User service response method and device, equipment and medium

Abstract

The invention provides a user service response method and device, equipment and a medium, and the method comprises the steps: obtaining a user request corresponding to a target work order, and extracting a keyword of the user request; identifying the user request according to the keyword, and matching a corresponding service agent for each sub-question when a plurality of sub-questions are identified; and inputting each sub-question into the corresponding matched service agent to obtain a sub-answer corresponding to each sub-question, generating a corresponding target answer according to each sub-answer, and feeding back the target answer to the target work order. According to the response mode disclosed by the invention, the complex question is divided into a plurality of sub-questions, so that the user request is understood, different service agents are called to respectively answer, the capability of processing a plurality of different types is achieved, a good response effect is achieved for different types of sub-questions, and the accuracy and accuracy of a target answer can be effectively improved.

Description

Technical Field

[0001] This application for invention relates to the field of artificial intelligence, and particularly to a user service response method, apparatus, device and medium. Background Art

[0002] In the products or services provided by enterprises, customer service is a crucial link. Traditional customer service mainly relies on manual processing, which has problems such as long response time, high cost, and prone to human errors. With the development of artificial intelligence technology, some artificial intelligence-based customer service systems have gradually been widely applied. These systems can use intelligent robots to perform simple dialogue responses through preset rules and templates. Although the response efficiency of the system is effectively improved, this existing technology often can only handle specific types of problems, is difficult to understand complex user requests and give accurate answers, and is prone to giving relatively inaccurate or even completely wrong response solutions. Summary of the Invention

[0003] This application for invention provides a user service response method, apparatus, device and medium to solve the technical problem of how to improve the accuracy of answers.

[0004] To solve the above technical problem, this application for invention provides a user service response method, including:

[0005] Obtain the user request corresponding to the target work order, and extract the keywords of the user request;

[0006] Identify the user request according to the keywords. When it is identified that the user request contains multiple sub-questions, respectively match corresponding service agents for each sub-question;

[0007] Input each sub-question into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-question, generate a target answer corresponding to the user request according to each sub-answer, and feedback the target answer to the target work order.

[0008] As a preferred solution, the identifying the user request according to the keywords includes:

[0009] Obtain a plurality of preset word sets, and respectively match the word sets corresponding to each keyword according to the mapping relationship between the keywords and the word sets;

[0010] Identify the sub-questions included in the user request according to the keywords and the matched word sets.

[0011] As a preferred solution, the inputting each sub-question into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-question includes:

[0012] Input each sub - problem into the corresponding and matched service agent respectively, so that each service agent calls at least one reply tool respectively to obtain sub - answers corresponding to each sub - problem; the types of the reply tools include a noun knowledge base, a semantic knowledge base, a knowledge graph, and external tools.

[0013] As a preferred solution, the semantic knowledge base is pre - configured to store the association between sub - problems and semantic information.

[0014] Each of the service agents calls at least one reply tool respectively to obtain sub - answers corresponding to each sub - problem, including:

[0015] The service agent retrieves the semantic knowledge base to obtain the semantic information corresponding to the sub - problem.

[0016] Obtain the user's historical service requests, and obtain the answer correction information corresponding to the sub - problem according to the historical service requests and the knowledge graph.

[0017] Retrieve the noun knowledge base according to the specific keywords of the sub - problem to obtain noun information.

[0018] Call the external tool according to the sub - problem to obtain external information.

[0019] Integrate the semantic information, answer correction information, noun information, and external information to obtain the sub - answer corresponding to the sub - problem.

[0020] As a preferred solution, the service agent integrates the semantic information, answer correction information, noun information, and external information to obtain the sub - answer corresponding to the sub - problem, including:

[0021] The service agent calculates the correlation degrees of the semantic information, answer correction information, noun information, and external information with the sub - problem through a reflection algorithm.

[0022] Eliminate the information with a correlation degree lower than a preset correlation degree threshold as noise to obtain the sub - answer corresponding to the sub - problem.

[0023] As a preferred solution, the service agent calls the external tool according to the sub - problem to obtain external information, including:

[0024] When the sub - problem is a sub - problem in the email field, the service agent queries the email server to obtain sending and receiving log, domain name system information, and blacklist information.

[0025] As a preferred solution, extracting the keywords of the user request includes:

[0026] Extract multiple characters of the user request.

[0027] Perform statistical analysis on the multiple characters, identify high-frequency words, and thus determine the target extraction length of the characters;

[0028] Obtain the keywords of the user request according to the high-frequency words and the target extraction length.

[0029] Correspondingly, the present invention application also provides a user service response device, including an extraction module, a matching module, and a feedback module; wherein,

[0030] The extraction module is used to obtain the user request corresponding to the target work order and extract the keywords of the user request;

[0031] The matching module is used to identify the user request according to the keywords. When it is identified that the user request contains multiple sub-questions, corresponding service agents are respectively matched for each sub-question;

[0032] The feedback module is used to input each sub-question into the corresponding matched service agent respectively, obtain sub-answers corresponding to each sub-question, generate a target answer corresponding to the user request according to each sub-answer, and feedback the target answer to the target work order.

[0033] As a preferred solution, the matching module identifies the user request according to the keywords, including:

[0034] The matching module obtains a plurality of preset word sets, and respectively matches the word sets corresponding to each keyword according to the mapping relationship between the keyword and the word set;

[0035] Identify the sub-questions included in the user request according to the keyword and the matched word set.

[0036] As a preferred solution, the feedback module inputs each sub-question into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-question, including:

[0037] The feedback module inputs each sub-question into the corresponding matched service agent respectively, so that each service agent respectively calls at least one reply tool to obtain sub-answers corresponding to each sub-question; the types of the reply tools include a noun knowledge base, a semantic knowledge base, a knowledge graph, and external tools.

[0038] As a preferred solution, the semantic knowledge base is pre-configured to store the association between sub-questions and semantic information;

[0039] Each service agent respectively calls at least one reply tool to obtain sub-answers corresponding to each sub-question, including:

[0040] The service agent retrieves the semantic knowledge base to obtain semantic information corresponding to the sub-question;

[0041] Obtain the user's historical service requests, and obtain answer correction information corresponding to the sub-question according to the historical service requests and the knowledge graph;

[0042] Retrieve the noun knowledge base according to the specific keywords of the sub-question to obtain noun information;

[0043] Call the external tool according to the sub-question to obtain external information;

[0044] Integrate the semantic information, answer correction information, noun information, and external information to obtain a sub-answer corresponding to the sub-question.

[0045] As a preferred solution, the service agent integrates the semantic information, answer correction information, noun information, and external information to obtain a sub-answer corresponding to the sub-question, including:

[0046] The service agent calculates the correlation between the semantic information, answer correction information, noun information, and external information and the sub-question through a reflection algorithm;

[0047] Eliminate the information with a correlation lower than the preset correlation threshold as noise to obtain a sub-answer corresponding to the sub-question.

[0048] As a preferred solution, the service agent calls the external tool according to the sub-question to obtain external information, including:

[0049] When the sub-question is a sub-question in the email field, the service agent queries the email server to obtain send and receive log, domain name system information, and blacklist information.

[0050] As a preferred solution, the extraction module extracts the keywords of the user request, including:

[0051] The extraction module extracts multiple characters of the user request;

[0052] Perform statistical analysis on the multiple characters to identify high-frequency words, thereby determining the target extraction length of the characters;

[0053] Obtain the keywords of the user request according to the high-frequency words and the target extraction length.

[0054] Correspondingly, the present invention application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the user service response method described above is implemented.

[0055] Correspondingly, the present invention application also provides a computer-readable storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the user service response method described above.

[0056] Compared with the prior art, the present invention application has the following beneficial effects:

[0057] The present invention application provides a user service response method, device, equipment and medium. The user service response method includes: obtaining a user request corresponding to a target work order, and extracting keywords of the user request; identifying the user request according to the keywords. When it is identified that the user request includes multiple sub-questions, respectively matching corresponding service agents for each sub-question; inputting each sub-question into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-question, generating a target answer corresponding to the user request according to each sub-answer, and feeding back the target answer to the target work order. By extracting the keywords of the user request, identifying the user request according to the keywords, identifying the complex problem of the user request as multiple sub-questions, and respectively matching service agents for each sub-question to obtain sub-answers of each sub-question, so as to obtain a target answer corresponding to the user request and feed it back to the target work order. The response method of the present invention application divides the complex problem into multiple sub-questions, thereby understanding the user request and calling different service agents to answer respectively. It has the ability to process various different types, and has a good response effect on different types of sub-questions, which can effectively improve the accuracy and precision of the target answer. Description of the Drawings

[0058] Figure 1 : It is a schematic flowchart of an embodiment of the user service response method provided by the present invention application.

[0059] Figure 2 : It is a schematic flowchart of an embodiment of the user service response device provided by the present invention application. Detailed Embodiments

[0060] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.

[0061] Embodiment 1

[0062] Please refer to Figure 1 ,Figure 1 A user service response method provided for this invention application, including steps S101 to S103; among them, each step is described in detail as follows:

[0063] Step S101, obtain the user request corresponding to the target work order, and extract the keywords of the user request.

[0064] In this embodiment, the user service response method can be applied to a customer service system, and the customer service system can be set in a computer device, and the computer device includes but is not limited to devices such as smart phones, laptop computers, tablet computers, desktop computers, physical servers, and cloud servers.

[0065] In this step, user requests can be managed through work orders (work documents). A work order is defined as a work plan composed of one or more tasks, and can be used as the basis for the superior department to issue tasks and the subordinate department to receive tasks. A work order can be independent, or it can be a part of a large project. At the same time, sub-work orders can also be defined for work orders.

[0066] This embodiment can obtain the above-mentioned target work order by reading multiple channels such as emails, Web forms, instant messaging tools, and / or application programming (Application Programming Interface, abbreviated as API) interfaces.

[0067] This embodiment obtains the user request corresponding to the target work order and extracts the keywords of the user request. The user request of this application contains the "question" that the user needs the customer service system to answer. This question can be described in declarative sentences or interrogative sentences, and this question can be described in forms such as Chinese, English, or even code.

[0068] For example, the user's question is "My XXX module has a fault. Please analyze the cause of the fault." At this time, the customer service system can answer this question (or rather, for this user request). This step sorts the target work order to ensure that the corresponding question can match the correct service agent in the subsequent steps.

[0069] In a preferred implementation manner, the extracting the keywords of the user request includes: extracting multiple characters of the user request; performing statistical analysis on the multiple characters to identify high-frequency words, so as to determine the target extraction length of the characters; and obtaining the keywords of the user request according to the high-frequency words and the target extraction length.

[0070] In this embodiment, n characters can be extracted from historical work orders, such as 50, 100, or 200 characters, for data statistics, to determine which character segments can be used for classification, identify high-frequency words or topics, so as to judge which information segments are valuable for sorting, to determine the optimal character length (target extraction length), and thus obtain the keywords of the user request. In addition to improving accuracy, this embodiment can also effectively save tokens (computing power costs).

[0071] Step S102, identify the user request according to the keyword. When it is identified that the user request contains multiple sub-questions, match corresponding service agents for each sub-question respectively.

[0072] In this step, the identifying the user request according to the keyword includes: obtaining a plurality of preset word sets, and respectively matching the word sets corresponding to each keyword according to the mapping relationship between the keyword and the word sets; identifying the sub-questions included in the user request according to the keyword and the matched word sets.

[0073] In this embodiment, each word set can respectively correspond to a service agent. For example, for service agents A, B, and C, they can respectively correspond to different word sets, and the mapping relationship between the service agent and the word set is pre-constructed. In this step, by matching the keyword with the word set, the specific sub-questions are further identified.

[0074] For example, for service agents A, B, and C, the corresponding word sets can be respectively:

[0075] A: The word set includes bounce, unable to send, unable to send email, abnormal email sending;

[0076] B: The word set includes unable to log in, login failed;

[0077] C: The word set includes spam, phishing email, fraud email, etc.

[0078] That is, service agent A is used to answer sub-questions related to bounce, unable to send, etc., service agent B is used to answer sub-questions related to login anomalies, and service agent C is used to answer sub-questions such as spam.

[0079] Further, the user requests in this embodiment can be distinguished by "simple questions" and "complex questions". For example, specifically, it can be judged according to dimensions such as work order keywords, classification, and complexity score through a rule engine algorithm. If the keywords include "password", "reset", and "login", all belonging to the same word set, then the user request is identified as a "simple question" at this time. The "simple question" can directly call the semantic knowledge base for answering.

[0080] When the user request contains keywords of multiple word sets, it is recognized as a "complex problem" at this time. Subsequently, corresponding agents need to be matched for each of the multiple sub-problems it contains.

[0081] Exemplarily, when the user request is "XXX function cannot be used normally", it can be divided or recognized into multiple sub-problems at this time, such as sub-problems like "consultation on this normal function", "whether there have been requests for configuration modification in history", and "what is the disposal method to solve this function", etc. At this time, according to the keywords among them, the corresponding word sets and service agents can be determined.

[0082] Step S103, input each sub-problem into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-problem, generate a target answer corresponding to the user request according to each of the sub-answers, and feedback the target answer to the target work order.

[0083] In this step, the inputting each sub-problem into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-problem includes: inputting each sub-problem into the corresponding matched service agent respectively, so that each service agent calls at least one reply tool respectively to obtain sub-answers corresponding to each sub-problem; the types of the reply tools include but are not limited to noun knowledge base, semantic knowledge base, knowledge graph, and external tools.

[0084] For a certain sub-problem, the reply tools that the service agent can call can be any one or two or three or four of the noun knowledge base, semantic knowledge base, knowledge graph, and external tools. Integrate the information respectively feedback by the reply tools to obtain its corresponding sub-answer, so as to ensure the accuracy of the obtained sub-answer.

[0085] For the application scenario where all four reply tools are called, in a preferred embodiment, the inputting each sub-problem into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-problem includes: the service agent retrieves the semantic knowledge base to obtain semantic information corresponding to the sub-problem; obtains the historical service requests of the user, and obtains answer correction information corresponding to the sub-problem according to the historical service requests and the knowledge graph; retrieves the noun knowledge base according to the specific keywords of the sub-problem to obtain noun information; calls the external tool according to the sub-problem to obtain external information; integrates the semantic information, answer correction information, noun information, and external information to obtain the sub-answer corresponding to the sub-problem.

[0086] In this embodiment, the semantic knowledge base can be pre-configured to store the association between sub-problems and semantic information. Query the semantic knowledge base for the sub-problem to find the corresponding semantic information.

[0087] The knowledge graph mainly associates sub-questions and the logical relationships before and after the functions involved. For example, the sub-question may be related to a certain function configuration or value-added function in historical data. Therefore, by querying the knowledge graph, it is possible to check whether there is relevant content needed in the historical data. For example, if the sub-question is "My XXX function cannot be used", and it is learned from querying the historical data that the user has configured "I do not want to use the XXX function" for this function, then the knowledge graph can be updated to obtain the above answer correction information "The configuration was modified on xx / xx".

[0088] For example, for the above sub-question "Consultation on this normal function", the semantic knowledge base can be called; for "Whether there has been a request for configuration modification in history", the knowledge graph can be called; and for "What is the handling method to solve this function", the semantic knowledge base can be called.

[0089] The noun knowledge base described in this embodiment is mainly a professional noun knowledge base or a proper noun knowledge base, which is used to analyze nouns or names in some specific fields to achieve differentiation between different fields.

[0090] For example, for the keyword "car", the meaning of "car" in the fields of "means of transportation" and "toy" can be distinguished by querying the noun knowledge base. Or, in the email field, the web version is called webmail and the mobile version is called hxphone. At this time, the retrieval knowledge can be added through the noun knowledge base: hxphone is a mobile application; webmail is an interface application to obtain noun information.

[0091] For obtaining sub-answers in some fields, external tools can be called for querying. For example, when the sub-question is a sub-question in the email field, the service agent can query the email server to obtain send and receive logs, Domain Name System (DNS) information, and blacklist information (such as the Realtime Blackhole List, abbreviated as RBL, which is an anti-spam technology. Through DNS query, RBL can check whether an IP address is listed in the blacklist to determine whether the IP address has sent spam. If the IP address is in the RBL list, the email server will reject emails from that IP address to prevent the spread of spam), and determine the send and receive logs, DNS information, and blacklist information as the external information.

[0092] After obtaining at least two of the above semantic information, answer correction information, noun information, and external information, the relevance between the semantic information, answer correction information, noun information, and external information and the sub-question can be calculated respectively through a reflection algorithm; the information with a relevance lower than the preset relevance threshold is removed as noise, and the information with a relevance higher than or equal to the above relevance threshold is retained, and the sub-answer corresponding to the sub-question is obtained by integration.

[0093] Then, the sub-answers are integrated and analyzed to obtain the target answer corresponding to the user request, and the target answer is fed back to the target work order. And, before feeding the target answer back to the target work order, the target answer can be confirmed by means of manual quality inspection, the target answers that do not meet the requirements are marked, and they are used as corpus for the tuning of the service agent. After the performance index of the service agent, such as its accuracy rate, reaches a certain level, it can be separated from this manual processing to achieve fully automatic user service response.

[0094] Correspondingly, please refer to Figure 2 , the present invention application also provides a user service response device 200, including an extraction module 201, a matching module 202, and a feedback module 203; wherein,

[0095] The extraction module 201 is used to obtain the user request corresponding to the target work order and extract the keywords of the user request.

[0096] The matching module 202 is used to identify the user request according to the keywords. When it is identified that the user request contains multiple sub-questions, corresponding service agents are respectively matched for each sub-question.

[0097] The feedback module 203 is used to input each sub-question into the corresponding matched service agent respectively, obtain the sub-answer corresponding to each sub-question, generate the target answer corresponding to the user request according to each sub-answer, and feed the target answer back to the target work order.

[0098] As a preferred solution, the matching module 202 identifies the user request according to the keywords, including:

[0099] The matching module 202 obtains a plurality of preset word sets, and respectively matches the word sets corresponding to each keyword according to the mapping relationship between the keyword and the word set;

[0100] According to the keyword and the matched word set, the sub-questions included in the user request are identified.

[0101] As a preferred solution, the feedback module 203 inputs each sub-question into the corresponding matched service agent respectively, and obtains the sub-answer corresponding to each sub-question, including:

[0102] The feedback module 203 inputs each sub-question into the correspondingly matched service agent, so that each service agent calls at least one reply tool respectively to obtain sub-answers corresponding to each sub-question; the types of the reply tools include a noun knowledge base, a semantic knowledge base, a knowledge graph, and external tools.

[0103] As a preferred solution, the semantic knowledge base is pre-configured to store the association between sub-questions and semantic information;

[0104] Each service agent calls at least one reply tool respectively to obtain sub-answers corresponding to each sub-question, including:

[0105] The service agent retrieves the semantic knowledge base to obtain the semantic information corresponding to the sub-question;

[0106] Obtain the user's historical service requests, and obtain the answer correction information corresponding to the sub-question according to the historical service requests and the knowledge graph;

[0107] Retrieve the noun knowledge base according to the specific keywords of the sub-question to obtain noun information;

[0108] Call the external tool according to the sub-question to obtain external information;

[0109] Integrate the semantic information, answer correction information, noun information, and external information to obtain the sub-answer corresponding to the sub-question.

[0110] As a preferred solution, the service agent integrates the semantic information, answer correction information, noun information, and external information to obtain the sub-answer corresponding to the sub-question, including:

[0111] The service agent calculates the correlation degree between the semantic information, answer correction information, noun information, and external information and the sub-question through a reflection algorithm;

[0112] Eliminate the information with a correlation degree lower than the preset correlation degree threshold as noise to obtain the sub-answer corresponding to the sub-question.

[0113] As a preferred solution, the service agent calls the external tool to obtain external information according to the sub-question, including:

[0114] When the sub-question is a sub-question in the email field, the service agent queries the email server to obtain the sending and receiving log, domain name system information, and blacklist information.

[0115] As a preferred solution, the extraction module 201 extracts the keywords of the user request, including:

[0116] The extraction module 201 extracts a plurality of characters of the user request;

[0117] Perform statistical analysis on the plurality of characters to identify high-frequency words, thereby determining the target extraction length of the characters;

[0118] Obtain the keyword of the user request according to the high-frequency word and the target extraction length.

[0119] Correspondingly, the present invention application also provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the user service response method described above is implemented.

[0120] The so-called processor may be a central processing unit (CPU), or may also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or the processor may also be any conventional processor, etc. The processor is the control center of the terminal, and connects various parts of the entire terminal through various interfaces and lines.

[0121] The memory can be used to store the computer program. By running or executing the computer program stored in the memory, and calling the data stored in the memory, various functions of the terminal are realized. The memory may mainly include a program storage area and a data storage area. Among them, the program storage area can store an operating system, application programs required for at least one function (such as a sound playback function, an image playback function, etc.), etc.; the data storage area can store data created according to the use of the mobile phone (such as audio data, phone book, etc.), etc. In addition, the memory may include high-speed random access memory, and may also include non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one magnetic disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0122] Correspondingly, the present invention application also provides a computer-readable storage medium, which includes a stored computer program. When the computer program runs, it controls the device where the computer-readable storage medium is located to execute the user service response method described above.

[0123] Among them, if the modules integrated in the user service response device / terminal are implemented in the form of software function units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on such an understanding, to implement all or part of the processes in the above-mentioned embodiment methods of the present invention, it can also be completed by instructing relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate forms, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0124] Compared with the prior art, the present invention application has the following beneficial effects:

[0125] The present invention application provides a user service response method, device, equipment and medium. The user service response method includes: obtaining a user request corresponding to a target work order, and extracting keywords of the user request; identifying the user request according to the keywords. When it is identified that the user request includes multiple sub-questions, respectively matching corresponding service agents for each sub-question; inputting each sub-question into the corresponding matched service agent respectively to obtain sub-answers corresponding to each sub-question, generating a target answer corresponding to the user request according to each sub-answer, and feeding back the target answer to the target work order. The present invention application extracts the keywords of the user request, identifies the user request according to the keywords, identifies the complex problem of the user request as multiple sub-questions, and respectively matches service agents for each sub-question to obtain sub-answers for each sub-question, so as to obtain a target answer corresponding to the user request and feed it back to the target work order. The response method of the present invention application divides the complex problem into multiple sub-questions, thereby understanding the user request and calling different service agents to answer respectively. It has the ability to handle various different types, and has a good response effect on different types of sub-questions, which can effectively improve the accuracy and precision of the target answer.

[0126] The specific embodiments described above further elaborate on the objectives, technical solutions, and beneficial effects of the present invention. It should be understood that the above are only specific embodiments of the present invention and are not used to limit the protection scope of the present invention. In particular, for those skilled in the art, any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A user service response method, characterized in that: include: Obtain the user request corresponding to the target work order, and extract keywords of the user request; Identify the user request according to the keyword, and when it is identified that the user request contains multiple sub-questions, match a corresponding service agent to each sub-question; Each sub-question is input into the corresponding matched service agent to obtain a sub-answer corresponding to each sub-question, a target answer corresponding to the user request is generated according to each sub-answer, and the target answer is fed back to the target work order.

2. A user service response method as claimed in claim 1, characterized in that: The identifying the user request according to the keyword includes: Acquire a plurality of preset word sets, and match the word sets corresponding to the keywords respectively according to the mapping relationship between the keywords and the word sets; According to the keywords and the matched word set, the sub-questions included in the user request are identified.

3. A user service response method as claimed in claim 1, characterized in that: The step of inputting each sub-question into the corresponding matched service agent to obtain a sub-answer corresponding to each sub-question includes: Each sub-question is input into the corresponding matched service agent, so that each service agent calls at least one reply tool to obtain a sub-answer corresponding to each sub-question; the types of reply tools include noun knowledge base, semantic knowledge base, knowledge graph and external tools.

4. A user service response method as claimed in claim 3, characterized in that: Each of the service agents respectively calls at least one answering tool to obtain a sub-answer corresponding to each sub-question, including: The service agent searches the semantic knowledge base to obtain semantic information corresponding to the sub-question; Obtain the user's historical service requests, and obtain answer correction information corresponding to the sub-question based on the historical service requests and the knowledge graph; Searching the noun knowledge base according to the specific keywords of the sub-question to obtain noun information; Calling the external tool to obtain external information according to the sub-problem; The semantic information, answer correction information, noun information and external information are integrated to obtain sub-answers corresponding to the sub-questions.

5. A user service response method as claimed in claim 3, characterized in that: The service agent integrates the semantic information, answer correction information, noun information and external information to obtain a sub-answer corresponding to the sub-question, including: The service agent calculates the relevance of the semantic information, answer correction information, noun information and external information to the sub-question through a reflection algorithm; The information with a correlation lower than the preset correlation threshold is removed as noise to obtain the sub-answer corresponding to the sub-question.

6. A user service response method as claimed in claim 3, characterized in that: The service agent calls the external tool to obtain external information according to the sub-problem, including: When the sub-problem is a mail domain sub-problem, the service agent queries the mail server, obtains the mail sending and receiving logs, domain name system information and blacklist information, and determines the mail sending and receiving logs, domain name system information and blacklist information as the external information.

7. A user service response method as claimed in claim 1, characterized in that: The extracting the keyword requested by the user includes: Extracting a plurality of characters requested by the user; Performing statistical analysis on the plurality of characters to identify high-frequency words, thereby determining a target extraction length of the characters; The keywords requested by the user are obtained according to the high-frequency words and the target extraction length.

8. A user service answering device, characterized in that: It includes extraction module, matching module and feedback module; among them, The extraction module is used to obtain the user request corresponding to the target work order and extract the keywords of the user request; The matching module is used to identify the user request according to the keyword, and when it is identified that the user request contains multiple sub-questions, match the corresponding service agent for each sub-question; The feedback module is used to input each sub-question into the corresponding matched service agent, obtain a sub-answer corresponding to each sub-question, generate a target answer corresponding to the user request based on each sub-answer, and feed back the target answer to the target work order.

9. A terminal device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the user service response method as claimed in any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the user service response method according to any one of claims 1 to 7.

Images