An AI-based intelligent customer service telephone repair method and system

Abstract

The present invention provides an AI-based intelligent customer service telephone repair reporting method and system, including: establishing a maintenance content knowledge base, establishing a first knowledge graph, the first knowledge graph including entity information of multiple maintenance subjects; establishing a second knowledge graph according to the maintenance content knowledge base, wherein the second knowledge graph includes entity information of fault types corresponding to the maintenance subjects; establishing a first key-value pair list of corresponding maintenance subject fault type guidance voice data, and searching for guidance voice data according to the first key-value pair list; obtaining user voice data to search the first knowledge graph, searching for corresponding fault type entity information from the corresponding second knowledge graph according to the first knowledge graph entity information and user voice keywords, obtaining corresponding guidance voice data to establish a maintenance guidance dialogue with the user; judging whether the currently searched fault type is correct according to the user voice data in the maintenance guidance dialogue, so as to determine the fault type or initiate the guidance dialogue again.

Description

Technical Field

[0001] The present invention relates to the field of intelligent repair reporting technology, and in particular to an AI-based intelligent customer service telephone repair reporting method and system. Background Art

[0002] At present, the existing intelligent repair technology usually involves staff receiving repair calls from customers, analyzing the content of the repair calls, determining the repair parts or systems that the customer may need, and further preparing a repair order and uploading it to the system for automatic distribution. However, the above repair technology solution requires that the repair service staff who answer the phone have strong professional capabilities and need to quickly understand the type of equipment failure described by the customer's repair call, as well as the parts or systems that need to be repaired, etc. Otherwise, the repair order produced is prone to the problem of mismatch between the repair content and the actual equipment failure. In addition, since the customer is not a professional, it is easy for the fault description to be inaccurate in the repair call. Therefore, the existing repair solution will lead to more professional and description barriers between the repair service staff and the customer's repair needs, thereby affecting the progress of equipment repair. Summary of the invention

[0003] One of the inventive purposes of the present invention is to provide an AI-based intelligent customer service telephone repair reporting method and system, wherein the method and system pre-construct a repair service knowledge structure of a dual knowledge graph, wherein the knowledge structure of the first knowledge graph is used to establish and store different types of repair reporting subjects and establish similarity relationships between different repair reporting subjects, and the knowledge structure of the second knowledge graph is used to establish and store the fault types of the repair reporting subjects. The first knowledge graph and the second knowledge graph can be used to relatively determine all complete repair reporting contents for subsequent determination of fault types.

[0004] Another inventive object of the present invention is to provide an AI-based intelligent customer service telephone repair method and system, wherein the method and system establish a repair dialogue guidance strategy based on relevant keywords recognized by the user's voice information, wherein the dialogue guidance strategy will be implemented based on the first knowledge graph and the second knowledge graph knowledge structure, and a fault description voice list of each fault type entity information is established based on the fault type based on the second knowledge graph, the most similar fault type is identified based on the voice information and the voice information in the corresponding fault voice list is fed back, and the fault type is further determined or the fault type is re-searched based on the user feedback information.

[0005] Another inventive object of the present invention is to provide an AI-based intelligent customer service telephone repair method and system, wherein the method and system perform dialogue guidance operations based on user feedback information. When the user's description of the fault type in the feedback is not the actual fault type that occurred, the closest fault type is found based on the entity similarity relationship of the second knowledge graph, and the feedback voice information is regenerated to the user, and then it is determined again based on the user's dialogue feedback whether it is the fault type that needs to be resolved, thereby achieving guided dialogue and quickly querying the target fault type.

[0006] Another inventive object of the present invention is to provide an AI-based intelligent customer service telephone repair reporting method and system, wherein the method and system establish a maintenance instruction key-value pair list for each fault type, obtain the target fault type through the above-mentioned guided dialogue, and search the corresponding maintenance instruction key-value pair list according to the target fault type to find the maintenance instruction information corresponding to the target fault type, and send the maintenance instruction information to the corresponding maintenance personnel to realize automated maintenance services without manually preparing related maintenance forms, thereby reducing the waste of human resources.

[0007] In order to achieve at least one of the above-mentioned invention purposes, the present invention further provides an AI-based intelligent customer service telephone repair method, the method comprising:

[0008] Establishing a maintenance content knowledge base, and establishing a first knowledge graph based on the maintenance content knowledge base, wherein the first knowledge graph includes entity information of multiple maintenance subjects;

[0009] Establishing a second knowledge graph according to the maintenance content knowledge base, wherein the second knowledge graph includes fault type entity information corresponding to the maintenance subject;

[0010] Establish a first key-value pair list of guidance voice data corresponding to the fault type of the maintenance subject, and search for the guidance voice data according to the first key-value pair list;

[0011] Obtain user voice data to search for entity information of the first knowledge graph, further search for corresponding fault type entity information from the corresponding second knowledge graph based on the entity information of the first knowledge graph and user voice keywords, and obtain corresponding guidance voice data to establish a guidance dialogue with the user for maintenance;

[0012] It is determined whether the fault type currently being searched is correct based on the user voice data in the guiding dialogue, and is used to determine the fault type or initiate the guiding dialogue again.

[0013] According to one of the preferred embodiments of the present invention, the method includes: using a speech recognition model to identify keywords in the user's repair reporting voice data, searching the first knowledge graph based on the identified voice data keywords, searching for a first target maintenance subject from the entity information of the first knowledge graph, and searching for a first fault type of the first target maintenance subject in the entity information of the second knowledge graph based on the first target maintenance subject, obtaining the corresponding first guidance voice data stored in the first key-value pair list based on the first fault type found, and returning the first guidance voice data to establish a guidance dialogue with the user.

[0014] According to another preferred embodiment of the present invention, after the first guidance voice data is returned and a guidance dialogue is established, the user voice data in the guidance dialogue is obtained, and the evaluation keywords of the user in the guidance dialogue are identified, and it is determined whether the first fault type currently being searched is a real fault type based on the evaluation keywords of the user in the guidance dialogue.

[0015] According to another preferred embodiment of the present invention, the method for constructing the guidance voice data includes: obtaining status information of the corresponding equipment fault type based on historical maintenance data and maintenance instruction manuals of different equipment, constructing standard question-based guidance voice data based on the status information of the corresponding equipment fault type, and using the question-based guidance voice data with the corresponding fault type entity as the key name and its own guidance voice data as the key value to establish the first key-value pair list.

[0016] According to another preferred embodiment of the present invention, the method for processing the guided dialogue includes: using voice recognition technology to identify evaluation keywords in the guided dialogue, and determining whether the evaluation keywords are negative evaluation keywords, and if they are negative evaluation keywords, re-searching the second knowledge graph for the second fault type entity information that is most similar to the previous first fault type, and finding the second guidance voice data corresponding to the second fault type according to the first key-value pair list, and further returning the second guidance voice data to reconstruct the guided dialogue with the user.

[0017] According to another preferred embodiment of the present invention, a threshold value for the number of guided dialogue construction times is set, and if the number of guided dialogues currently reconstructed based on the user's negative evaluation keywords in the guided dialogue is greater than or equal to the threshold value for the number of guided dialogue construction times, then the first knowledge graph is returned to obtain the entity information of the closest new second maintenance subject in the first knowledge graph according to the cosine similarity calculation method of the entity information, and the third fault type existing in the second knowledge graph is searched based on the second maintenance subject and the user's voice keywords, and the guidance voice data corresponding to the third fault type is searched based on the first key-value pair list, and the guidance voice data corresponding to the third fault type is returned, and the guidance dialogue is reconstructed with the user.

[0018] According to another preferred embodiment of the present invention, the method for processing the guided dialogue includes: when negative evaluation keywords in the guided dialogue are identified by voice recognition technology, and new keywords in the guided dialogue are identified and compared with maintenance subject keywords in the entity information of the first knowledge graph at the same time, if there are new maintenance subject keywords in the guided dialogue and there are negative evaluation keywords, then the fault type corresponding to the second knowledge graph is searched in the new maintenance subject in the first knowledge graph, and the corresponding guidance voice data is searched according to the first key-value pair list to return and reconstruct the guided dialogue.

[0019] According to another preferred embodiment of the present invention, the method includes: constructing a second key-value pair list, wherein the second key-value pair list uses the fault type entity information of the second knowledge graph as the key name, and the corresponding maintenance strategy as the key value, and when there are positive evaluation keywords in the user voice data in the guided dialogue, the second key-value pair list is queried according to the fault type entity information of the human knowledge graph currently constructing the guided dialogue to obtain the corresponding maintenance strategy, and the maintenance strategy is sent to the target maintenance personnel to perform the maintenance operation.

[0020] In order to achieve at least one of the above-mentioned invention purposes, the present invention further provides an AI-based intelligent customer service telephone repair reporting system, which executes the above-mentioned AI-based intelligent customer service telephone repair reporting method.

[0021] The present invention further provides a computer-readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the above-mentioned AI-based intelligent customer service telephone repair reporting method. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 Shown is a flow chart of an AI-based intelligent customer service telephone repair reporting method of the present invention.

[0023] Figure 2 Shown is a schematic diagram of the structure of the first knowledge graph in the present invention.

[0024] Figure 3 Shown is a schematic diagram of the structure of the second knowledge graph in the present invention.

[0025] Figure 4 Shown is a schematic diagram of the first key-value pair list in the present invention.

[0026] Figure 5 Shown is a schematic diagram of the second key-value pair list in the present invention. DETAILED DESCRIPTION

[0027] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0028] It is to be understood that the term "one" should be understood as "at least one" or "one or more", that is, in one embodiment, the number of an element may be one, while in another embodiment, the number of the element may be multiple, and the term "one" should not be understood as a limitation on the quantity.

[0029] Please combine Figure 1-Figure 5 , the present invention discloses an AI-based intelligent customer service telephone repair method and system, wherein the method mainly includes the following steps: first, it is necessary to establish a maintenance knowledge base for different devices, wherein the maintenance knowledge base can be established based on, but not limited to, the historical maintenance data of the equipment, professional maintenance instruction manuals, and the compilation of professional maintenance personnel, etc. The content in the maintenance knowledge base includes, but is not limited to: equipment name, part name, structure name, system name, fault type name, state information presented by the corresponding fault type, and the equipment and maintenance method that need to be repaired. The establishment of the above maintenance knowledge base is used to construct a knowledge graph. Since different equipment corresponds to different parts and fault types, the present invention sets a knowledge structure of a dual knowledge graph, which can perform knowledge query at the maintenance subject level and the fault type level respectively. The present invention also uses existing speech recognition technology to identify the voice data keywords of the user in the telephone repair, and converts the voice data keywords into text keywords to query the knowledge graph, find the corresponding entity information, and judge the equipment fault type mentioned by the current user in the telephone repair process according to the found entity information. The present invention establishes guidance voice data based on fault type confirmation, further identifies the user's voice data and finds the corresponding fault type, obtains the corresponding guidance voice data and returns it to the user to build a guidance dialogue, and judges whether the current user's telephone repair equipment and fault type are accurate based on the guidance dialogue. If inaccurate, find an approximate fault type until the currently found fault type is the fault type confirmed by the user. The above-mentioned knowledge graph and guidance dialogue construction method can effectively avoid the problem of unsatisfactory repair content and progress due to unclear semantic expression of the fault and the fact that the fault type requires professional judgment through guidance confirmation.

[0030] For more information, see Figure 2The present invention establishes a first knowledge graph of the maintenance subject based on the maintenance knowledge base constructed as above, wherein each entity naming method of the first knowledge graph is the name of the device or device component, for example, the entity information names constructed by the first knowledge graph include: air conditioner, remote control, air conditioner fan, air conditioner pipeline, heater, television, screen and other equipment or equipment component names, wherein each device and equipment component in the first knowledge graph has a certain similarity relationship during the construction process, for example, it can be calculated that there is a certain attribute relationship and semantic relationship between the air conditioner and the air conditioner pipeline, so for the triples (entity, attribute, relationship) of the first knowledge graph of the air conditioner, there is a certain similarity relationship in the corresponding first knowledge graph, in the present invention, the (entity, attribute, relationship) of each knowledge graph node can be quantified into a feature vector, and the triple feature vectors of different nodes can be used to calculate the similarity of different knowledge nodes of the knowledge graph including but not limited to the cosine similarity calculation method. Since the similarity calculation method of the knowledge graph is a prior art, the present invention does not improve how to calculate the similarity, so the present invention will not be described in detail. Please refer to Figure 2 , the node of the first knowledge graph is P n Different knowledge nodes have certain connection relationships, and the lengths of different connection lines can be expressed as similarity. The shorter the connection line, the higher the similarity.

[0031] After completing the construction of the above-mentioned first knowledge graph, a second knowledge graph of the fault type of each entity name in the above-mentioned first knowledge graph is further constructed, wherein the second knowledge graph described in the present invention is constructed according to the entity name of each first knowledge graph. For example, the air conditioner is used as an entity name in the first knowledge graph, and in the second knowledge graph, entity names of different fault types including but not limited to air conditioner power outage, refrigeration failure, no wind fault, etc. can be constructed, and knowledge node triplets of the second knowledge graph corresponding to the fault type are constructed according to the entity names, attributes and relationships of the different fault types. The present invention is based on the fact that the triple information of the second knowledge graph can be converted into a feature vector of the knowledge node of the corresponding fault type, and the similarity data of the knowledge nodes of different fault types can be calculated according to the above-mentioned method including but not limited to the cosine similarity calculation method. Please refer to Figure 3 , each knowledge node in the second knowledge graph is defined as V n For the convenience of description, each knowledge node V in the second knowledge graph n It also indicates the corresponding node entity information name.

[0032] After completing the construction of the above-mentioned first knowledge graph P and second knowledge graph V, a mapping relationship between the corresponding first knowledge graph entity information and the corresponding second knowledge graph is established. After obtaining the entity name of the corresponding maintenance subject of the first knowledge graph, the corresponding fault type of the second knowledge graph can be found according to the entity name of the first knowledge graph.

[0033] The present invention further constructs guidance voice data based on the entity information of the second knowledge graph. Since different fault types have different manifestations and states, the present invention designs guidance voice data of different fault types according to the existing corresponding equipment professional maintenance instructions or historical maintenance data, wherein the guidance voice data is question-based guidance voice data; for example, when the fault type is air-conditioning power failure, the following question-based guidance voice data can be designed according to the professional maintenance instructions or historical maintenance data: "Is there a circuit power failure in your air-conditioning?" For another example, when the fault type is a cooling fault, the designed guidance voice data can be: "Does your air-conditioning blow cold air after turning on the cooling mode?" etc. The above guidance voice will be designed according to the manifestation of the specific fault type, and the present invention will not elaborate on this. Furthermore, the present invention establishes a first key-value pair list K1 for each guiding voice data established above, wherein the first key-value pair list is established with the entity name corresponding to the second knowledge graph, for example, the entity name of the second knowledge graph is "refrigeration failure" and the "refrigeration failure" is used as the key name key; and the guiding voice data corresponding to the entity name "Does your air conditioner blow cold air after turning on the cooling mode?" is used as the key value vlue, and all the fault types and guiding voice data corresponding to the second knowledge graph are constructed into corresponding key-value pairs key-vlue to obtain the first key-value pair list. Please refer to Figure 3 The first key-value pair list is displayed, where each entity information V of each second knowledge graph n Each corresponds to the guiding voice data of a question.

[0034] The present invention further constructs a corresponding maintenance strategy based on the entity information of the second knowledge graph, and constructs a second key-value pair list K2 with the maintenance strategy and the entity information name of the second knowledge graph, wherein the entity information name of the second knowledge graph is used as the key name key, and the corresponding maintenance strategy is used as the key value vlue, and all the entity information names of the second knowledge graph and the corresponding maintenance strategy are formed into key-value pairs to form the second key-value pair list. Please refer to Figure 4 Each entity information V of the second knowledge graph displayed n Each corresponds to a maintenance policy.

[0035] After completing the above-mentioned first knowledge graph P, second knowledge graph V, first key-value pair list K1, and second key-value pair list K2, the present invention further performs the following specific customer service telephone repair method: first, the user's first repair voice data is obtained, and the existing voice recognition technology is used to identify the keywords in the customer's first voice repair data, and the keywords are compared with the entity information in the first knowledge graph P to obtain the maintenance subject information based on the first knowledge graph, and further according to the entity information name of the first knowledge graph P, the second knowledge graph corresponding to the maintenance subject is searched, and according to the keywords in the customer's first voice repair data, the entity information V in the second knowledge graph V is searched. n , used to determine the corresponding fault type data of the maintenance subject in the current customer's first voice repair data, and further according to the entity information V corresponding to the second knowledge graph n , search the corresponding guiding voice data question in the first key-value pair list K1 according to the corresponding entity letter n , the found guiding voice data question n , converted into speech through text-to-speech technology, the guiding speech data is automatically broadcast to the user to construct the first guiding dialogue. At this time, after receiving the first guiding dialogue, the user will judge whether the fault state described by the server in the current dialogue is consistent with the real state. If it is consistent, the evaluation keywords will be fed back. It is worth mentioning that the evaluation keywords described in the present invention include negative evaluation keywords and positive evaluation keywords, wherein the negative and positive defined in the present invention are positive or negative relative to the fault type, that is, the server judges correctly as a positive keyword, and the server judges incorrectly as a negative keyword. For example, in the above-mentioned air-conditioning power failure, for the guiding speech "Is there a circuit power failure in your air-conditioning?" If the user replies "yes", the current server judges correctly relative to the air-conditioning power failure, that is, "yes" is a positive keyword. For another example: the guiding speech data corresponding to the refrigeration failure is "Does your air-conditioning blow cold air after turning on the refrigeration mode?" When the user replies "yes", the current server judges incorrectly relative to the refrigeration failure, and the user replies "yes" as a negative keyword. It should be noted that the above negative keywords and positive keywords can be pre-stored in the form of data sets, and the corresponding keyword quantity and type are set according to language habits, which will not be described in detail in the present invention.

[0036] When the user voice data collected by the first guidance dialogue is a negative keyword, further based on the previous fault type judgment, the (entity, attribute, relationship) triple of the previous fault type knowledge graph node is obtained, and the (entity, attribute, relationship) triple of the corresponding fault type knowledge graph node is converted and encoded into a feature vector, and the second fault type knowledge graph node closest to the previous fault type knowledge graph node is calculated according to the above-mentioned cosine similarity calculation method, the entity information name of the second fault type knowledge graph node is extracted, and the second guidance voice data based on the second fault type is found according to the first key-value pair list, the second guidance voice data is returned, a second guidance dialogue with the user is established, and the negative keywords and positive keywords in the second guidance dialogue are judged to determine whether it is necessary to find the fault type again and establish a guidance dialogue.

[0037] When there are positive keywords in the voice conversation recognized by voice recognition technology, the system extracts the entity information name of the second knowledge graph node according to the positive keywords, and obtains the corresponding maintenance policy from the second key-value pair list K2 according to the entity information name of the second knowledge graph node. The corresponding maintenance policy is sent to the target maintenance personnel to perform maintenance operations.

[0038] It is worth mentioning that in the present invention, due to the problem of unclear language expression by the user himself, the maintenance subject information may not be recognized by voice recognition technology from the user's first voice repair data. Therefore, when the keywords recognized by the user's first voice repair data do not match the entity information of the first knowledge graph node, the system directly finds the maintenance subject with the most repair reports or the highest repair probability and the corresponding fault type through historical data statistics, and returns the guidance voice data to build a guidance dialogue. When the keywords expressed by the user in the subsequent guidance dialogue contain the maintenance subject keywords of the first knowledge graph and include the fault type keywords of the second knowledge graph, the guidance dialogue is directly constructed again according to the keywords recognized in the subsequent guidance dialogue to confirm the existing maintenance subject and fault type.

[0039] In another preferred embodiment of the present invention, in order to avoid the problem of the guided dialogue being entangled in local dialogue, the present invention sets a guided dialogue threshold for the guided dialogue constructed for each fault type of the maintenance subject. If the number of guided dialogues S currently reconstructed based on the user's negative evaluation keywords in the guided dialogue is greater than or equal to the threshold number of guided dialogue construction times S n, then return to the first knowledge graph, further obtain the entity information of the second maintenance subject in the first knowledge graph that is closest to the last maintenance subject according to the cosine similarity calculation method of the entity information, and search for the third fault type existing in the second knowledge graph according to the second maintenance subject entity information and the user's voice keywords, and again search for the guidance voice data corresponding to the third fault type according to the first key-value pair list, return the guidance voice data corresponding to the third fault type, and reconstruct the guidance dialogue with the user.

[0040] The embodiments disclosed in the present invention, the process described above with reference to the flowchart can be implemented as a computer software program. The embodiments disclosed in the present invention include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program contains a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from the network through the communication part, and / or installed from a removable medium. When the computer program is executed by the central processing unit (CPU), the above-mentioned functions defined in the method of the present application are executed. It should be noted that the above-mentioned computer-readable medium of the present application can be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium can be, for example, but not limited to, a system, device or device of an electrical, magnetic, optical, electromagnetic, infrared segment, or semiconductor, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to, an electrical connection with one or more wire segments, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present application, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In the present application, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, which carries a computer-readable program code. Such propagated data signals may take a variety of forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the above. A computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which may send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device. The program code embodied on the computer readable medium may be transmitted using any appropriate medium, including but not limited to wireless, electrical wire, optical cable, RF, etc., or any suitable combination of the foregoing.

[0041] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present invention. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0042] Those skilled in the art should understand that the embodiments of the present invention described above and shown in the accompanying drawings are only examples and do not limit the present invention. The objectives of the present invention have been fully and effectively achieved. The functional and structural principles of the present invention have been demonstrated and explained in the embodiments. Without departing from the principles, the implementation methods of the present invention may be deformed or modified in any way.

Claims

1. An AI-based intelligent customer service telephone repair method, characterized in that: The method comprises: Establishing a maintenance content knowledge base, and establishing a first knowledge graph based on the maintenance content knowledge base, wherein the first knowledge graph includes entity information of multiple maintenance subjects; Establishing a second knowledge graph according to the maintenance content knowledge base, wherein the second knowledge graph includes fault type entity information corresponding to the maintenance subject; Establish a first key-value pair list of guidance voice data corresponding to the fault type of the maintenance subject, and search for the guidance voice data according to the first key-value pair list; Obtain user voice data to search for entity information of the first knowledge graph, further search for corresponding fault type entity information from the corresponding second knowledge graph based on the entity information of the first knowledge graph and user voice keywords, and obtain corresponding guidance voice data to establish a guidance dialogue with the user for maintenance; Judging whether the fault type currently being searched is correct based on the user voice data in the guiding dialogue, so as to determine the fault type or initiate the guiding dialogue again; A threshold for the number of guided dialogue construction times is set. If the number of guided dialogues currently reconstructed based on the user's negative evaluation keywords in the guided dialogue is greater than or equal to the threshold for the number of guided dialogue construction times, return to the first knowledge graph and obtain the entity information of the closest new second maintenance subject in the first knowledge graph according to the cosine similarity calculation method of the entity information, and search for the third fault type existing in the second knowledge graph based on the second maintenance subject and the user's voice keywords, and find the guidance voice data corresponding to the third fault type based on the first key-value pair list, return the guidance voice data corresponding to the third fault type, and reconstruct the guidance dialogue with the user.

2. According to claim 1, an AI-based intelligent customer service telephone repair method is characterized in that: The method includes: using a speech recognition model to identify keywords in user repair voice data, searching the first knowledge graph according to the identified voice data keywords, searching for a first target maintenance subject from entity information in the first knowledge graph, and searching for a first fault type of the first target maintenance subject in entity information in the second knowledge graph based on the first target maintenance subject, obtaining corresponding first guidance voice data stored in the first key-value pair list according to the first fault type found, and returning the first guidance voice data to establish a guidance dialogue with the user.

3. According to claim 2, an AI-based intelligent customer service telephone repair method is characterized in that: After the first guidance voice data is returned and a guidance dialogue is established, the user voice data in the guidance dialogue is obtained, and the evaluation keywords of the user in the guidance dialogue are identified, and it is determined whether the first fault type currently being searched is a real fault type based on the evaluation keywords of the user in the guidance dialogue.

4. According to claim 1, the AI-based intelligent customer service telephone repair method is characterized in that: The method for constructing the guidance voice data includes: obtaining status information of the corresponding equipment fault type based on historical maintenance data and maintenance instruction manuals of different equipment, constructing standard question-based guidance voice data based on the status information of the corresponding equipment fault type, and using the question-based guidance voice data with the corresponding fault type entity name as the key name and its own guidance voice data as the key value to establish the first key-value pair list.

5. According to claim 1, the AI-based intelligent customer service telephone repair method is characterized in that: The method for processing the guided dialogue includes: using voice recognition technology to identify evaluation keywords in the guided dialogue, and determining whether the evaluation keywords are negative evaluation keywords; if they are negative evaluation keywords, re-searching the second knowledge graph for the second fault type entity information name that is most similar to the previous first fault type, and finding the second guidance voice data corresponding to the second fault type according to the first key-value pair list, and further returning the second guidance voice data to reconstruct the guided dialogue with the user.

6. The AI-based intelligent customer service telephone repair method according to claim 1, characterized in that: The method for processing the guided dialogue includes: when negative evaluation keywords in the guided dialogue are identified by voice recognition technology, and new keywords in the guided dialogue are identified and compared with maintenance subject keywords in the entity information of the first knowledge graph at the same time, if there are new maintenance subject keywords in the guided dialogue and negative evaluation keywords, then the fault type corresponding to the second knowledge graph is searched in the new maintenance subject in the first knowledge graph, and the corresponding guidance voice data is searched according to the first key-value pair list to return and reconstruct the guided dialogue.

7. The AI-based intelligent customer service telephone repair method according to claim 1, characterized in that: The method includes: constructing a second key-value pair list, wherein the second key-value pair list uses the fault type entity information of the second knowledge graph as the key name and the corresponding maintenance strategy as the key value; when there are positive evaluation keywords in the user voice data in the guiding dialogue, the second key-value pair list is queried according to the fault type entity information of the second knowledge graph currently constructing the guiding dialogue to obtain the corresponding maintenance strategy, and the maintenance strategy is sent to the target maintenance personnel to perform the maintenance operation.

8. An AI-based intelligent customer service telephone repair system, characterized in that: The system executes an AI-based intelligent customer service telephone repair reporting method as described in any one of claims 1-7 above.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and the computer program is executed by a processor to implement an AI-based intelligent customer service telephone repair reporting method as described in any one of claims 1 to 7.

Images