Customer service question and answer method, device and equipment based on natural language processing and medium

Abstract

The embodiment of the invention relates to the technical field of artificial intelligence, and discloses a customer service question answering method and device based on natural language processing, equipment and a medium. The method comprises the following steps: acquiring dialogue context information in question and answer interaction between a user and a customer service; performing service classification according to the dialogue context information to obtain a service classification result; wherein the service classification result comprises account controlled services, website information services and general dialogue services; and calling a service agent under the corresponding service according to a service classification result, and performing information identification on the dialogue context information through the service agent to obtain a question and answer reply message. By dividing customer service questions and answers in multiple service scenes, the user intention can be identified more accurately, and question and answer reply is finer and more reliable.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology and can be applied to the field of financial technology, and in particular to a customer service question and answer method, device, equipment and medium based on natural language processing. Background Art

[0002] With the development of computer technology, financial institutions' services have gradually become information-based and intelligent. Users can conduct online business without leaving home. As a result, the demand for online customer service has greatly increased. However, due to the limited number of manual customer service and the long service time each time, users may have to wait in line for a long time during the peak consultation period.

[0003] In order to improve service quality, intelligent customer service technology has emerged. However, the existing intelligent customer service technology can only recover by keywords, and the degree of intelligence is insufficient. It is not accurate enough to identify the user's intention and often gives irrelevant answers. Especially in special application scenarios, there is a lack of processing solutions. Summary of the invention

[0004] The present invention provides a customer service question and answer method, device, equipment and medium based on natural language processing to more accurately identify user intentions and make question and answer responses more precise and reliable.

[0005] According to one aspect of the present invention, a customer service question and answer method based on natural language processing is provided, the method comprising:

[0006] Obtain conversation context information during Q&A interaction between users and customer service staff;

[0007] Classifying services according to the conversation context information to obtain service classification results; wherein the service classification results include: account-controlled services, branch information services, and general conversation services;

[0008] The business agent corresponding to the business is retrieved according to the business classification result, and the conversation context information is identified by the business agent to obtain a question and answer reply message.

[0009] According to another aspect of the present invention, a customer service question-and-answer device based on natural language processing is provided, the device comprising:

[0010] The dialogue context information acquisition module is used to obtain the dialogue context information in the question-and-answer interaction between the user and the customer service;

[0011] A service classification result determination module, configured to classify services according to the conversation context information to obtain service classification results; wherein the service classification results include: account-controlled services, branch information services, and general conversation services;

[0012] The question and answer reply message determination module is used to retrieve the business intelligence entity under the corresponding business according to the business classification result, and obtain the question and answer reply message by performing information recognition on the dialogue context information through the business intelligence entity.

[0013] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0014] at least one processor; and

[0015] a memory communicatively connected to the at least one processor; wherein,

[0016] The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the customer service question and answer method based on natural language processing described in any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, wherein the computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the customer service question and answer method based on natural language processing described in any embodiment of the present invention when executed.

[0018] According to another aspect of the present invention, a computer program product is provided, including a computer program, which, when executed by a processor, implements the customer service question and answer method based on natural language processing described in any embodiment of the present invention.

[0019] The technical solution of the embodiment of the present invention obtains the conversation context information in the question and answer interaction between the user and the customer service; classifies the business according to the conversation context information to obtain the business classification result; wherein the business classification result includes: account controlled business, branch information business, and general conversation business; according to the business classification result, the business intelligence body under the corresponding business is retrieved, and the conversation context information is identified by the business intelligence body to obtain the question and answer reply message, which solves the problem that the customer service question and answer is not smart enough. By dividing the customer service question and answer into multiple business scenarios, the user intention can be identified more accurately, making the question and answer reply more precise and reliable.

[0020] It should be understood that the contents described in this section are not intended to identify the key or important features of the embodiments of the present invention, nor are they intended to limit the scope of the present invention. Other features of the present invention will become easily understood through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0022] Figure 1 is a flow chart of a customer service question and answer method based on natural language processing provided according to an embodiment of the present invention;

[0023] Figure 2 This is an application example diagram of a customer service question-and-answer method based on natural language processing provided according to an embodiment of the present invention;

[0024] Figure 3 is a structural schematic diagram of a customer service question-and-answer device based on natural language processing provided according to an embodiment of the present invention;

[0025] Figure 4 It is a structural schematic diagram of an electronic device for implementing the customer service question-and-answer method based on natural language processing according to an embodiment of the present invention. DETAILED DESCRIPTION

[0026] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0027] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0028] Figure 1This is a flowchart of a customer service question and answer method based on natural language processing provided according to an embodiment of the present invention. This embodiment is applicable to situations where users interact with online customer service questions and answers. The method can be executed by a customer service question and answer device based on natural language processing. The customer service question and answer device based on natural language processing can be implemented in the form of hardware and / or software. The customer service question and answer device based on natural language processing can be configured in an electronic device, which can be a mobile device such as a mobile phone, a tablet computer (PAD), a wearable device, or a personal computer (PC, Personal Computer), etc. Figure 1 As shown, the method includes:

[0029] Step 110: Obtain conversation context information in the question-and-answer interaction between the user and the customer service.

[0030] The conversation context information includes the current information of the user's interaction with the customer service Q&A, and may also include the historical information of the user's interaction with the customer service Q&A. For example, the conversation context information may be the current information of the user's interaction with the customer service Q&A and the four pieces of information before the current information. The conversation context information may be text information obtained based on multimodal information. For example, the conversation context information may be text information obtained based on one or more of text, language, and pictures. The acquisition of the conversation context information is only obtained after the user's authorization, and the acquisition method is reasonable and legal.

[0031] In an optional implementation of an embodiment of the present invention, obtaining conversation context information in a question-and-answer interaction between a user and a customer service representative includes: obtaining multimodal conversation information in a question-and-answer interaction between a user and a customer service representative; and converting both voice information and image information in the multimodal conversation information into text information to obtain conversation context information.

[0032] Among them, converting voice information into text information can be to perform noise reduction processing on voice information in multimodal dialogue information through spectral subtraction, divide the audio signal into short frame audio, and use fast Fourier transform to convert the short frame audio in the time domain into a frequency domain signal; filter the frequency domain signal through a Mel filter, and use discrete cosine transform to compress the signal to obtain audio features; finally, convert the audio features into text data.

[0033] Specifically, common non-human audio, such as car horns and pet calls, can be extracted through spectral subtraction, and then the non-human noise spectrum is subtracted from the input audio spectrum to achieve noise reduction processing of speech information. When segmenting the audio signal, each frame can contain 500 milliseconds of audio data. In order to reduce the edge effect between frames, a window function, such as a Hamming window, can be applied to each frame. Features that are more relevant to the human auditory system can be extracted through a Mel filter. Converting audio features into text data can be achieved through a pre-trained network model. For example, audio features are converted into text data through a converter model (Transformer model) in a neural network. In order to ensure the reliability of the conversion, a Transformer model can be generated by training a large amount of annotated speech data and annotated text data. During the training process, the model parameters can be optimized through the back propagation algorithm.

[0034] The conversion of image information into text information can be achieved by using a pre-trained convolutional neural network (CNN) and / or recurrent neural network (RNN). Exemplarily, historical image information in customer service interactions can be used for annotation, such as annotation of text bounding boxes, character categories, and text sequences. CNN and RNN are trained, CNN is used to extract features from images, and RNN is used to perform sequence modeling on the extracted features to capture the dependencies between characters. During training, the Connectionist Temporal Classification (CTC) loss function can be used for model optimization to generate text sequences.

[0035] By converting voice information and image information into text information in different ways, we can ensure that the source of conversation context information is rich and comprehensive, improve the accuracy of text information determination, and thus improve the intelligence of customer service questions and answers and achieve accurate responses.

[0036] Step 120: Classify services according to the conversation context information to obtain service classification results.

[0037] Among them, the business classification results include: account-controlled business, branch information business, and general conversation business. The business classification of the conversation context information can be obtained based on the classification of the pre-trained support vector machine (SVM) model. Specifically, it can be labeled according to the historical conversation information in the question-and-answer interaction between the user and the customer service, and labeled as account-controlled class, branch information class or general conversation class. Then, the text is converted into a numerical feature vector using a bag-of-words model, and then the support vector machine is used for model training. Therefore, the trained SVM can be used to classify the conversation context information.

[0038] The technical solution of the embodiment of the present invention can be applied to the field of financial technology, and the business classification can be a business in the field of financial technology. Specifically, the account controlled business can refer to the restrictions on the use of the user's account in the financial institution. The branch information business can refer to the user's desire to find information related to the financial institution's branches. The general dialogue business can refer to the user asking questions about problems that arise in the daily business processing of the financial institution. By classifying the business, different question and answer replies can be made under different businesses, the reply granularity can be refined, and the reply accuracy can be improved.

[0039] Step 130: retrieve the business agent corresponding to the business according to the business classification result, identify the conversation context information through the business agent, and obtain the question and answer reply message.

[0040] Among them, the business intelligent entities corresponding to different business classification results may be different. Figure 2 FIG. 1 is an application example diagram of a customer service question-and-answer method based on natural language processing provided according to an embodiment of the present invention. Figure 2 As shown, when conducting customer service Q&A, the conversation context information can be obtained, and the conversation context information can be classified into business categories to obtain the business classification results. When the business classification result is an account-controlled business, the conversation context information is input into the controlled intelligent agent to obtain the question and answer reply message corresponding to the account-controlled business. When the business classification result is a branch information business, the conversation context information is input into the branch intelligent agent to obtain the question and answer reply message corresponding to the branch information business. When the business classification result is a general conversation business, the conversation context information is input into the general intelligent agent to obtain the question and answer reply message corresponding to the general conversation business.

[0041] In the embodiment of the present invention, the controlled agent may be a network model generated by training data corresponding to the account controlled business. The controlled agent is used to query and reply the control status of the user's financial resource account. The branch agent may be a network model generated by training data corresponding to the branch information business. The branch agent is used to query and reply the branch information of the financial institution. The general agent may be a network model generated by training data corresponding to the general dialogue business. The general agent is used to query and reply the solution to the problem related to finance.

[0042] Account controlled services, branch information services, and general conversation services query different databases and have different reply mechanisms. By classifying services, the embodiments of the present invention can achieve detailed replies in the Q&A interaction between users and customer service in actual applications, improve the accuracy of replies, and ensure user satisfaction.

[0043] In an optional implementation of an embodiment of the present invention, a business agent under the corresponding business is retrieved according to the business classification result, and the conversation context information is identified by the business agent to obtain a question and answer reply message, including: when the business classification result is an account-controlled business, the agent corresponding to the account-controlled business is used to perform the following operations: query the corresponding resource identification information according to the user account information; perform keyword extraction on the conversation context information, and when the keyword extraction result matches the resource identification information, query the reason for resource control; and use the reason for resource control as the question and answer reply message.

[0044] Among them, the resource identification information can be the user's financial resource account information in a financial institution, such as bank card, credit card, and mobile phone number. When the user interacts with the customer service, the corresponding resource identification information can be queried according to the user account information. For example, the bank card or credit card information under the name can be queried through the mobile phone number, ID number, or user name provided by the user. When the resource identification information is determined, it can be queried in the database maintained by the financial institution according to the user's authorization. When the keyword extraction result in the dialogue context information matches the resource identification information, the user's resource control reason can be queried. Among them, keyword extraction can be achieved in a variety of ways, such as keyword extraction through a hidden Markov model. Alternatively, the keyword extraction result can be obtained through text recognition technology. Exemplarily, the keywords extracted in the context information can be compared with the user's bank card or credit card. If the comparison is passed, the resource control reason is queried. Afterwards, the resource control reason can be filled in according to the preset template as a question and answer reply message. When the keyword does not pass the comparison with the user's bank card or credit card, the user can be requested to provide the resource identification information to be queried. By operating the intelligent entities corresponding to account-controlled businesses, the user's input information when interacting with customer service can be reduced, which simplifies the interactive operations, improves the accuracy of intent recognition, and enhances user satisfaction.

[0045] In an optional implementation of an embodiment of the present invention, a business agent corresponding to the business is retrieved according to the business classification result, and the conversation context information is identified by the business agent to obtain a question and answer reply message, including: when the business classification result is an outlet information type business, the agent corresponding to the outlet information type business is used to perform the following operations: performing word segmentation processing on the conversation context information to obtain a word segmentation set; when the word segmentation set matches the outlet information in a preset data set, determining the target outlet information; and using the target outlet information as the question and answer reply message.

[0046] Among them, there are many ways to process word segmentation. For example, word segmentation can be performed through a natural language processing model. Exemplarily, the natural language processing model can be a bidirectional encoder (Bidirectional Encoder Representations from Transformers, BERT) model. There are many ways to match the word segmentation set with the network point information in the preset data set, such as determining whether the two match by similarity calculation. Optionally, the Jaccard similarity coefficient or the Jaccard distance is used to determine whether the word segmentation set matches the network point information in the preset data set. The target network point information whose matching degree with the preset data set meets the preset screening conditions can be used as a question and answer reply message. For example, the top 10 network point information that meets the requirements in the context information can be used as a question and answer reply message. Through the operation of the intelligent body corresponding to the network point information business, it is possible to quickly query and return appropriate network point information for users, thereby improving the efficiency and accuracy of information return.

[0047] In an optional implementation of an embodiment of the present invention, a business agent corresponding to the business is retrieved according to the business classification result, and the conversation context information is identified by the business agent to obtain a question and answer reply message, including: when the business classification result is a general conversation type business, the agent corresponding to the general conversation type business is used to perform the following operations: entity recognition is performed on the conversation context information to obtain multiple target entities in the general business; and an entity connection relationship is established for each target entity through a relationship extraction model; when the entity connection relationship matches the target text in a preset corpus, the reply corresponding to the target text is used as the question and answer reply message.

[0048] Among them, entity recognition can be achieved through a named entity recognition algorithm. Entities include but are not limited to: date, finance, account, transfer, deposit, location and other bank-related entities. The relationship extraction model can be a pre-trained network model. For example, the user question, the entity in the question, and the specific relationship between the question and the entity can be manually annotated to generate a sample set. The recurrent neural network model in deep learning is used to build a relationship extraction model based on the sample set. Thus, the relationship extraction model can establish an entity connection relationship based on context information and entity recognition results. For example, the entity connection relationship can be "account-query balance", "transfer-handling fee", "credit card-application process". When matching the target text, the text in the preset corpus can be pre-processed by word segmentation, stop word removal and other pre-processing operations. Then, the entity connection relationship and the text in the corpus are converted into word vectors. For example, the text vector in the preset corpus can be represented by a bag of words model. After that, a similarity calculation algorithm such as cosine similarity is used to determine the similarity between the entity connection relationship and the text in the bag of words model, so that the text with the highest similarity to the entity connection relationship in the bag of words model is used as the target text. Then, a question-answer reply message is generated based on the target text.

[0049] For example, based on the calculated similarity value, the similarity between texts can be determined. The similarity value range is between [-1, 1]. The closer the value is to 1, the more similar the texts are, and the closer the value is to -1, the less similar the texts are. If the similarity is lower than 0.7, the call is automatically transferred to manual customer service. If the similarity is higher than 0.7, the template filling method is used to generate the question and answer reply message.

[0050] Through the operation of the intelligent agent corresponding to the general dialogue business, problem solutions can be provided for the user's basic business services to ensure the accuracy of intent recognition.

[0051] The technical solution of the embodiment of the present invention can improve the precision and accuracy of questions and answers by using intelligent agents corresponding to account-controlled services, branch information services, and general dialogue services to respond to dialogue context information based on business classification results, and can provide solutions for various types of businesses.

[0052] Based on the above implementation, optionally, entity recognition is performed on the conversation context information to obtain multiple target entities in the general business, including: classifying the conversation context information into general business to obtain a general business tag; under the general business tag, performing entity recognition on the conversation context information to obtain multiple target entities in the general business.

[0053] Exemplarily, historical conversation context information in general services, such as answers to common questions, can be collected and annotated with general business tags. General business tags can be tags for bank card transfer-related services, credit card loan-related services, car loan-related services, and mortgage-related services. Then, the data is cleaned and preprocessed, such as removing noise, correcting errors, standardizing text, cleaning, word segmentation, and removing stop words. The importance of words to the text can also be added to the text data, such as measuring the importance of words to the text by word frequency (TF) and inverse document frequency (IDF). After that, the data set can be divided into training set and test set, such as using 80% of the data as the training set and 20% of the data as the test set. Model training is performed using the training set data and the corresponding tags. Classification is performed by finding the optimal hyperplane in the feature space. For example, the soft margin SVM method can be used. During the training process, the parameters of the SVM algorithm are adjusted by the cross-validation method. Thus, the general business classification of the conversation context information can be achieved through the trained SVM.

[0054] When performing target entity recognition, it can be performed under a general business tag, which can refine the granularity of entity recognition and thereby improve the recognition reliability of business-related entities, thereby providing users with accurate question-and-answer response messages.

[0055] The technical solution of this embodiment obtains the conversation context information in the question and answer interaction between the user and the customer service; classifies the business according to the conversation context information to obtain the business classification result; wherein the business classification result includes: account controlled business, branch information business, and general conversation business; according to the business classification result, the business intelligence body under the corresponding business is retrieved, and the conversation context information is identified by the business intelligence body to obtain the question and answer reply message, which solves the problem that the customer service question and answer is not smart enough. By dividing the customer service question and answer into multiple business scenarios, the user intention can be identified more accurately, making the question and answer reply more precise and reliable.

[0056] Figure 3 Schematic diagram of a customer service question-and-answer device based on natural language processing provided according to an embodiment of the present invention. Figure 3 As shown, the device includes: a dialogue context information acquisition module 310, a business classification result determination module 320 and a question and answer reply message determination module 330. Among them:

[0057] The dialogue context information acquisition module 310 is used to acquire dialogue context information in the question-and-answer interaction between the user and the customer service;

[0058] The service classification result determination module 320 is used to classify services according to the conversation context information to obtain service classification results; wherein the service classification results include: account-controlled services, branch information services, and general conversation services;

[0059] The question and answer reply message determination module 330 is used to retrieve the business agent under the corresponding business according to the business classification result, and to obtain the question and answer reply message by performing information recognition on the dialogue context information through the business agent.

[0060] Optionally, the dialog context information acquisition module 310 includes:

[0061] A multimodal dialogue information acquisition unit, used to acquire multimodal dialogue information in the question-and-answer interaction between the user and the customer service;

[0062] The conversation context information acquisition unit is used to convert the voice information and image information in the multimodal conversation information into text information to obtain the conversation context information.

[0063] Optionally, the question and answer reply message determination module 330 includes:

[0064] The first question-answer reply message determination unit is configured to, when the business classification result is an account-controlled business, perform the following operations through an agent corresponding to the account-controlled business:

[0065] Query the corresponding resource identification information according to the user account information;

[0066] Extract keywords from the conversation context information, and when the keyword extraction result matches the resource identification information, query the reason why the resource is controlled;

[0067] The reason why the resource is controlled is used as a question and answer reply message.

[0068] Optionally, the question and answer reply message determination module 330 includes:

[0069] The second question-answer reply message determining unit is configured to, when the service classification result is a branch information service, perform the following operations through an agent corresponding to the branch information service:

[0070] Perform word segmentation on the conversation context information to obtain a word segmentation set;

[0071] When the word segmentation set matches the network point information in the preset data set, the target network point information is determined;

[0072] The target network point information is used as a question and answer reply message.

[0073] Optionally, the question and answer reply message determination module 330 includes:

[0074] The third question-answer reply message determination unit is configured to, when the service classification result is a general conversation service, perform the following operations through an agent corresponding to the general conversation service:

[0075] Perform entity recognition on the conversation context information to obtain multiple target entities in the general business; and establish entity connection relationships for each target entity through a relationship extraction model;

[0076] When the entity connection relationship matches the target text in the preset corpus, the reply corresponding to the target text is used as the question-answer reply message.

[0077] Optionally, the third question-answer reply message determining unit is specifically configured to:

[0078] Performing general business classification on the conversation context information to obtain a general business label;

[0079] Under the general business tag, entity recognition is performed on the conversation context information to obtain multiple target entities in the general business.

[0080] The customer service question and answer device based on natural language processing provided in the embodiment of the present invention can execute the customer service question and answer method based on natural language processing provided in any embodiment of the present invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0081] In the technical solution of the embodiment of the present invention, the collected information is information and data authorized by the user or fully authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of the relevant data comply with the relevant laws, regulations and standards of the relevant countries and regions, take necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation entrances for users to choose to authorize or refuse.

[0082] When big data analysis based on personal information is conducted and artificial intelligence technology is used for automated decision-making, users are provided with corresponding operation entrances for them to choose to agree or reject the automated decision-making results; if the user chooses to reject, the expert decision-making process will be entered.

[0083] Figure 4 A schematic diagram of the structure of an electronic device 10 that can be used to implement an embodiment of the present invention is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processing, cellular phones, smart phones, wearable devices (such as helmets, glasses, watches, etc.) and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely examples and are not intended to limit the implementation of the present invention described and / or required herein.

[0084] like Figure 4As shown, the electronic device 10 includes at least one processor 11, and a memory connected to the at least one processor 11, such as a read-only memory (ROM) 12, a random access memory (RAM) 13, etc., wherein the memory stores a computer program that can be executed by at least one processor, and the processor 11 can perform various appropriate actions and processes according to the computer program stored in the read-only memory (ROM) 12 or the computer program loaded from the storage unit 18 to the random access memory (RAM) 13. In the RAM 13, various programs and data required for the operation of the electronic device 10 can also be stored. The processor 11, the ROM 12, and the RAM 13 are connected to each other through a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0085] A number of components in the electronic device 10 are connected to the I / O interface 15, including: an input unit 16, such as a keyboard, a mouse, etc.; an output unit 17, such as various types of displays, speakers, etc.; a storage unit 18, such as a disk, an optical disk, etc.; and a communication unit 19, such as a network card, a modem, a wireless communication transceiver, etc. The communication unit 19 allows the electronic device 10 to exchange information / data with other devices through a computer network such as the Internet and / or various telecommunication networks.

[0086] The processor 11 may be a variety of general and / or special processing components with processing and computing capabilities. Some examples of the processor 11 include, but are not limited to, a central processing unit (CPU), a graphics processing unit (GPU), various special artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, a digital signal processor (DSP), and any appropriate processor, controller, microcontroller, etc. The processor 11 executes the various methods and processes described above, such as a customer service question-and-answer method based on natural language processing.

[0087] In some embodiments, the customer service question-and-answer method based on natural language processing may be implemented as a computer program, which is tangibly contained in a computer-readable storage medium, such as a storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on the electronic device 10 via the ROM 12 and / or the communication unit 19. When the computer program is loaded into the RAM 13 and executed by the processor 11, one or more steps of the customer service question-and-answer method based on natural language processing described above may be performed. Alternatively, in other embodiments, the processor 11 may be configured to execute the customer service question-and-answer method based on natural language processing in any other appropriate manner (e.g., by means of firmware).

[0088] Various implementations of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), load programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various implementations can include: being implemented in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which can be a special purpose or general purpose programmable processor that can receive data and instructions from a storage system, at least one input device, and at least one output device, and transmit data and instructions to the storage system, the at least one input device, and the at least one output device.

[0089] Computer programs for implementing the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, so that when the computer program is executed by the processor, the functions / operations specified in the flow chart and / or block diagram are implemented. The computer program may be executed entirely on the machine, partially on the machine, partially on the machine and partially on a remote machine as a stand-alone software package, or entirely on a remote machine or server.

[0090] In the context of the present invention, a computer-readable storage medium may be a tangible medium that may contain or store a computer program for use by or in combination with an instruction execution system, device or equipment. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the foregoing. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, 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 foregoing.

[0091] To provide interaction with a user, the systems and techniques described herein may be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and a pointing device (e.g., a mouse or trackball) through which the user can provide input to the electronic device. Other types of devices may also be used to provide interaction with the user; for example, the feedback provided to the user may be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user may be received in any form (including acoustic input, voice input, or tactile input).

[0092] The systems and techniques described herein may be implemented in a computing system that includes backend components (e.g., as a data server), or a computing system that includes middleware components (e.g., an application server), or a computing system that includes frontend components (e.g., a user computer with a graphical user interface or a web browser through which a user can interact with implementations of the systems and techniques described herein), or a computing system that includes any combination of such backend components, middleware components, or frontend components. The components of the system may be interconnected by any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include: a local area network (LAN), a wide area network (WAN), a blockchain network, and the Internet.

[0093] A computing system may include a client and a server. The client and the server are generally remote from each other and usually interact through a communication network. The client and server relationship is generated by computer programs running on the corresponding computers and having a client-server relationship with each other. The server may be a cloud server, also known as a cloud computing server or cloud host, which is a host product in the cloud computing service system to solve the defects of difficult management and weak business scalability in traditional physical hosts and VPS services.

[0094] It should be understood that the various forms of processes shown above can be used to reorder, add or delete steps. For example, the steps described in the present invention can be executed in parallel, sequentially or in different orders, as long as the desired results of the technical solution of the present invention can be achieved, and this document does not limit this.

[0095] The above specific implementations do not constitute a limitation on the protection scope of the present invention. It should be understood by those skilled in the art that various modifications, combinations, sub-combinations and substitutions can be made according to design requirements and other factors. Any modification, equivalent substitution and improvement made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A customer service question and answer method based on natural language processing, characterized in that: include: Obtain conversation context information during Q&A interaction between users and customer service staff; Classifying services according to the conversation context information to obtain service classification results; wherein the service classification results include: account-controlled services, branch information services, and general conversation services; The business agent corresponding to the business is retrieved according to the business classification result, and the conversation context information is identified by the business agent to obtain a question and answer reply message.

2. The method according to claim 1, characterized in that Obtain conversation context information between users and customer service, including: Obtain multimodal dialogue information in the Q&A interaction between users and customer service; The voice information and the image information in the multimodal dialogue information are converted into text information to obtain the dialogue context information.

3. The method according to claim 1, characterized in that Retrieving a business agent corresponding to the business according to the business classification result, and performing information recognition on the conversation context information by the business agent to obtain a question-answer reply message, including: When the business classification result is an account-controlled business, the following operations are performed through the agent corresponding to the account-controlled business: Query the corresponding resource identification information according to the user account information; Perform keyword extraction on the conversation context information, and when the keyword extraction result matches the resource identification information, query the reason why the resource is controlled; The reason why the resource is controlled is used as a question-answer reply message.

4. The method according to claim 1, characterized in that: Retrieving a business agent corresponding to the business according to the business classification result, and performing information recognition on the conversation context information by the business agent to obtain a question-answer reply message, including: When the service classification result is a branch information service, the following operations are performed through the agent corresponding to the branch information service: Performing word segmentation processing on the conversation context information to obtain a word segmentation set; When the word segmentation set matches the network point information in the preset data set, determining the target network point information; The target network point information is used as a question and answer reply message.

5. The method according to claim 1, characterized in that Retrieving a business agent corresponding to the business according to the business classification result, and performing information recognition on the conversation context information by the business agent to obtain a question-answer reply message, including: When the service classification result is a general conversation service, the following operations are performed through the agent corresponding to the general conversation service: Performing entity recognition on the conversation context information to obtain multiple target entities in the general business; and establishing entity connection relationships for each target entity through a relationship extraction model; When the entity connection relationship matches the target text in the preset corpus, the reply corresponding to the target text is used as the question and answer reply message.

6. The method according to claim 5, characterized in that Perform entity recognition on the conversation context information to obtain multiple target entities in the general business, including: Performing general business classification on the conversation context information to obtain a general business label; Under the general business tag, entity recognition is performed on the conversation context information to obtain multiple target entities in the general business.

7. A customer service question and answer device based on natural language processing, characterized in that: include: The dialogue context information acquisition module is used to obtain the dialogue context information in the question-and-answer interaction between the user and the customer service; A service classification result determination module, configured to classify services according to the conversation context information to obtain service classification results; wherein the service classification results include: account-controlled services, branch information services, and general conversation services; The question and answer reply message determination module is used to retrieve the business intelligence entity under the corresponding business according to the business classification result, and obtain the question and answer reply message by performing information recognition on the dialogue context information through the business intelligence entity.

8. An electronic device, characterized in that: The electronic device comprises: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, and the computer program is executed by the at least one processor so that the at least one processor can execute the customer service question and answer method based on natural language processing as described in any one of claims 1-6.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer instructions, and the computer instructions are used to enable a processor to implement the customer service question-and-answer method based on natural language processing according to any one of claims 1 to 6 when executed.

10. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the customer service question-and-answer method based on natural language processing according to any one of claims 1 to 6.

Images