Large model enhanced customer service specification detection and analysis method and system

Abstract

The invention relates to the field of customer service management, and discloses a large model enhanced customer service specification detection and analysis method and system, and the method comprises the following steps: S1, accessing a service specification St and recording data Rt; s2, analyzing the St by a large model, and disassembling the St into detection key points c1,..., cn; s3, constructing a cue word template, and optimizing to generate prompt1,..., promptn; s4, converting the voice into the text, and converting the Rt into the text Ct; s5, instantiating the Ct to a cue word template, submitting large model detection, and generating result1,..., resultn; s6, comprehensively analyzing a detection result, and generating a conclusion; and S7, providing improvement suggestions based on the generated report. Through analysis and key point splitting of business specifications by a large model, the detection accuracy of the system reaches 97% or above, and meanwhile, the conditions of wrong detection and missing detection are effectively reduced through high-precision conversion of the voice-to-text technology.

Description

Technical Field

[0001] The present invention relates to the technical field of customer service management, and in particular to a large-model enhanced customer service specification detection and analysis method and system. Background Art

[0002] With the rapid development of enterprise informatization and customer service systems, customer service standard detection and analysis play a vital role in ensuring business compliance and improving customer satisfaction. The core of customer service standard detection is to monitor the operations of customer service personnel during business handling to ensure that they strictly follow the established business specifications to reduce violations and reduce business risks.

[0003] First, traditional customer service standard testing mostly relies on manual sampling or basic automated testing technology. Due to the lack of in-depth analysis of business specifications and item-by-item testing, the accuracy of testing is low, usually around 75%. In addition, key operations are easily missed or mis-tested during manual testing, making it difficult to comprehensively and accurately evaluate the implementation of customer service staff's business specifications, which brings greater business risks to enterprises.

[0004] Secondly, the existing manual sampling method not only has a low detection coverage rate and cannot fully detect all customer service recordings, but also has a long detection feedback cycle, usually taking several days or even weeks. This lag makes it impossible for customer service personnel to correct their irregular operations in a timely manner, which in turn affects the execution efficiency of subsequent business and reduces customer experience.

[0005] Finally, the existing customer service standard detection system usually only stays at the simple compliance judgment, and it is difficult to conduct in-depth analysis of non-standard operations and generate targeted improvement suggestions. When faced with a large amount of detection data, it is difficult for management to grasp the overall standard implementation from a macro perspective, and there is a lack of effective tools to optimize customer service operation processes and training strategies, which affects the efficiency and effectiveness of business management. Summary of the invention

[0006] In view of the deficiencies in the prior art, the present invention provides a large-model enhanced customer service specification detection and analysis method and system, which solves the problems of insufficient detection accuracy and serious false detection and missed detection.

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a large model enhanced customer service specification detection and analysis method, comprising the following steps:

[0008] S1. Standard input: access business standard specification S_t and recording data R_t;

[0009] S2, automatic specification analysis: the business specification S_t is understood and split into key points through the big model, and several detection points c1, c2, …, cn are generated;

[0010] S3, automatic prompt word template construction: Based on the built-in general template, construct the initial prompt word template and use the large model to optimize it, and generate the prompt word template prompt_1, prompt_2, ..., prompt_n for each key point;

[0011] S4, audio recording to text: convert the audio recording R_t into audio recording text C_t through speech-to-text technology;

[0012] S5. Key point detection: instantiate the recording text C_t into the generated prompt word template prompt_1, prompt_2,…, prompt_n, submit the large model for detection, and obtain the detection results result_1, result_2,…, result_n;

[0013] S6. Comprehensive detection analysis: Through the large model, the results of all detection points resu l t_1, resu l t_2,…, resu l t_n are comprehensively analyzed to generate the final detection conclusion;

[0014] S7. Generation of test analysis report: Statistical analysis is performed based on the test results of a certain period of time, and an analysis report is generated, which also provides reasons for non-compliance with the specifications and improvement suggestions.

[0015] Preferably, in the step S2, the automatic specification parsing decomposes and extracts key points of the business specification S_t through a large model;

[0016] The key points extracted include identity authentication, business processing and necessary reminder items.

[0017] Preferably, in the step S3, the automatic prompt word template construction includes a built-in universal prompt word template for key point detection, and adaptively optimizing each key point through a large model to generate prompt word templates suitable for different business specifications.

[0018] Preferably, in the step S4, the recording-to-text conversion uses speech recognition technology to convert the recording data R_t of the customer service handling business into the recording text C_t, and ensures that the accuracy of the transcribed text reaches a preset standard.

[0019] Preferably, in the step S6, the comprehensive analysis summarizes all key point detection results through prompt words, and requests the large model to output the final conclusion;

[0020] The final conclusion includes whether the specification is met and the specific reasons for not meeting the specification.

[0021] Preferably, in the step S7, the detection and analysis report is generated by a large model, which includes the current status of compliance with the specifications, statistics of records of non-compliance with the specifications, cause analysis and targeted improvement measures.

[0022] A large model-enhanced customer service specification detection and analysis system, including a business specification access module, a specification intelligent understanding module, a recording data access module, a detection result storage module and a report generation module;

[0023] The business specification access module is used to access the business specifications of all businesses that need to be detected and analyzed and store them in a database;

[0024] The specification intelligent understanding module decomposes the business specifications into key points based on the big model technology, generates detection key points and stores them in the specification template library;

[0025] The recording data access module is used to access the recording data of each business system in real time;

[0026] The recording intelligent detection module is used to convert the recording data into text through speech-to-text technology, and perform standardized detection and analysis based on detection points;

[0027] The detection result storage module stores the detection analysis results and returns them to the relevant interface;

[0028] The report generation module is used to perform statistical analysis on the standard detection and analysis results within a period of time, generate a standard detection and analysis report, and provide intelligent improvement suggestions.

[0029] Preferably, the business specification access module supports access to standard specifications of multiple business systems, and automatically classifies and stores the accessed specifications;

[0030] The specification intelligent understanding module parses and splits the business specification through a large model, breaks down the specification S_t into multiple key points c1, c2, ..., cn, and generates corresponding prompt word templates prompt_1, prompt_2, ..., prompt_n.

[0031] Preferably, the recording intelligent detection module converts the recording data R_t into text C_t based on the speech-to-text technology, and instantiates the detection of the standard detection key point template through the large model to generate the detection results result_1, result_2, ..., result_n;

[0032] The report generation module obtains the number and ratio of non-compliant records for each business and each customer service staff through statistical analysis, and analyzes the reasons for non-compliance with the standards. The statistical analysis submits the analysis results to the big model through the built-in prompt word template for further standard detection and analysis, and generates intelligent improvement measures to reduce non-standard operations.

[0033] Preferably, the detection and analysis report includes the number of records that do not meet the standards for each business, the number of records that do not meet the standards for each person, and statistics on reasons for non-compliance.

[0034] The present invention provides a large model enhanced customer service specification detection and analysis method and system. It has the following beneficial effects:

[0035] 1. The present invention uses a large model to parse and split the business specifications, and combines it with the optimized prompt word template to perform item-by-item detection. The system achieves a detection accuracy of more than 97%. At the same time, the high-precision conversion of speech-to-text technology ensures the reliability of the recorded text. The precise coordination of each link makes the detection results more comprehensive and accurate, effectively reducing the situation of false detection and missed detection, and providing enterprises with a more reliable customer service specification implementation evaluation.

[0036] 2. The present invention accesses all recording data of customer service when handling business in real time, ensuring that each recording undergoes complete standard testing and realizes full detection. At the same time, after the business is completed, the system can complete the standard inspection and generate feedback results within about 10 seconds, helping customers to learn and improve quickly. This real-time and comprehensiveness greatly improves the detection efficiency and avoids the lag problem caused by traditional manual post-inspection.

[0037] 3. The present invention generates intelligent detection and analysis reports regularly. The system not only provides detailed non-conformity records and cause analysis, but also generates targeted improvement suggestions through large models, such as strengthening specific business training or reminding key specifications. The management can grasp the implementation of customer service specifications from a macro perspective, and optimize the process through the suggestions in the report, further reduce business risks, and improve customer service quality and customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a flow chart of the method steps of the present invention;

[0039] Figure 2 It is a schematic diagram of the system module of the present invention. DETAILED DESCRIPTION

[0040] The following will be combined with the drawings in the specification 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.

[0041] Embodiment 1:

[0042] Please refer to the attached Figure 1 The embodiment of the present invention provides a large model enhanced customer service specification detection and analysis method, comprising the following steps:

[0043] S1. Standard input: access business standard specification S_t and recording data R_t;

[0044] S2, automatic specification analysis: the business specification S_t is understood and split into key points through the big model, and several detection points c1, c2, …, cn are generated;

[0045] S3, automatic prompt word template construction: Based on the built-in general template, construct the initial prompt word template and use the large model to optimize it, and generate the prompt word template prompt_1, prompt_2, ..., prompt_n for each key point;

[0046] S4, audio recording to text: convert the audio recording R_t into audio recording text C_t through speech-to-text technology;

[0047] S5. Key point detection: instantiate the recording text C_t into the generated prompt word template prompt_1, prompt_2,…, prompt_n, submit the large model for detection, and obtain the detection results result_1, result_2,…, result_n;

[0048] S6. Comprehensive detection analysis: Through the large model, the results of all detection points resu l t_1, resu l t_2,…, resu l t_n are comprehensively analyzed to generate the final detection conclusion;

[0049] S7. Generation of test analysis report: Statistical analysis is performed based on the test results of a certain period of time, and an analysis report is generated, which also provides reasons for non-compliance with the specifications and improvement suggestions;

[0050] In step S2, automatic specification parsing decomposes and extracts key points of the business specification S_t through a large model;

[0051] Key points extraction includes identity verification, business processing and required reminders;

[0052] In step S3, the automatic prompt word template construction includes a built-in universal key point detection prompt word template, and adaptively optimizes each key point through a large model to generate prompt word templates suitable for different business specifications;

[0053] In step S4, voice recording to text conversion uses speech recognition technology to convert the customer service recording data R_t into voice recording text C_t, and ensures that the accuracy of the transcribed text meets the preset standard;

[0054] In step S6, the comprehensive analysis summarizes all key point detection results through prompt words and requests the large model to output the final conclusion;

[0055] The final conclusion includes whether the specification is met and the specific reasons for non-compliance;

[0056] In step S7, the inspection and analysis report is generated through the large model, which includes the current status of compliance with the specifications, statistics of non-compliance with the specifications, cause analysis, and targeted improvement measures.

[0057] Specifically, in step S1, the system accesses the business standard specification S_t and the recording data R_t to ensure that the system can fully capture the business standards and actual customer service recordings, ensuring the integrity of the data and the coverage of the business specifications from the source. Then in step S2, the system uses the big model to deeply understand and split the key points of the business specification S_t, and decomposes the complex business specifications into multiple independent detection points c1, c2,…, cn (such as identity verification, business processing, reminder items, etc.), making the specification detection more detailed, ensuring that each key link can be evaluated separately to avoid missed detection or false detection. At the same time, this split ensures that the system can accurately identify and detect the compliance of various business processes in complex and changing business scenarios. In step S3, the system automatically constructs the initial prompt word template based on the built-in general template, and uses the big model for adaptive optimization to generate prompts for each detection point. Word template prompt_1, prompt_2,…, prompt_n. Through this optimized prompt word template, the system can better adapt to the needs of different business scenarios, greatly improve the detection accuracy and stability, avoid the limitations brought by fixed prompt word templates, and the average detection accuracy reaches more than 97%, which is much higher than the 75% accuracy of traditional systems. In step S4, the recording R_t is converted into text C_t with high precision through speech recognition technology to ensure the integrity and accuracy of the text, achieving a transcription accuracy of more than 95%, providing reliable basic data for subsequent detection. Step S5 is the core of key point detection. The system instantiates the recording text C_t to the previously generated prompt word template prompt_1, prompt_2,…, prompt_n, and submits the instantiated prompt words to the large model for item-by-item detection to obtain the detection result resu for each key point lt_1, resu lt_2,…, resu lt_n. This process ensures that each specification point can be strictly tested one by one, reducing the risk of errors and omissions. The entire testing process can be completed within 10 seconds, which is much more efficient than the existing manual sampling. Customers can quickly learn based on the feedback results to ensure immediate application in the next business transaction.In step S6, the system uses a large model to comprehensively analyze all key test results resu lt_1, resu lt_2, …, resu lt_n, and generates the final standard test conclusion. This conclusion not only indicates whether the business specifications are met, but also explains in detail the specific reasons for non-compliance with the specifications, ensuring the comprehensiveness, transparency and interpretability of the test results. Finally, in step S7, the system generates an analysis report based on the standard test data over a period of time. The report includes the compliance of each business specification, the statistics of non-compliance records, the cause analysis and targeted improvement measures. Through these reports, the management can clearly understand the overall situation of customer service specification execution and take corresponding optimization measures (such as strengthening training or reminders for a certain business), so as to continuously improve the quality and compliance of customer service. The automated analysis and improvement suggestions of the report can help enterprises reduce more than 50% of non-standard operations. ;

[0058] Embodiment 2:

[0059] Please refer to the attached Figure 2 , a large model enhanced customer service specification detection and analysis system, including a business specification access module, a specification intelligent understanding module, a recording data access module, a detection result storage module and a report generation module;

[0060] The business specification access module is used to access the business specifications of all businesses that need to be detected and analyzed and store them in the database;

[0061] The specification intelligent understanding module breaks down the business specifications into key points based on the big model technology, generates detection points and stores them in the specification template library;

[0062] The recording data access module is used to access the recording data of each business system in real time;

[0063] The recording intelligent detection module is used to convert recording data into text through speech-to-text technology, and perform standardized detection and analysis based on detection points;

[0064] The test result storage module stores the test analysis results and returns them to the relevant interface;

[0065] The report generation module is used to perform statistical analysis on the results of standard inspection and analysis within a period of time, generate standard inspection and analysis reports, and provide intelligent improvement suggestions;

[0066] The business specification access module supports access to standard specifications of multiple business systems and automatically classifies and stores the accessed specifications;

[0067] The specification intelligent understanding module parses and splits the business specifications through a large model, breaks down the specification S_t into multiple key points c1, c2, …, cn, and generates corresponding prompt word templates prompt_1, prompt_2, …, prompt_n;

[0068] The recording intelligent detection module converts the recording data R_t into text C_t based on the speech-to-text technology, and instantiates the standard detection point template through the large model to generate the detection results resu l t_1, resu l t_2,…, resu l t_n;

[0069] The report generation module obtains the number and ratio of non-compliant records for each business and each customer service staff through statistical analysis, and analyzes the reasons for non-compliance. The statistical analysis submits the analysis results to the big model through the built-in prompt word template for further standard detection and analysis, and generates intelligent improvement measures to reduce non-standard operations;

[0070] The detection and analysis report includes the number of non-compliant records for each business, the number of non-compliant records for each person, and the statistics of the reasons for non-compliant records.

[0071] Specifically, first, the business specification access module accesses the standard specifications of each business system and stores them in categories to ensure that the system can obtain and manage multiple business specifications. Then, the specification intelligent understanding module parses and disassembles the business specifications based on the big model technology, generates detection points and prompt word templates, and ensures that the specifications are accurately refined. The recording data access module accesses the recording data R_t of the business system in real time to ensure the real-time and comprehensiveness of the data. Then, the recording intelligent detection module converts the recording data into text C_t through speech-to-text technology, and uses the big model to analyze it item by item based on the detection points to generate the detection results of each point resu l t_1, resu l t_2,…, resu l t_n, the test result storage module stores the analysis results in the database and feeds them back to the relevant interface for the business system or management to call. The report generation module regularly or actively generates analysis reports based on the test data. By counting the non-compliant records of each business and customer service personnel, it deeply analyzes the reasons for non-compliance with the regulations and further generates intelligent improvement suggestions to help enterprises optimize management processes. The advantage of this system is that it can fully detect all recordings to ensure that there is no missed detection. The system's detection accuracy is as high as over 97%. The system optimizes the prompt word template through a large model and detects the business points item by item to ensure the accuracy of the test results. After accessing the recording data, the system can complete the detection and feedback within 10 seconds, which greatly improves the timeliness of business feedback and avoids the lag caused by traditional manual sampling. At the same time, the generation of intelligent reports helps management understand the compliance execution of each business from a macro perspective and provide specific improvement suggestions, ultimately improving the compliance and management efficiency of customer service.

[0072] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A large model-enhanced customer service specification detection and analysis method, characterized in that: The following steps are involved: S1. Standard input: access business standard specification S_t and recording data R_t; S2, automatic specification analysis: the business specification S_t is understood and split into key points through the big model, and several detection points c1, c2, …, cn are generated; S3, automatic prompt word template construction: Based on the built-in general template, construct the initial prompt word template and use the large model to optimize it, and generate the prompt word template prompt_1, prompt_2, ..., prompt_n for each key point; S4, audio recording to text: convert the audio recording R_t into audio recording text C_t through speech-to-text technology; S5, key point detection: instantiate the recording text C_t to the generated prompt word template prompt_1, prompt_2, ..., prompt_n, submit the large model for detection, and obtain the detection results result_1, result_2, ..., result_n; S6, comprehensive detection analysis: through the large model, the results of all detection points (result_1, result_2,…,result_n) are comprehensively analyzed to generate the final detection conclusion; S7. Generation of test analysis report: Statistical analysis is performed based on the test results of a certain period of time, and an analysis report is generated, which also provides reasons for non-compliance with the specifications and improvement suggestions.

2. A large model enhanced customer service specification detection and analysis method according to claim 1, characterized in that: In the step S2, the automatic specification parsing decomposes and extracts key points of the business specification S_t through the large model; The key points extracted include identity authentication, business processing and necessary reminder items.

3. A large model enhanced customer service specification detection and analysis method according to claim 1, characterized in that: In the step S3, the automatic prompt word template construction includes a built-in universal key point detection prompt word template, and adaptively optimizing each key point through a large model to generate prompt word templates suitable for different business specifications.

4. A large model enhanced customer service specification detection and analysis method according to claim 1, characterized in that: In the step S4, the recording-to-text conversion uses speech recognition technology to convert the recording data R_t of the customer service handling business into the recording text C_t, and ensures that the accuracy of the transcribed text meets the preset standard.

5. A large model enhanced customer service specification detection and analysis method according to claim 1, characterized in that: In the step S6, the comprehensive analysis summarizes all key point detection results through prompt words and requests the large model to output the final conclusion; The final conclusion includes whether the specification is met and the specific reasons for not meeting the specification.

6. A large model enhanced customer service specification detection and analysis method according to claim 1, characterized in that: In the step S7, the inspection and analysis report is generated through the large model, which includes the current status of compliance with the specifications, statistics of non-compliance records, cause analysis, and targeted improvement measures.

7. A large model enhanced customer service specification detection and analysis system, according to a large model enhanced customer service specification detection and analysis method according to any one of claims 1 to 6, characterized in that: It includes business specification access module, specification intelligent understanding module, recording data access module, test result storage module and report generation module; The business specification access module is used to access the business specifications of all businesses that need to be detected and analyzed and store them in a database; The specification intelligent understanding module decomposes the business specifications into key points based on the big model technology, generates detection key points and stores them in the specification template library; The recording data access module is used to access the recording data of each business system in real time; The recording intelligent detection module is used to convert the recording data into text through speech-to-text technology, and perform standardized detection and analysis based on detection points; The detection result storage module stores the detection analysis results and returns them to the relevant interface; The report generation module is used to perform statistical analysis on the standard detection and analysis results within a period of time, generate a standard detection and analysis report, and provide intelligent improvement suggestions.

8. A large model enhanced customer service specification detection and analysis system according to claim 7, characterized in that: The business specification access module supports access to standard specifications of multiple business systems and automatically classifies and stores the accessed specifications; The specification intelligent understanding module parses and splits the business specification through a large model, breaks down the specification S_t into multiple key points c1, c2, ..., cn, and generates corresponding prompt word templates prompt_1, prompt_2, ..., prompt_n.

9. A large model enhanced customer service specification detection and analysis system according to claim 7, characterized in that: The recording intelligent detection module converts the recording data R_t into text C_t based on the speech-to-text technology, and instantiates the standard detection key point template through the large model to generate the detection results result_1, result_2, ..., result_n; The report generation module obtains the number and ratio of non-compliant records for each business and each customer service staff through statistical analysis, and analyzes the reasons for non-compliance with the standards. The statistical analysis submits the analysis results to the big model through the built-in prompt word template for further standard detection and analysis, and generates intelligent improvement measures to reduce non-standard operations.

10. A large model enhanced customer service specification detection and analysis system according to claim 7, characterized in that: The detection and analysis report includes the number of records that do not meet the standards for each business, the number of records that do not meet the standards for each person, and statistics on reasons for non-compliance.

Images