 304, bus; 305, I / O interface; 306, input part; 307, output part; 308, storage part; 309, communication part; 310, drive; 311, removable media. DETAILED DESCRIPTION
[0026] The present application is further described in detail below in conjunction with the accompanying drawings.
[0027] This specific embodiment is merely an explanation of the present application and is not a limitation of the present application. After reading this specification, those skilled in the art may make modifications to the present embodiment without any creative contribution as needed, but such modifications are protected by the patent law as long as they are within the scope of the claims of the present application.
[0028] In order to make the purpose, technical solution and advantages of the embodiments of the present application clearer, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.
[0029] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.
[0030] The embodiment of the present application provides an AI customer service interaction control method, which is applied to a service platform, can assist in solving various customer needs, is applicable to various industries, and can meet the needs of multiple scenarios, which can enhance the customer's experience during consultation.
[0031] The main process of the AI customer service interaction control method provided in the embodiment of the present application is described as follows.
[0032] like Figure 1 As shown: Step S100: Acquire real-time call audio.
[0033] Among them, the call voice is the conversation between the customer and the customer service after the call is connected. Generally, when a customer calls the customer service phone, the AI customer service is responsible for answering the call and communicating with the customer. Before the customer calls the customer service phone, they usually access the pending user demand data through channels such as multi-channel APP / mini program / public account / web page / telephone, such as data in the form of voice, text, pictures, etc. When accessing the pending user demand data, a unified interface is provided to allow users to directly input the pending user demand data without the need for multi-level menu access, which can improve the efficiency of user demand access.
[0034] Furthermore, when the AI customer service cannot provide a solution to the customer's problem, the customer will request manual service. Therefore, in different scenarios, the call voice can be a conversation between the customer and the AI customer service, a conversation between the customer and the manual customer service, or a conversation between the customer and the AI customer service and the manual customer service in turn.
[0035] In order to provide more targeted services, when a customer calls the customer service hotline, the first thing to do is to determine the customer service method this time, that is, manual customer service or AI customer service.
[0036] Specifically, when answering a customer's call, obtain the customer's contact number.
[0037] Then, the communication number is matched with the preset number library to obtain a matching result. Among them, the number library is a database that records the customer's communication number, including customer information and the corresponding communication number. The number library can be pre-stored in a device with a storage function such as a memory. In a specific embodiment, the number library can be obtained by collecting the customer's registration information. When a customer uses an unfamiliar number to call the customer service phone, the customer's communication number can also be updated to the number library. The matching results are divided into two situations: successful matching and failed matching. When the communication number is matched in the number library, the matching result is a successful matching, and the customer information corresponding to the communication number can be directly fed back at this time. When the communication number is not matched in the number library, the information of failed matching is fed back as the matching result.
[0038] In a specific embodiment, the process of matching the communication number with the number library can be implemented by a binary tree algorithm.
[0039] Furthermore, the customer file is obtained according to the matching result. Among them, the customer file records the customer situation, such as the customer's level and historical service records. The customer's level can be divided into ordinary customers and VIP customers. Historical service records include but are not limited to service time, service status and service satisfaction. The service status is divided into resolved and unresolved. Service satisfaction is divided into satisfied and dissatisfied. For the case of successful matching, the corresponding customer file can be retrieved through the customer information. For the case of failed matching, it can be considered that the customer is calling the customer service phone for the first time. Although there is no customer file corresponding to the customer at this time, the customer's level can be defaulted to an ordinary customer.
[0040] Finally, customer service methods are assigned based on customer profiles. In a specific embodiment, AI customer service can be assigned to ordinary customers, and manual customer service can be assigned to VIP customers. In addition, manual customer service can be assigned to customers who have made complaints within a week. The complaint status is that the service status shows that it is not resolved or the service satisfaction is unsatisfactory.
[0041] Step S200: Recognize the call voice, generate voice text, and obtain sound features.
[0042] There are two methods for recognizing the call voice, namely, recognizing the content of the call voice and recognizing the sound of the call voice, thereby obtaining the sound features. The sound features are the parameters of the call voice. The sound features include but are not limited to volume, pitch and speech speed.
[0043] Specifically, the content of the call speech can be recognized by automatic speech recognition technology to recognize the call speech and convert the call speech into speech text. The sound of the call speech can be recognized by analyzing the amplitude and energy of the call speech to determine the volume, by analyzing the frequency of the call speech to determine the pitch, and by analyzing the phoneme duration, syllable rate, and pronunciation interval of the call speech to determine the speech speed.
[0044] Step S300: Analyze the customer's emotional state based on the voice text and sound features.
[0045] Optionally, step S300 includes the following steps: (step S310 to step S330) Step S310: extracting keywords from the speech text.
[0046] Among them, keywords are words that can reflect the emotional color of customers.
[0047] In a specific embodiment, a comparison table can be pre-set. The comparison table includes a number of words and corresponding emotional colors. Emotional colors can be divided into three types: positive, negative and neutral. In the comparison table, there are many words that can reflect each emotional color. The comparison table can be pre-stored in a device with a storage function such as a memory. Then, the words in the voice text that are consistent with the words in the comparison table are extracted as keywords.
[0048] Step S320: quantify the keywords, volume, speech speed and pitch to obtain a multi-dimensional matrix.
[0049] Among them, the multidimensional matrix is a matrix composed of the results obtained by digitizing keywords, volume, speech rate and pitch. It can be understood that volume, pitch and speech rate are all numerical values and can be directly written into the multidimensional matrix, while keywords need to be processed numerically. There are many ways to digitize keywords, and the embodiment of the present application is explained by taking a processing method as an example. Specifically, each keyword extracted is classified according to the emotional color according to the comparison table, and the number of keywords corresponding to each emotional color is obtained. Then, the number of keywords, volume, speech rate and pitch of different emotional colors are written into the multidimensional matrix in order.
[0050] Of course, we can also only count the number of keywords with negative sentiment. Alternatively, we can divide the comparison table in more detail, divide negative sentiment into multiple levels, and determine the degree of negativity of the words corresponding to each negative sentiment in turn. Then, we can count the number of keywords in each level of keywords with negative sentiment, and then write the number of keywords in each level into the multidimensional matrix.
[0051] Step S330: Input the multidimensional matrix into the trained neural network to obtain the emotional state.
[0052] Since this application requires real-time analysis of the customer's emotional state in order to determine whether to convert AI customer service to manual customer service, in order to simplify the algorithm, the emotional state here can be divided into only angry state and non-angry state.
[0053] It can be understood that by training the neural network, the neural network can be enabled to identify and classify emotional states. That is, when a multidimensional matrix is input into the neural network, the output result of the neural network is the emotional state reflected by the keywords, volume, speech speed and pitch in the multidimensional matrix. In the embodiment of the present application, the neural network is a generative adversarial neural network.
[0054] Specifically, the training process of the neural network is as follows: First, obtain several sound slice samples and the classification label of each sound slice sample. The sound sample slice is a segment of a preset duration captured from each call voice. The preset duration can be set to 5 seconds. The classification label is the emotional state reflected by the sound slice sample. After obtaining several sound slice samples, the staff can identify and annotate the emotional state reflected by the sound slice samples to determine the classification label of each sound slice sample.
[0055] Then, a number of sound slice samples are input into the neural network to obtain an output result corresponding to each sound slice sample.
[0056] The loss function is then calculated based on the deviation between the classification label and the output result, and the network parameters of the neural network are adjusted based on the loss function. The technical means involved in this step are relatively mature technical means in the relevant field, so they will not be described in detail here.
[0057] Repeat the above steps until the recognition rate of the neural network for the emotional state reaches a preset value, and the neural network obtained at this time is a trained neural network. Among them, the recognition rate of the emotional state can be obtained by the number of sound slice samples and the number of sound slice samples that can correctly identify the emotional state, that is, the recognition rate of the emotional state is the ratio of the number of sound slice samples that correctly identify the emotional state to the number of sound slice samples. In the embodiment of the present application, the preset value of the recognition rate of the emotional state is 98%.
[0058] Considering that the more data there is in a multidimensional matrix, the more computing resources it occupies, the multidimensional matrix can also be reduced in dimension through a kernel algorithm before being input into the neural network.
[0059] It should also be noted that since the analysis of emotional state is based on real-time calls, it is necessary to analyze the emotional state at regular intervals. Specifically, the interception interval can be preset, which can be set to 5 seconds. When the call lasts 5 seconds, the entire content of the call voice is identified, the keywords and sound features within these 5 seconds are determined, and then the emotional state is analyzed through the neural network. When the call lasts 10 seconds, the 5-10 second portion of the call voice is identified, the keywords and sound features within these 5 seconds are determined, and then the emotional state is analyzed through the neural network. When the call duration does not reach an integer multiple of the interception interval, the emotional state is not analyzed. And so on, until the call ends. When the time between the last analysis of the emotional state and the end of the call does not reach 5 seconds, the emotional state of the call content within this time period also needs to be analyzed.
[0060] Of course, in other embodiments, natural language processing technology can also be used to analyze the emotional state.
[0061] Step S400: Analyze the processing progress according to the voice text and the acquired work order status.
[0062] It is understood that when customers call customer service, they usually have one or more issues that need to be resolved or consulted. The progress of problem resolution is the processing progress. In different scenarios, the processing progress can be divided into different stages, for example, the processing progress is stagnant due to failure to understand the customer's needs, the processing progress is stagnant due to multiple solutions that cannot meet the customer's needs, or the processing progress is normal.
[0063] The specific method for analyzing the processing progress is as follows: First, extract key sentences in the speech text. Among them, the key sentences are interrogative sentences or declarative sentences with strong emotional colors. Interrogative sentences can be extracted by identifying words with interrogative sentence characteristics. Declarative sentences with strong emotional colors can be determined by evaluating the degree of emotional color of each sentence. Specifically, the number of keywords representing each emotional color in each sentence is determined according to the comparison table, and different scores are assigned to keywords of each emotional color, and then the degree of emotional color of the sentence is determined according to the score of the keyword of each emotional color and the corresponding number of keywords. The degree of emotional color can be presented in the form of scores. Further, the average value of the emotional color degree of all current sentences is calculated in real time. If the emotional color degree of the current sentence is higher than the fluctuation range of the average value, it means that the emotional color is relatively strong. The fluctuation range of the average value can be set to 100%+5%. It should be noted that when judging whether a sentence is a declarative sentence with a strong emotional color, all interrogative sentences need to be excluded. That is, only the emotional color degree of all declarative sentences is evaluated.
[0064] Then, the content of the key sentences is analyzed. If the analysis shows that the customer repeats the same question twice, it means that the processing progress is stagnant due to the failure to understand the customer's needs. In this case, the processing progress is considered to be abnormal. In addition, when the analysis shows that multiple solutions proposed for the customer's questions are rejected multiple times, it means that none of the multiple solutions can meet the customer's needs, resulting in a stagnation of the processing progress. In this case, the processing progress is also considered to be abnormal.
[0065] It should be noted that during the communication between the customer and the AI customer service, the AI customer service may need to provide a specific solution. When the AI customer service provides a specific solution, the solution will be sent to the customer's mobile phone during the call for the customer to confirm.
[0066] Furthermore, the current work order status needs to be considered when analyzing the processing progress. A work order is a document that records, tracks, and processes customer needs. It is understood that when a customer calls the customer service number and is connected, the system automatically creates a work order and fills it out according to the content of the voice text. However, if the system fails and the work order cannot be generated, the work order status is abnormal. At the same time, the processing progress is also abnormal. On the contrary, the processing progress is normal.
[0067] Step S500: Determine whether to transfer to manual customer service based on the emotional state and processing progress.
[0068] Specifically, when the emotional state is angry or the processing progress is abnormal, it will be transferred to manual customer service. It should be noted that when the emotional state is angry through neural network analysis, the staff needs to further determine whether the customer is in an angry state. When the staff confirms that the customer is in an angry state, it is necessary to transfer the AI customer service to manual customer service.
[0069] Step S600: Calling manual customer service when transfer to manual customer service is required.
[0070] It is understandable that switching AI customer service to manual customer service is to better help customers solve problems and solve some problems that AI customer service cannot solve. After switching AI customer service to manual customer service, it is also necessary to obtain the call voice between the customer and the manual customer service and convert the call voice into voice text. Specifically, it is necessary to extract the questions raised by the customer and the solutions proposed by the manual customer service from the voice text, and then store them in the processing library for AI customer service to learn. Of course, in the process of communication between manual customer service and customers, according to customer needs. You can also select the most matching words from the script library and recommend it to the manual customer service work interface, so that the manual customer service can modify the words and then respond.
[0071] At the same time, the corresponding content is extracted from the voice text to fill in the work order. When the customer confirms the solution proposed by the manual customer service, the work order is completed, and the work order task is generated and automatically transferred. The work order flow strategy of the work order task can be configured in advance according to the needs. When the work order task is issued to the staff, the staff will mark the work order status. The marked content can be other content such as resolved, unresolved, etc. Among them, for the work orders marked as unresolved, they will also be pushed to the manual customer service for follow-up.
[0072] In the process of serving customers, we will also receive complaints from customers. For these complaints, AI customer service can be adjusted through statistics and customer feedback. That is, the multi-branch tree algorithm is used to adjust the broadcast order of the broadcast content in each branch. Specifically, when a customer inquires or complains, the branch weight will be increased, and the functions commonly used by customers will be broadcast in advance to dynamically optimize the processing flow. Of course, by taking statistics on customer cases, the multi-branch tree branches can also be weighted based on factors such as seasons and related activities, the causes of complaints can be analyzed, and relevant reports can be issued to the leadership. In addition, the entire process of creating and allocating work orders, transferring and changing work orders, and processing and replying to work orders can also generate visual reports, which can be pushed to designated departments as a basis for assessment, so as to implement SLA settings and reminders at each stage.
[0073] AI customer service can not only realize the above functions, but also call intelligent outbound calls to perform corresponding surveys and other tasks based on surveys or other outbound call needs, and statistically process relevant data.
[0074] It is worth noting that the AI customer service interaction control method provided in this application can be applied to usage scenarios in various industries such as smart government affairs, follow-up work orders, disaster warnings, maintenance work orders, smart water affairs, smart drainage, budget work orders, property work orders, field service work orders, and fault work orders. Before putting the service platform into use, it is first necessary to obtain the scenario information of using AI customer service, and then determine the work order processing flow configuration data that matches the scenario information, and then configure the work order system according to the work order processing flow configuration data. In a specific embodiment, the work order system supports work order classification customization function, work order template customization function, field customization function, form combination customization function, and report customization function.
[0075] Figure 2 An AI customer service interactive control system is provided for one embodiment of the present application.
[0076] The cloud service-based pipeline inspection data analysis method provided in the embodiment of the present application can be well applied to the above-mentioned scenario, and can achieve the effect of improving inspection efficiency and shortening inspection cycle.
[0077] like Figure 2The AI customer service interactive control system shown includes an acquisition module 21, a recognition module 22, a sentiment analysis module 23, a progress analysis module 24, a determination module 25 and a switching module 26, wherein: The acquisition module 21 is used to acquire the real-time call voice.
[0078] The recognition module 22 is used to recognize the call voice, generate voice text, and obtain sound features.
[0079] The emotion analysis module 23 is used to analyze the emotional state of the customer based on the voice text and the sound features.
[0080] The progress analysis module 24 is used to analyze the processing progress according to the voice text and the acquired work order status, wherein the work order is a document for recording, tracking and processing customer needs.
[0081] The determination module 25 is used to determine whether to transfer to manual customer service according to the emotional state and the processing progress.
[0082] The switching module 26 is used to call the manual customer service when it is necessary to switch to the manual customer service.
[0083] This embodiment is only one possible implementation method and does not limit other possible implementation methods.
[0084] Figure 3 A schematic diagram of the structure of an intelligent terminal suitable for implementing the embodiments of the present application is shown.
[0085] like Figure 3 As shown, the intelligent terminal includes a central processing unit (CPU) 301, which can perform various appropriate actions and processes according to the program stored in the read-only memory (ROM) 302 or the program loaded from the storage part to the random access memory (RAM) 303. In the RAM 303, various programs and data required for system operation are also stored. The CPU 301, ROM 302 and 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.
[0086] The following components are connected to the I / O interface 305: an input section 306 including a keyboard, a mouse, etc.; an output section 307 including a cathode ray tube (CRT), a liquid crystal display (LCD), etc., and a speaker, etc.; a storage section 308 including a hard disk, etc.; and a communication section 309 including a network interface card such as a LAN card, a modem, etc. The communication section 309 performs communication processing via a network such as the Internet. A drive 310 is also connected to the I / O interface 305 as needed. A removable medium 311, such as a magnetic disk, an optical disk, a magneto-optical disk, a semiconductor memory, etc., is installed on the drive 310 as needed, so that a computer program read therefrom is installed into the storage section 308 as needed.
[0087] In particular, according to an embodiment of the present application, the above reference flow chart Figure 1 The described process can be implemented as a computer software program. For example, an embodiment of the present application includes a computer program product, which includes a computer program carried on a machine-readable medium, and the computer program contains program code for executing the method shown in the flow chart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part 309, and / or installed from the removable medium 311. When the computer program is executed by the central processing unit (CPU) 301, the above-mentioned functions defined in the system of the present application are executed.
[0088] It should be noted that the computer-readable medium shown in the present application may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, 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), 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 the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present application, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer-readable medium may be transmitted using any appropriate medium, including but not limited to: wireless, wire, optical cable, RF, etc., or any suitable combination of the above.
[0089] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present application. 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 aforementioned module, a 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 and / or flow chart, and the combination of the boxes in the block diagram and / 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.
[0090] The units or modules involved in the embodiments described in the present application may be implemented by software or hardware. The units or modules described may also be set in a processor. For example, they may be described as: a processor including: an acquisition module 21, a recognition module 22, a sentiment analysis module 23, a progress analysis module 24, a determination module 25, and a switching module 26. The names of these units or modules do not, in some cases, constitute limitations on the units or modules themselves. For example, the acquisition module 21 may also be described as a "module for acquiring real-time call voice."
[0091] As another aspect, the present application further provides a computer-readable storage medium, which may be included in the intelligent terminal described in the above embodiment; or may exist independently and not be assembled into the intelligent terminal. The above computer-readable storage medium stores one or more programs, and when the above programs are used by one or more processors to execute the AI customer service interaction control method described in the present application.
[0092] The above description is only a preferred embodiment of the present application and an explanation of the technical principles used. Those skilled in the art should understand that the scope of application involved in the present application is not limited to the technical solution formed by a specific combination of the above technical features, but should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the aforementioned application concept. For example, the above features are replaced with (but not limited to) technical features with similar functions applied in the present application.
Claims
1. An AI customer service interaction control method, characterized in that: include: Get real-time call audio; Recognize the call voice, generate voice text, and obtain sound features; Analyzing the emotional state of the customer based on the voice text and the sound characteristics; Analyzing the processing progress according to the voice text and the obtained work order status, where the work order is a document for recording, tracking and processing customer needs; Determining whether to transfer to manual customer service according to the emotional state and the processing progress; Call human customer service when you need to be transferred to human customer service.
2. The AI customer service interaction control method according to claim 1, characterized in that: The voice features include volume, speaking speed and pitch, and the analyzing the customer's emotional state according to the voice text and the voice features includes: Extracting keywords from the speech text; Quantifying the keywords, the volume, the speech speed and the pitch to obtain a multidimensional matrix; The multidimensional matrix is input into a trained neural network to obtain the emotional state.
3. The AI customer service interaction control method according to claim 2, characterized in that: After the keywords, the volume, the speech speed and the pitch are digitized to obtain a multidimensional matrix, we have: Perform dimensionality reduction processing on the multidimensional matrix.
4. The AI customer service interaction control method according to claim 2, characterized in that: The training method of the neural network includes: Acquire a number of sound slice samples and a classification label of each of the sound slice samples; Inputting the plurality of sound slice samples into the neural network to obtain an output result corresponding to each of the sound slice samples; Calculating a loss function according to the deviation between the classification label and the output result, and adjusting the network parameters of the neural network according to the loss function; The above steps are repeatedly performed until the recognition rate of the neural network for the emotional state reaches a preset value.
5. The AI customer service interaction control method according to claim 1, characterized in that: The analyzing the processing progress according to the voice text and the acquired work order status includes: Extracting key sentences from the speech text; When the customer repeats the same question twice or rejects the proposed solution multiple times according to the key statement analysis, the processing progress is abnormal; When the work order status is abnormal, the processing progress is abnormal.
6. The AI customer service interaction control method according to claim 1, characterized in that: Before obtaining the real-time call audio: When answering a customer's call, obtain the customer's contact number; Matching the communication number with a preset number library to obtain a matching result; Acquire the customer profile according to the matching result; A customer service method is assigned based on the customer profile.
7. The AI customer service interaction control method according to claim 1, characterized in that: Also includes: After switching to manual customer service, the voice text is analyzed to determine the customer's problem and the corresponding solution; The customer problems and the corresponding solutions are stored in a processing library.
8. An AI customer service interactive control system, characterized in that: include, An acquisition module (21), used for acquiring real-time call voice; A recognition module (22), used for recognizing the call voice, generating voice text, and obtaining sound features; A sentiment analysis module (23), used for analyzing the emotional state of the customer based on the speech text and the sound features; A progress analysis module (24), used for analyzing the processing progress according to the voice text and the acquired work order status, wherein the work order is a document for recording, tracking and processing customer needs; A determination module (25), configured to determine whether to transfer to manual customer service according to the emotional state and the processing progress; as well as, The switching module (26) is used to call the manual customer service when it is necessary to switch to the manual customer service.
9. An intelligent terminal, characterized in that: It comprises a memory and a processor, wherein the memory stores a computer program that can be loaded by the processor and execute the AI customer service interaction control method as described in any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that: A computer program is stored which can be loaded by a processor and execute the AI customer service interaction control method as described in any one of claims 1 to 7.