Customer support and feedback analysis processing method based on large language model

Abstract

The invention discloses a customer support and feedback analysis processing method based on a large language model, and the method comprises the following specific steps: obtaining customer support and feedback data, and judging whether the customer support and feedback data are of a voice type or not; preprocessing the client support and feedback data to obtain input data of a large language model; based on a plurality of independent large language models, performing sentiment analysis, theme recognition and question classification on the input data at the same time to obtain corresponding analysis results; scoring the analysis results and selecting the analysis result with the highest score as an initial optimal solution; circularly feeding back the initial optimal solution to a plurality of independent large language models for further analysis, and obtaining a target optimal solution through convergence evaluation; and matching the target optimal solution with the knowledge base. By combining the large language model and the knowledge base, client support requests and feedback data are processed in real time, the response speed and accuracy are improved, deep feedback analysis is provided, and service decisions are supported.

Description

Technical Field

[0001] The present invention relates to the field of customer support and feedback analysis, and in particular to a customer support and feedback analysis processing method based on a large language model. Background Art

[0002] Enterprises generally handle customer inquiries and feedback through customer support systems. However, traditional customer support systems usually rely on manual customer service or basic text analysis tools, which have significant limitations in processing speed and accuracy. Especially when faced with a large number of diverse customer needs, traditional methods are difficult to provide accurate responses and feedback analysis in real time and efficiently. In addition, traditional systems usually lack the ability to support multiple languages ​​and deep semantic understanding, and cannot meet the needs of global business.

[0003] In recent years, with the development of natural language processing technology, large language models have shown excellent text understanding and generation capabilities. However, there are still many challenges in applying these technologies to actual business scenarios, such as customer support and feedback analysis, especially when dealing with complex and diverse customer feedback, which requires a combination of rich business knowledge and contextual understanding to provide high-quality responses and analysis.

[0004] Therefore, it is urgent to propose a real-time customer support and feedback analysis method. Summary of the invention

[0005] The present invention provides a customer support and feedback analysis and processing method based on a large language model. By combining a large language model and a knowledge base, customer support requests and feedback data are processed in real time, the response speed and accuracy are improved, and in-depth feedback analysis is provided to support business decisions.

[0006] The technical solution adopted by the present invention is:

[0007] A customer support and feedback analysis and processing method based on a large language model comprises the following specific steps:

[0008] S1. Obtain customer support and feedback data, and determine whether the customer support and feedback data is of voice type: if so, perform voice transcription on the customer support and feedback data; if not, proceed to the next step;

[0009] S2. Preprocessing the customer support and feedback data to obtain input data for a large language model;

[0010] S3, based on multiple independent large language models, respectively, the input data is subjected to sentiment analysis, topic identification and question classification to obtain corresponding analysis results;

[0011] S4, scoring the analysis results and selecting the analysis result with the highest score as the initial optimal solution;

[0012] S5, looping and feeding back the initial optimal solution to multiple independent large language models for further analysis, and obtaining the target optimal solution through convergence evaluation;

[0013] S6. Match the target optimal solution with the knowledge base: if the match is successful, output the target optimal solution; if the match is unsuccessful, perform manual processing and update the knowledge base, and then output the result of the manual processing.

[0014] Furthermore, the step S1 specifically includes:

[0015] Through the API data interface, multiple data sources are integrated to obtain customer support and feedback data including but not limited to text and voice types. For voice-type customer support and feedback data, voice transcription is performed first, and the transcribed customer support and feedback data proceed to the next step.

[0016] Further, S21, language detection is performed on the customer support and feedback data to identify the language type of the obtained customer support and feedback data;

[0017] S22, removing irrelevant characters from the customer support and feedback data and performing unified coding processing to obtain text-normalized customer support and feedback data;

[0018] S23. Noise filtering is performed on the text-normalized customer support and feedback data to obtain input data for the large language model.

[0019] Compared with the prior art, the present invention has the following beneficial effects:

[0020] 1) Combining large language models with knowledge bases to achieve efficient processing of complex, multilingual customer feedback, and to implement multiple independent large model self-evaluation systems and circulation systems;

[0021] 2) Significantly improve the efficiency and quality of customer support through real-time analysis and automated response;

[0022] 3) The dynamic update mechanism of the knowledge base ensures the system's adaptability to new issues and trends, and improves the overall capabilities of customer support. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments:

[0024] Figure 1 It is a flow chart of the customer support and feedback analysis and processing method based on a large language model of the present invention. DETAILED DESCRIPTION

[0025] Example

[0026] like Figure 1 As shown, a customer support and feedback analysis and processing method based on a large language model includes the following specific steps:

[0027] S1. Obtain customer support and feedback data, and determine whether the customer support and feedback data is of voice type: if so, perform voice transcription on the customer support and feedback data; if not, proceed to the next step;

[0028] Specifically, multiple data sources are integrated through the API data interface to obtain customer support and feedback data including but not limited to text and voice types. For the voice type customer support and feedback data, voice transcription is first performed, and the transcribed customer support and feedback data enters the next step;

[0029] S2. Preprocessing the customer support and feedback data to obtain input data for a large language model;

[0030] Specifically, language detection is performed on customer support and feedback data to identify the language type of the customer support and feedback data; irrelevant characters are removed from the customer support and feedback data and the encoding format is unified to obtain text-normalized customer support and feedback data; noise is filtered on the text-normalized customer support and feedback data to obtain input data for the large language model to ensure the accuracy of the text data;

[0031] S3, based on multiple independent large language models, respectively, the input data is subjected to sentiment analysis, topic identification and question classification to obtain corresponding analysis results;

[0032] S4, scoring the analysis results and selecting the analysis result with the highest score as the initial optimal solution;

[0033] S5, looping and feeding back the initial optimal solution to multiple independent large language models for further analysis, and obtaining the target optimal solution through convergence evaluation;

[0034] S6. Match the target optimal solution with the knowledge base: if the match is successful, output the target optimal solution; if the match is unsuccessful, perform manual processing and update the knowledge base, and then output the result of the manual processing.

[0035] The present invention combines a large language model and a knowledge base to acquire and preprocess customer support and feedback data, perform large language model analysis, knowledge base management and update, and manual assistance and feedback analysis, thereby achieving efficient processing of complex, multi-language customer feedback, and realizing multiple independent large model self-assessment systems and circulation systems, real-time analysis and automated response, significantly improving the efficiency and quality of customer support. The dynamic update mechanism of the knowledge base ensures the system's adaptability to new issues and trends, and enhances the overall capabilities of customer support.

[0036] The embodiments described above are merely descriptions of preferred implementation modes of the present invention and are not intended to limit the scope of the present invention. Without departing from the principles and essence of the present invention, various modifications and improvements made to the technical solutions of the present invention by ordinary technicians in this field should all fall within the scope of protection determined by the claims of the present invention.

Claims

1. A customer support and feedback analysis and processing method based on a large language model, characterized in that: The specific steps include: S1. Obtain customer support and feedback data, and determine whether the customer support and feedback data is of voice type: if so, perform voice transcription on the customer support and feedback data; if not, proceed to the next step; S2. Preprocessing the customer support and feedback data to obtain input data for a large language model; S3, based on multiple independent large language models, respectively, the input data is subjected to sentiment analysis, topic identification and question classification to obtain corresponding analysis results; S4, scoring the analysis results and selecting the analysis result with the highest score as the initial optimal solution; S5, looping and feeding back the initial optimal solution to multiple independent large language models for further analysis, and obtaining the target optimal solution through convergence evaluation; S6. Match the target optimal solution with the knowledge base: if the match is successful, output the target optimal solution; If the match is unsuccessful, manual processing is performed and the knowledge base is updated, and then the results of the manual processing are output.

2. A method for analyzing and processing customer support and feedback based on a large language model according to claim 1, characterized in that: The step S1 specifically includes: Through the API data interface, multiple data sources are integrated to obtain customer support and feedback data including but not limited to text and voice types. For voice-type customer support and feedback data, voice transcription is performed first, and the transcribed customer support and feedback data proceed to the next step.

3. The method for analyzing and processing customer support and feedback based on a large language model according to claim 1, characterized in that: The step S2 specifically includes: S21. Performing language detection on the customer support and feedback data to identify the language type of the customer support and feedback data; S22, removing irrelevant characters from the customer support and feedback data and performing unified coding processing to obtain text-normalized customer support and feedback data; S23. Noise filtering is performed on the text-normalized customer support and feedback data to obtain input data for the large language model.

Images