Intelligent customer service implementation method and device based on Internet of Vehicles, equipment and medium

Abstract

The invention discloses an intelligent customer service implementation method based on the Internet of Vehicles, and the method comprises the steps: receiving fault information of a fault phenomenon inputted by a user, and carrying out the preprocessing of the fault information; inputting the preprocessed fault information into a large language model, and outputting a fault analysis subtask; determining an available tool and tool input of the available tool from a given tool set according to the fault analysis sub-task, and executing the fault analysis sub-task by using the available tool to obtain a task execution result; when the task execution result meets the user requirement, outputting the task execution result; according to the method, the fault problem is disassembled into a plurality of sub-tasks by using the large language model, so that the effect and efficiency of Internet of Vehicles task execution can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field, and in particular to a method, device, equipment and medium for realizing intelligent customer service based on Internet of Vehicles. Background Art

[0002] The Internet of Vehicles customer service system is designed to perform various tasks and provide assistance by simulating the functions of human assistants. It can understand the fault information input by car company users and perform corresponding actions or provide information. In the current digital era, the requirements for the Internet of Vehicles customer service system are higher. It needs to be able to perform complex tasks and provide personalized and intelligent services.

[0003] The current Internet of Vehicles customer service system collects relevant data and performs a lot of labeling work for the Internet of Vehicles scenario. The process of collecting and labeling data requires a lot of manual participation, consumes a huge amount of manpower and material resources, is inefficient, and has low data quality. It is unable to use a lot of industry prior knowledge, resulting in unsatisfactory task execution results and inability to achieve truly high intelligence. Summary of the invention

[0004] The embodiment of the present invention provides a method for realizing intelligent customer service based on the Internet of Vehicles, which uses a large language model to decompose a fault problem into multiple subtasks, thereby improving the effect and efficiency of executing Internet of Vehicles tasks.

[0005] In a first aspect, an embodiment of the present invention provides a method for implementing intelligent customer service based on an Internet of Vehicles, including:

[0006] Receiving fault information of a fault phenomenon input by a user, and preprocessing the fault information;

[0007] Input the preprocessed fault information into the large language model and output the fault analysis subtask;

[0008] Determine an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and use the available tool to execute the fault analysis subtask to obtain a task execution result;

[0009] When the task execution result meets the user's requirements, the task execution result is output.

[0010] Furthermore, the tool set includes at least one of a data query interface tool, a fault analysis interface tool, a text2sql tool and a standardized output tool.

[0011] Further, the preprocessing of the fault information includes:

[0012] Based on preset business rules, determine whether the fault information is complete;

[0013] If not, outputting prompt information according to the business rule, so that the user can re-enter the fault information of the fault phenomenon according to the prompt information;

[0014] If so, the fault information is concatenated with a preset prompt word template to obtain concatenated fault information.

[0015] Further, determining a tool to be used from a given tool set according to the fault analysis subtask includes:

[0016] Vectorizing the fault analysis subtask and each tool in the tool set to obtain a fault analysis subtask vector and a tool vector for each tool;

[0017] Calculating the cosine similarity between the fault analysis subtask vector and each of the tool vectors;

[0018] The tool vectors whose cosine similarity is greater than a preset similarity threshold are eliminated, and the tool corresponding to the tool vector with the highest cosine similarity is selected from the eliminated tool vectors as the tool to be used.

[0019] Furthermore, the tool set includes a standardized output tool for filtering sensitive words in the fault information and assembling them into standard natural sentences. The standardized output tool needs to be trained, and the training process includes:

[0020] Acquire historical fault data of historical fault phenomena; wherein the historical fault data includes historical fault information input by a user and historical fault analysis results and standardized output results corresponding to the historical fault information;

[0021] Dividing the historical fault data into a training set and a test set according to a preset division ratio;

[0022] The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained standardized output tool.

[0023] Furthermore, the tool set includes a text2sql tool for converting natural language input into executable database query statements. The text2sql tool needs to be trained, and the training process includes:

[0024] Obtain historical natural query statements and corresponding historical database query statements;

[0025] Dividing the historical natural query statements and the historical database query statements into a training set and a test set according to a preset division ratio;

[0026] The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained text2sql tool.

[0027] Furthermore, the method further comprises:

[0028] When the task execution result does not meet the user requirements, the fault information and the task execution result are spliced, the spliced ​​text is input into the large language model, the fault analysis subtask is output, and the steps of determining the usable tools from the given tool set according to the fault analysis subtask and subsequent steps are repeated until the task execution result meets the user requirements.

[0029] In a second aspect, an embodiment of the present invention provides an intelligent customer service implementation device based on an Internet of Vehicles, including:

[0030] A data preprocessing module, used to receive fault information of a fault phenomenon input by a user, and preprocess the fault information;

[0031] A task generation module, used for inputting the preprocessed fault information into a large language model and outputting a fault analysis subtask;

[0032] A task execution module, used to determine an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and use the available tool to execute the fault analysis subtask to obtain a task execution result;

[0033] The result output module is used to output the task execution result when the task execution result meets the user's requirements.

[0034] In a third aspect, an embodiment of the present invention provides an electronic device, including:

[0035] Memory for storing computer programs;

[0036] A processor, configured to execute the computer program;

[0037] Among them, when the processor executes the computer program, it implements the method for realizing intelligent customer service based on the Internet of Vehicles as described in any one of the first aspects above.

[0038] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed, the method for realizing intelligent customer service based on the Internet of Vehicles as described in any one of the first aspects above is implemented.

[0039] Compared with the prior art, an embodiment of the present invention provides an intelligent customer service implementation method based on the Internet of Vehicles, which has the following beneficial effects: by receiving fault information of a fault phenomenon input by a user and preprocessing the fault information; inputting the preprocessed fault information into a large language model, and outputting a fault analysis subtask; determining an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and using the available tool to execute the fault analysis subtask to obtain a task execution result; when the task execution result meets the user's requirements, outputting the task execution result; the present invention uses a large language model to decompose the fault problem into multiple subtasks, which can improve the effect and efficiency of Internet of Vehicles task execution. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] In order to more clearly illustrate the technical features of the embodiments of the present invention, the drawings required for use in the embodiments of the present invention will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without paying any creative work.

[0041] Figure 1 It is a flowchart of an embodiment of a method for realizing intelligent customer service based on Internet of Vehicles provided by the present invention;

[0042] Figure 2 It is a structural schematic diagram of an embodiment of an intelligent customer service implementation device based on Internet of Vehicles provided by the present invention;

[0043] Figure 3 It is a structural schematic diagram of an embodiment of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0044] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 are within the scope of protection of the present invention.

[0045] In the first aspect, the embodiment of the present invention provides a method for realizing intelligent customer service based on the Internet of Vehicles, see Figure 1 , which is a flow chart of an embodiment of a method for realizing intelligent customer service based on Internet of Vehicles provided by the present invention.

[0046] like Figure 1 As shown, the method comprises the following steps:

[0047] S1: receiving fault information of a fault phenomenon input by a user, and preprocessing the fault information;

[0048] S2: input the preprocessed fault information into the large language model, and output the fault analysis subtask;

[0049] S3: determining an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and using the available tool to execute the fault analysis subtask to obtain a task execution result;

[0050] S4: When the task execution result meets the user's requirements, output the task execution result.

[0051] In a specific implementation, firstly, fault information about a fault phenomenon input by a user is received, and the fault information is preprocessed, and the preprocessed fault information is input into a large language model for reasoning. The large language model decomposes the fault problem into one or more fault analysis subtasks according to the input Internet of Vehicles fault information, available tools, and the returned format. According to the fault analysis subtask, available tools and tool inputs of the available tools are determined from a given tool set, and the available tools are used to execute the fault analysis subtask to obtain a task execution result, and it is determined whether the task execution result can meet the user's requirements. When the user's requirements are met, the task execution result is fed back to the user.

[0052] In summary, the present invention receives fault information of a fault phenomenon input by a user and preprocesses the fault information; inputs the preprocessed fault information into a large language model, and outputs a fault analysis subtask; determines usable tools and tool inputs of the usable tools from a given tool set according to the fault analysis subtask, and uses the usable tools to execute the fault analysis subtask to obtain a task execution result; when the task execution result meets the user's requirements, outputs the task execution result; the present invention uses a large language model to decompose the fault problem into multiple subtasks, which can improve the effect and efficiency of the execution of Internet of Vehicles tasks.

[0053] In an optional implementation, the tool set includes at least one of a data query interface tool, a fault analysis interface tool, a text2sql tool, and a standardized output tool.

[0054] Specifically, the tool set is used for the interaction between the large language model and the external API to improve the large language model's ability to delimit, locate, analyze and process vehicle network faults. The tool set includes at least one of a data query interface tool, a fault analysis interface tool, a text2sql tool and a standardized output tool. The data query interface tool is used to execute database query statements and return query results. For example, when it is necessary to query the historical payment records of a user with an id of 123412, the input and output of the data query interface are as follows:

[0055] Input of the data query interface:

[0056] select payment_history from online_data where id=123412

[0057] Output of the data query interface:

[0058] id|payment_history

[0059] 12341|2024-03-15 00:00:00

[0060] 12341|2024-04-15 00:00:00

[0061] Furthermore, the fault analysis interface tool is used for fault analysis of the Internet of Vehicles. It determines whether there is a fault problem based on business rules, and returns abnormal data and possible fault problems based on the input dictionary format data. For example, when performing fault analysis on the Internet of Vehicles data transmission rate, the input Internet of Vehicles data transmission rate indicator data is received, and abnormal data points of the Internet of Vehicles data transmission rate are screened out according to the 3-sigma principle. The 3-sigma principle is to calculate the standard deviation σ and mean μ of the sequence under the assumption that the Internet of Vehicles data transmission rate indicator sequence is normally distributed. When the data point is not in the range of (μ-3σ, μ+3σ), it is an abnormal data point. After the abnormal data points of the Internet of Vehicles data transmission rate are screened out, the fault is determined to be an abnormal Internet of Vehicles data transmission rate fault according to the business rules, and the abnormal data points in dictionary format are output. The input and output of the fault analysis interface are as follows:

[0062] Input of the fault analysis interface:

[0063] {"id":["12341","12341","12341","12341"],

[0064] "data_transfer_rate":[89,40,30,97],

[0065] "process_time":["2024-03-15 00:00:00","2024-03-16 00:00:00","2024-03-1700:00:00","2024-03-18 00:00:00"]}

[0066] Output of the fault analysis interface:

[0067] {"id":["12341","12341"],

[0068] "data_transfer_rate":[40,30],

[0069] "process_time":["2024-03-17 00:00:00","2024-03-18 00:00:00"],

[0070] Is there a fault?: 1,

[0071] "Fault problem": "Vehicle network data transmission rate is abnormal"}

[0072] It can be understood that the text2sql tool is used to convert natural language input into executable database query statements. For example, when the payment record of the user with id 123412 needs to be queried, the input and output of the text2sql tool are as follows:

[0073] Input for the text2sql tool:

[0074] Query the payment record of user id 123412

[0075] Output of text2sql tool:

[0076] select payment_history from online_data where id=123412

[0077] The standardized output tool is used to filter out relevant sensitive words in the input sentences for different types of user input, and assemble them into standard natural language sentences and return them to the user. For example, the input and output of the standardized output tool are as follows:

[0078] Standardize the input to the output tool:

[0079] "The fault phenomenon is that the vehicle fails to connect to the Internet, "2024-04-24 15:20:35, 50 meters west of the intersection of C Road and D Road in B City, Province A"

[0080] Output of the standard output tool:

[0081] "After checking, it was detected that the vehicle failed to connect to the Internet at 15:20:35 on April 24, 2024, in the area between Road C and Road D in City B, Province A. It was found that the user's fee was insufficient, which caused the vehicle to be unable to connect to the Internet. Please recharge the fee as soon as possible."

[0082] In an optional implementation manner, the preprocessing the fault information includes:

[0083] Based on preset business rules, determine whether the fault information is complete;

[0084] If not, outputting prompt information according to the business rule, so that the user can re-enter the fault information of the fault phenomenon according to the prompt information;

[0085] If so, the fault information is concatenated with a preset prompt word template to obtain concatenated fault information.

[0086] Specifically, receive fault information about the fault phenomenon input by the user, and determine whether the fault information input by the user is complete based on the preset fault analysis business rules. If incomplete, guide the user to supplement the missing fault information. According to the preset business rules, output prompt information, prompt the user to supplement according to the prompt information, and re-enter the supplemented fault information. If complete, concatenate the fault information with the preset prompt word template in a string concatenation manner to obtain the concatenated fault information as the input of the large language model.

[0087] For example, when the input fault information is "The vehicle currently has a networking failure, please help me find out what the problem is", the fault information obtained after splicing with the preset prompt word template is as follows:

[0088] "Handling Internet of Vehicles faults must include the location of the fault, the time of the fault, and the fault phenomenon. Determine whether the input question contains the above elements and answer the question in the following format. You can select the tools in the following tool sets:

[0089] {text2sql tool, data query interface tool, data analysis interface tool, standardized output tool}

[0090] Please follow the following format enclosed in brackets:

[0091] ---

[0092] Question: Question that needs to be answered

[0093] Think about the steps you need to take to answer this question

[0094] Action: One of the tool names in the tool set must be among {text2sql tool, data query interface tool, data analysis interface tool, standardized output tool}

[0095] Action Input: Select the input required by the tool

[0096] Observation: Select tool returns results

[0097] ...(thinking / action / action input / observation can be repeated N times)

[0098] Thinking: Now you know the final answer

[0099] Result: Final Answer

[0100] ---

[0101] Start answering now, and follow the given format to reason step by step before giving your final answer.

[0102] Question: The vehicle is currently experiencing a networking failure. Please help me find out what the problem is."

[0103] In an optional implementation, determining a tool to be used from a given tool set according to the fault analysis subtask includes:

[0104] Vectorizing the fault analysis subtask and each tool in the tool set to obtain a fault analysis subtask vector and a tool vector for each tool;

[0105] Calculating the cosine similarity between the fault analysis subtask vector and each of the tool vectors;

[0106] The tool vectors whose cosine similarity is greater than a preset similarity threshold are eliminated, and the tool corresponding to the tool vector with the highest cosine similarity is selected from the eliminated tool vectors as the tool to be used.

[0107] Specifically, the fault analysis subtask and each tool in the toolset are respectively converted into a vector in a high-dimensional space to obtain the fault analysis subtask vector and the tool vector of each tool for mathematical operations. Each tool vector in the toolset is traversed to calculate its cosine similarity with the fault analysis subtask vector. Cosine similarity is a measure of the degree of similarity between two vectors in direction. Its value range is from -1 to 1. The closer the value is to 1, the more similar the two vectors are. A similarity threshold is set to determine whether the two vectors are similar enough. All tool vectors whose cosine similarity with the fault analysis subtask vector is greater than the threshold are eliminated. Among the eliminated tool vectors, the tool corresponding to the tool vector with the highest cosine similarity is selected as the tool to be used.

[0108] In an optional implementation, the tool set includes a standardized output tool for filtering sensitive words in the fault information and assembling them into standard natural sentences. The standardized output tool needs to be trained, and the training process includes:

[0109] Acquire historical fault data of historical fault phenomena; wherein the historical fault data includes historical fault information input by a user and historical fault analysis results and standardized output results corresponding to the historical fault information;

[0110] Dividing the historical fault data into a training set and a test set according to a preset division ratio;

[0111] The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained standardized output tool.

[0112] Specifically, from a business perspective, in order to unify the official standardized caliber response, the output is filtered for sensitive words, for example, the specific address is blurred. First, historical fault data of users on historical fault phenomena are collected, including historical fault information input by users, corresponding historical fault analysis results and corresponding standardized output results, which are used as data sets. The pre-trained large language model is used for full parameter adjustment training to meet the needs of standardized natural language. The historical fault data is divided into a training set and a test set according to a preset division ratio. Exemplarily, it can be divided in a ratio of 9:1, and the learning rate is set to 0.001, the batch size is 4, and the iteration cycle is 2. The pre-trained large language model is trained for full parameter adjustment using the training set, and the effect of fine-tuning the large language model is verified on the test set. The full parameter adjustment training is based on the original large language model without freezing all parameters. The data extracted from the training set is forward propagated. After the model output is obtained, the loss is calculated with the training set label. Finally, back propagation is performed according to the calculated loss value to update the large language model parameters to meet the needs of standardized output in the Internet of Vehicles scenario, and a standardized output tool is obtained.

[0113] In an optional implementation, the tool set includes a text2sql tool for converting natural language input into executable database query statements. The text2sql tool needs to be trained, and the training process includes:

[0114] Obtain historical natural query statements and corresponding historical database query statements;

[0115] Dividing the historical natural query statements and the historical database query statements into a training set and a test set according to a preset division ratio;

[0116] The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained text2sql tool.

[0117] Specifically, in order to meet the need of converting natural language input into executable database query statements in the Internet of Vehicles scenario, historical natural query statements input in the Internet of Vehicles scenario and corresponding historical database query statements are collected, and the historical natural query statements and the historical database query statements are divided into a training set and a test set according to a preset division ratio. Exemplarily, the division can be performed in a ratio of 9:1, and the optimizer is set to Adam, the learning rate is 0.001, the batch size is 4, and the iteration cycle is 4. The training set is used to perform full parameter adjustment training on the pre-trained large language model, and the effect of the fine-tuned text2sql is verified on the test set. The data extracted from the training set is forward propagated and calculated. After obtaining the model output, the loss is calculated with the training set label. Finally, back propagation is performed according to the calculated loss value to update the large language model parameters to obtain the text2sql tool.

[0118] In an optional embodiment, the method further includes:

[0119] When the task execution result does not meet the user requirements, the fault information and the task execution result are spliced, the spliced ​​text is input into the large language model, the fault analysis subtask is output, and the steps of determining the usable tools from the given tool set according to the fault analysis subtask and subsequent steps are repeated until the task execution result meets the user requirements.

[0120] Specifically, it is determined whether the obtained task execution result can respond to the user's requirements. If it meets the user's requirements, it can be directly output and fed back to the user. If it does not meet the user's requirements, steps S2 to S4 are repeated until the task execution result meets the user's requirements.

[0121] In a second aspect, an embodiment of the present invention provides an intelligent customer service implementation device based on the Internet of Vehicles, see Figure 2 , which is a structural diagram of an embodiment of an intelligent customer service implementation device based on the Internet of Vehicles provided by the present invention.

[0122] like Figure 2 As shown, the device comprises:

[0123] The data preprocessing module 21 is used to receive the fault information of the fault phenomenon input by the user and preprocess the fault information;

[0124] A task generation module 22, used for inputting the pre-processed fault information into a large language model and outputting a fault analysis subtask;

[0125] A task execution module 23, used to determine an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and use the available tool to execute the fault analysis subtask to obtain a task execution result;

[0126] The result output module 24 is used to output the task execution result when the task execution result meets the user's requirements.

[0127] In an optional implementation, the tool set includes at least one of a data query interface tool, a fault analysis interface tool, a text2sql tool, and a standardized output tool.

[0128] In an optional implementation, the data preprocessing module 21 is further used for:

[0129] Based on preset business rules, determine whether the fault information is complete;

[0130] If not, outputting prompt information according to the business rule, so that the user can re-enter the fault information of the fault phenomenon according to the prompt information;

[0131] If so, the fault information is concatenated with a preset prompt word template to obtain concatenated fault information.

[0132] In an optional implementation, the task execution module 23 is further used to:

[0133] Vectorizing the fault analysis subtask and each tool in the tool set to obtain a fault analysis subtask vector and a tool vector for each tool;

[0134] Calculating the cosine similarity between the fault analysis subtask vector and each of the tool vectors;

[0135] The tool vectors whose cosine similarity is greater than a preset similarity threshold are eliminated, and the tool corresponding to the tool vector with the highest cosine similarity is selected from the eliminated tool vectors as the tool to be used.

[0136] In an optional implementation, the tool set includes a standardized output tool for filtering sensitive words in the fault information and assembling them into standard natural sentences. The standardized output tool needs to be trained, and the training process includes:

[0137] Acquire historical fault data of historical fault phenomena; wherein the historical fault data includes historical fault information input by a user and historical fault analysis results and standardized output results corresponding to the historical fault information;

[0138] Dividing the historical fault data into a training set and a test set according to a preset division ratio;

[0139] The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained standardized output tool.

[0140] In an optional implementation, the tool set includes a text2sql tool for converting natural language input into executable database query statements. The text2sql tool needs to be trained, and the training process includes:

[0141] Obtain historical natural query statements and corresponding historical database query statements;

[0142] Dividing the historical natural query statements and the historical database query statements into a training set and a test set according to a preset division ratio;

[0143] The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained text2sql tool.

[0144] In an optional implementation, the device further includes a result verification module, and the result verification module is used to:

[0145] When the task execution result does not meet the user requirements, the fault information and the task execution result are spliced, the spliced ​​text is input into the large language model, the fault analysis subtask is output, and the steps of determining the usable tools from the given tool set according to the fault analysis subtask and subsequent steps are repeated until the task execution result meets the user requirements.

[0146] In a third aspect, an embodiment of the present invention provides an electronic device, see Figure 3 , which is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention.

[0147] like Figure 3 As shown, the device includes:

[0148] A memory 31, used for storing computer programs;

[0149] A processor 32, configured to execute the computer program;

[0150] Wherein, when the processor 32 executes the computer program, it implements the intelligent customer service implementation method based on the Internet of Vehicles as described in any of the above embodiments.

[0151] Exemplarily, the computer program may be divided into one or more modules / units, which are stored in the memory 31 and executed by the processor 32 to implement the present invention. The one or more modules / units may be a series of computer program instruction segments capable of implementing specific functions, which are used to describe the execution process of the computer program in the electronic device.

[0152] The processor 32 may be a central processing unit (CPU), or 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. A general-purpose processor may be a microprocessor or any conventional processor, etc.

[0153] The memory 31 can be used to store the computer program and / or module, and the processor 32 realizes various functions of the electronic device by running or executing the computer program and / or module stored in the memory 31, and calling the data stored in the memory 31. The memory 31 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, an application 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, a phone book, etc.), etc. In addition, the memory 31 can include a high-speed random access memory, and can also include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart memory card (Smart Media Card, SMC), a secure digital (SecureDigital, SD) card, a flash card (Flash Card), at least one disk storage device, a flash memory device, or other volatile solid-state storage devices.

[0154] It should be noted that the above electronic device includes, but is not limited to, a processor and a memory. Those skilled in the art can understand that Figure 3 The structural diagram is merely an example of the electronic device described above and does not constitute a limitation on the electronic device, and may include more components than shown in the figure, or a combination of certain components, or different components.

[0155] In a fourth aspect, an embodiment of the present invention further provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed, the method for realizing intelligent customer service based on the Internet of Vehicles as described in any of the above embodiments is implemented.

[0156] It should be understood that the present invention implements all or part of the processes in the above-mentioned intelligent customer service implementation method based on the Internet of Vehicles, and can also be completed by instructing related hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, it can implement the steps of the above-mentioned intelligent customer service implementation method based on the Internet of Vehicles. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electric carrier signal, telecommunication signal and software distribution medium, etc.

[0157] The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. It should be pointed out that for those skilled in the art, several equivalent obvious variations and / or equivalent substitutions can be made without departing from the technical principles of the present invention. These obvious variations and / or equivalent substitutions should also be regarded as the protection scope of the present invention.

Claims

1. A method for realizing intelligent customer service based on Internet of Vehicles, characterized in that: include: Receiving fault information of a fault phenomenon input by a user, and preprocessing the fault information; Input the preprocessed fault information into the large language model and output the fault analysis subtask; Determine an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and use the available tool to execute the fault analysis subtask to obtain a task execution result; When the task execution result meets the user's requirements, the task execution result is output.

2. The method for realizing intelligent customer service based on Internet of Vehicles according to claim 1, characterized in that: The tool set includes at least one of a data query interface tool, a fault analysis interface tool, a text2sql tool and a standardized output tool.

3. The method for realizing intelligent customer service based on Internet of Vehicles according to claim 1, characterized in that: The preprocessing of the fault information includes: Based on preset business rules, determine whether the fault information is complete; If not, outputting prompt information according to the business rule, so that the user can re-enter the fault information of the fault phenomenon according to the prompt information; If so, the fault information is concatenated with a preset prompt word template to obtain concatenated fault information.

4. The method for realizing intelligent customer service based on Internet of Vehicles according to claim 1, characterized in that: Determining the tool to be used from a given tool set according to the fault analysis subtask includes: Vectorizing the fault analysis subtask and each tool in the tool set to obtain a fault analysis subtask vector and a tool vector for each tool; Calculating the cosine similarity between the fault analysis subtask vector and each of the tool vectors; The tool vectors whose cosine similarity is greater than a preset similarity threshold are eliminated, and the tool corresponding to the tool vector with the highest cosine similarity is selected from the eliminated tool vectors as the tool to be used.

5. The method for realizing intelligent customer service based on Internet of Vehicles according to claim 1, characterized in that: The tool set includes a standardized output tool, which is used to filter sensitive words in the fault information and assemble them into standard natural sentences. The standardized output tool needs to be trained, and the training process includes: Acquire historical fault data of historical fault phenomena; wherein the historical fault data includes historical fault information input by a user and historical fault analysis results and standardized output results corresponding to the historical fault information; Dividing the historical fault data into a training set and a test set according to a preset division ratio; The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained standardized output tool.

6. The method for realizing intelligent customer service based on Internet of Vehicles according to claim 1, characterized in that: The tool set includes a text2sql tool for converting natural language input into executable database query statements. The text2sql tool needs to be trained, and the training process includes: Obtain historical natural query statements and corresponding historical database query statements; Dividing the historical natural query statements and the historical database query statements into a training set and a test set according to a preset division ratio; The large language model is trained using the training set, and the large language model is verified using the test set to obtain a trained text2sql tool.

7. The method for realizing intelligent customer service based on Internet of Vehicles according to claim 1, characterized in that: The method further comprises: When the task execution result does not meet the user requirements, the fault information and the task execution result are spliced, the spliced ​​text is input into the large language model, the fault analysis subtask is output, and the steps of determining the usable tools from the given tool set according to the fault analysis subtask and subsequent steps are repeated until the task execution result meets the user requirements.

8. An intelligent customer service implementation device based on Internet of Vehicles, characterized in that: include: A data preprocessing module, used to receive fault information of a fault phenomenon input by a user, and preprocess the fault information; A task generation module, used for inputting the preprocessed fault information into a large language model and outputting a fault analysis subtask; A task execution module, used to determine an available tool and a tool input of the available tool from a given tool set according to the fault analysis subtask, and use the available tool to execute the fault analysis subtask to obtain a task execution result; The result output module is used to output the task execution result when the task execution result meets the user's requirements.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, configured to execute the computer program; Wherein, when the processor executes the computer program, it implements the intelligent customer service implementation method based on the Internet of Vehicles as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed, the method for realizing intelligent customer service based on the Internet of Vehicles as described in any one of claims 1 to 7 is implemented.

Images