Intelligent data list distribution system based on ChartGPT driver

The ChartGPT-driven intelligent data list allocation system solves the problems of difficult rule maintenance and low performance of the Drools rule engine in list allocation systems, achieving efficient and accurate case allocation, improving processing efficiency and customer satisfaction, and reducing labor costs.

CN118917581APending Publication Date: 2024-11-08SHANGHAI CREATE FRAME SOFTWARE CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410931208.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-12
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

In existing technologies, list allocation systems based on the Drools rule engine suffer from difficulties in rule maintenance, performance issues, and low throughput, making it difficult to efficiently and accurately allocate customer service personnel to handle cases.

Method used

The ChartGPT-driven intelligent data list allocation system automatically assigns the most suitable customer service personnel to handle cases through data collection, preprocessing, feature extraction, model training, and application modules. It utilizes the ChartGPT model and intelligent dispatch algorithm to achieve self-optimization and upgrading by combining distributed computing and real-time data stream processing.

Benefits of technology

It improved case processing efficiency and dispatch accuracy, reduced error rate, alleviated manual workload, saved human resource costs, enhanced customer satisfaction, and possessed high performance and throughput.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118917581A_ABST
    Figure CN118917581A_ABST
Patent Text Reader

Abstract

The invention relates to the field of data processing, in particular to an intelligent data list distribution system based on a ChartGPT driver. The system comprises a data collection module, a data preprocessing module, a feature extraction and data set division module, a model training module, a model evaluation module and a model application module. The data collection module is used for collecting historical order sending data; the feature extraction and data set division module is used for extracting various features and dividing the corresponding training set into subsets according to the features to obtain training data; the model training module carries out decision tree model training by using a ChartGPT model and training data, in the training process, the decision tree model learns how to generate a chart according to natural language description converted from preprocessed data, and the chart represents the matching degree of a case and an employee. According to the method, distributed computing, real-time data stream processing and other technologies are adopted, high performance and throughput are achieved, self-optimization and upgrading are continuously conducted through ChartGPT, and the method is more intelligent.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data processing, and in particular to an intelligent data list distribution system based on ChartGPT drive. Background Art

[0002] In the Internet field, case list distribution is a very important link in customer service, telemarketing, debt collection and other scenarios. The traditional random distribution method is to distribute the case list to the designated personnel after the case list is imported into the system through specific data verification and data confusion. The list distribution based on the Drools rule engine is slightly more flexible in processing. The system needs to integrate the Drools rule engine into the system and call it through the API interface. It also needs to design a series of rules to automatically distribute it to the corresponding personnel for processing based on customer information, case type, case status and other factors. The brief flow chart is as follows: Figure 1 As shown. It is necessary to design modules such as variable factors, case division rules, case division decisions, and case division records; combining one or more rules is a decision, and each decision is bound to a person. At the same time, the distribution mode can be set: average distribution, proportional distribution, etc. The decision rules are converted into Drools rule language and handed over to the rule engine for execution, and the case division result data can be obtained.

[0003] However, the above-mentioned list distribution system implemented by the Drools engine has the following defects:

[0004] Difficulty in rule maintenance. Since the rules of the Drools engine are written in the Drools language, if there are many rules or the rules change frequently, rule maintenance may become difficult. In addition, the Drools engine needs to match and execute a large number of rules, which may cause performance problems. If there are many rules or the rules are complex, it may affect the response speed and throughput of the system. Summary of the invention

[0005] The purpose of the present invention is to address the problems existing in the background technology and propose an intelligent data list distribution system driven by ChartGPT. By adopting the ChartGPT model and intelligent dispatching algorithm, the most suitable customer service personnel are automatically assigned to handle cases, thereby improving case processing efficiency and reducing error rate. It adopts distributed computing, real-time data stream processing and other technologies, has high performance and throughput, and uses ChartGPT to continuously perform self-optimization and upgrading, making it more intelligent.

[0006] The technical solution of the present invention is an intelligent data list distribution system driven by ChartGPT, including a data collection module, a data preprocessing module, a feature extraction and data set partitioning module, a model training module, a model evaluation module and a model application module; the data collection module is used to collect historical dispatch data; the data preprocessing module is used to clean and format the collected data, process missing values ​​and outliers, and convert non-numerical attributes into numerical attributes; the feature extraction and data set partitioning module is used to extract various features, and divide the corresponding training set into subsets according to the features to obtain training data; the model training module uses the ChartGPT model and training data to train the decision tree model. During the training process, the decision tree model learns how to generate a chart based on the natural language description converted from the preprocessed data, and the chart represents the degree of matching between the case and the employee. During training, the features that have the greatest impact on the dispatch result are selected as the input of the decision tree model; the model evaluation module uses a cross-validation method or test data to evaluate the performance of the trained decision tree model. If the performance is not satisfactory, the parameters of the model are adjusted, or other features are selected to retrain the model; the model application module uses the trained decision tree model for the actual dispatch process.

[0007] Preferably, the historical dispatch data includes various attributes of the case, the attributes of the employee, and the results of the dispatch. Data mining technology is used for data collection and processing. The data of each business system is collected through Kafka and processed by stream computing through Kafka-Flink.

[0008] Preferably, when converting the preprocessed data into a natural language description, it is generated through a template or automatically generated using NLP technology.

[0009] Preferably, model training ends when all features are used up or all instances in each subset belong to the same category.

[0010] Preferably, deep learning technology is used to predict the matching degree between each case and each employee, inputting the chart data generated by ChartGPT, outputting the predicted matching degree, and then assigning the case to the employee with the highest matching degree based on the matching degree.

[0011] Preferably, it also includes a visual display module for providing a visual display interface, on which customer service personnel can view case information and processing progress.

[0012] Compared with the prior art, the present invention has the following beneficial technical effects:

[0013] The present invention adopts the ChartGPT model and uses an intelligent dispatching method to automatically assign the most suitable customer service personnel to handle cases according to factors such as customer information, case type, and case status, thereby improving case handling efficiency, dispatching accuracy, reducing error rates, improving customer satisfaction, enhancing customer stickiness, saving human resources, and reducing labor costs. At the same time, the system can reduce the burden of manual dispatching, improve work efficiency, and reduce labor costs. The system uses technologies such as distributed computing and real-time data stream processing, has high performance and throughput, and uses ChartGPT to continuously optimize and upgrade itself, making it more intelligent. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 A flowchart for the existing list distribution based on the Drools rule engine;

[0015] Figure 2 It is a flow chart of the intelligent data list distribution system driven by ChartGPT according to the present invention. DETAILED DESCRIPTION

[0016] like Figure 2 As shown, this embodiment proposes an intelligent data list distribution system based on ChartGPT drive, including a data collection module, a data preprocessing module, a feature extraction and data set partitioning module, a model training module, a model evaluation module, a model application module and a visualization display module.

[0017] The data collection module is used to collect historical dispatch data, which includes various attributes of cases (such as case type, case amount, case location, etc.), employee attributes (such as employee ability, employee experience, etc.), and dispatch results (such as whether the case was successfully handled and the processing time, etc.). Data mining technology is used for data collection and processing. Data from various business systems are collected through Kafka and stream computing is performed through Kafka-Flink.

[0018] The data preprocessing module is used to clean and format the collected data, handle missing values ​​and outliers, convert non-numeric attributes into numerical attributes, and convert the preprocessed data into natural language descriptions through template generation or automatic generation using NLP technology.

[0019] The feature extraction and data set partitioning module is used to extract various features and divide the corresponding training set into subsets according to the features to obtain training data.

[0020] The model training module uses the ChartGPT model and training data to train the decision tree model. During the training process, the decision tree model learns how to generate a chart based on the natural language description converted from the preprocessed data. The chart represents the degree of match between the case and the employee. The model learns how to predict the dispatch result based on the attributes of the case and the employee. During training, the feature that has the greatest impact on the dispatch result is selected as the input of the decision tree model. First, an optimal feature is selected and the training set is divided into subsets according to this feature; then, the above process is repeated for each subset until a certain termination condition is met (for example, all features have been used up, or all instances in each subset belong to the same category). The feature that has the greatest impact on the dispatch result is selected as the input of the decision tree model. Feature selection can be performed through correlation analysis, principal component analysis, and other methods. Based on business knowledge and data analysis results, select the features that have the greatest impact on the dispatch result. For example, employee ability and experience are important features. Design an intelligent dispatching algorithm that uses deep learning technology to predict the degree of match between each case and each employee. It inputs chart data generated by ChartGPT, outputs the predicted degree of match, and then assigns the case to the employee with the highest match based on the degree of match.

[0021] The model evaluation module uses the cross-validation method to evaluate the performance of the trained decision tree model. If the performance is not satisfactory, the model parameters are adjusted, or other features are selected to retrain the model. When there is a new case that needs to be dispatched, the case features can be input into the decision tree model, and the model will output a predicted dispatch result (i.e. the customer service staff who is most suitable to handle the case). Use test data to evaluate the performance of the model. The main evaluation indicators can be the dispatch success rate, processing time, etc.

[0022] When using test data to evaluate the performance of the model, the main evaluation indicators are the order success rate and processing time. According to the evaluation results of the model, adjust the parameters and optimize the model. As time goes by, customer needs and customer service staff's capabilities may change, so the decision tree model needs to be updated regularly. Regularly collect new order data to train the model and continuously optimize the model results.

[0023] The model application module applies the trained decision tree model to the actual dispatching process.

[0024] The visual display module is used to provide a visual display interface where customer service personnel can view case information and processing progress, and can use Web front-end technology for interface design and development.

[0025] The present invention adopts the ChartGPT model and uses an intelligent dispatching method to automatically assign the most suitable customer service personnel to handle cases according to factors such as customer information, case type, and case status, thereby improving case handling efficiency, dispatching accuracy, and reducing error rates, thereby improving customer satisfaction, enhancing customer stickiness, saving human resources, and reducing labor costs. It can also be customized and optimized according to specific business needs and data characteristics, providing better support for work in the fields of customer service, telemarketing, and collection. At the same time, the system can reduce the burden of manual dispatching, improve work efficiency, and reduce labor costs. The system uses technologies such as distributed computing and real-time data stream processing, has high performance and throughput, and uses ChartGPT to continuously perform self-optimization and upgrades, making it more intelligent.

[0026] The embodiments of the present invention are described in detail above with reference to the accompanying drawings, but the present invention is not limited thereto, and various changes can be made within the knowledge scope of technicians in the relevant technical field without departing from the purpose of the present invention.

Claims

1. An intelligent data list distribution system based on ChartGPT driver, characterized in that: include: Data collection module, used to collect historical dispatch data; Data preprocessing module, which is used to clean and format the collected data, handle missing values ​​and outliers, and convert non-numeric attributes into numerical attributes; The feature extraction and data set partitioning module is used to extract various features and divide the corresponding training set into subsets according to the features to obtain training data; Model training module, using ChartGPT model and training data to train decision tree model. During the training process, the decision tree model learns how to generate a chart based on the natural language description converted from the preprocessed data. The chart represents the matching degree between the case and the employee. During the training, the features that have the greatest impact on the dispatch result are selected as the input of the decision tree model. Model evaluation module, which uses cross-validation method or test data to evaluate the performance of the trained decision tree model. If the performance is not satisfactory, the model parameters are adjusted or other features are selected to retrain the model. The model application module applies the trained decision tree model to the actual order dispatching process.

2. According to claim 1, a smart data list distribution system based on ChartGPT drive is characterized in that: Historical dispatch data includes various attributes of cases, attributes of employees, and dispatch results. Data mining technology is used for data collection and processing. Data from various business systems are collected through Kafka and processed through stream computing through Kafka-Flink.

3. According to claim 1, a ChartGPT-driven intelligent data list distribution system is characterized in that: When converting preprocessed data into natural language descriptions, it is generated through templates or automatically using NLP technology.

4. According to claim 1, a ChartGPT-driven intelligent data list distribution system is characterized in that: When all features are used up or all instances in each subset belong to the same category, model training ends.

5. According to claim 1, a ChartGPT-driven intelligent data list distribution system is characterized in that: Use deep learning technology to predict the degree of match between each case and each employee, input the chart data generated by ChartGPT, output the predicted degree of match, and then assign the case to the employee with the highest match based on the degree of match.

6. According to claim 1, a ChartGPT-driven intelligent data list distribution system is characterized in that: It also includes a visual display module, which is used to provide a visual display interface where customer service personnel can view case information and processing progress.