Omnichannel retail enterprise cross-channel return management system based on large language model

Abstract

The invention relates to the field of big data analysis, and discloses an omni-channel retail enterprise cross-channel return management system based on a big language model, which comprises a return application module, a return information integration module, an automatic return processing module, a refund and compensation module, an intelligent analysis module and a feedback module, the goods return application module is used for initiating goods return requests in different sales channels, the system automatically examines and approves the goods return requests meeting conditions, the goods return information integration module is used for integrating goods return data from different sales channels, and the automatic goods return processing module automatically generates corresponding goods return processes according to goods return types. The refund and compensation module automatically calculates and processes refund according to the state of a returned commodity and a return policy, the intelligent analysis module is used for analyzing return data and identifying a return trend and potential problems, and the feedback module is used for collecting feedback and optimizing user experience. The method has the advantages of realizing automatic goods return processing and providing personalized goods return service.

Description

Technical Field

[0001] The present invention relates to the field of big data analysis, and in particular to a cross-channel return management system for omni-channel retail enterprises based on a large language model. Background Art

[0002] In the rapid development of omni-channel retail enterprises in recent years, a diversified retail model with parallel and complementary online and offline channels has gradually been formed. Consumers shop through multiple channels such as online malls, physical stores, and mobile terminals, and shopping methods are becoming increasingly diversified and flexible. In this process, returns, as an important part of after-sales service, have gradually shown cross-channel characteristics and complexity. In particular, with the rise of e-commerce and changes in consumer shopping behavior, the difficulty of cross-channel return management has continued to increase. Consumers hope that after purchasing goods on any channel, they can conveniently return them according to their own needs, whether in physical stores, online platforms or through mobile terminals. Traditional return management models are usually limited to a single channel and cannot effectively achieve information sharing and collaborative work between different channels. The data fragmentation between channels makes the return process cumbersome and inefficient. Many existing return management systems lack automation support, especially when operating across channels, and often rely on manual processing processes, resulting in cumbersome and error-prone return operations. More manual intervention not only increases the operating costs of enterprises, but also prolongs the processing time of the return process, affecting the consumer experience. At the same time, existing systems usually lack intelligent analysis and prediction of return reasons, and cannot accurately grasp the trend and pattern of returns, thus affecting inventory management and supply chain decisions. Therefore, it is necessary to design a cross-channel return management system for omni-channel retail enterprises based on a large language model to improve return efficiency. Summary of the invention

[0003] 1. Technical issues to be resolved

[0004] In view of the shortcomings of the prior art, the present invention provides a cross-channel return management system for omni-channel retail enterprises based on a large language model, which has the advantages of being able to integrate information across channels, realize automated return processing, and provide personalized return services, thereby solving the problems in the above-mentioned background technology.

[0005] (II) Technical solution

[0006] In order to achieve the above-mentioned purpose of being able to integrate information across channels, realize automated return processing, and provide personalized return services, the present invention provides the following technical solutions: a cross-channel return management system for omni-channel retail enterprises based on a large language model, including a return application module, a return information integration module, an automated return processing module, a refund and compensation module, an intelligent analysis module, and a feedback module;

[0007] The return application module is used to initiate return requests in different sales channels, and the system automatically approves eligible return requests;

[0008] The return information integration module is used to integrate return data from different sales channels;

[0009] The automated return processing module automatically generates a corresponding return process according to the return type;

[0010] The refund and compensation module automatically calculates and processes refunds based on the status of the returned goods and the return policy;

[0011] The intelligent analysis module is used to analyze return data and identify return trends and potential problems;

[0012] The feedback module is used to collect feedback and optimize user experience.

[0013] Preferably, the return application module further includes integrating the return data of different sales channels into a unified platform through API and middleware technology, realizing real-time or scheduled data synchronization through standardized data interfaces, utilizing ETL technology to extract, clean, convert and load data, combining message queues to realize asynchronous data transmission, and processing the return data that requires immediate feedback in real time through stream data processing technology, while non-real-time data is synchronized and cleaned through scheduled tasks or batch processing.

[0014] Preferably, the return information integration module further includes connecting with different sales channels through API, automatically receiving and integrating return applications, and using OCR technology to scan return orders or vouchers from offline store channels, automatically extracting return information and importing it into the system, dynamically managing return policies through a rule engine, automatically determining return conditions, and analyzing historical data in combination with a machine learning model to predict the approval results of return requests, using NLP technology to analyze the return application text, and automatically identifying whether the reason for the return is compliant.

[0015] Preferably, the automated return processing module further includes automatically generating and executing the return processing flow according to different return conditions through the combination of the rule engine and the workflow engine. The rule engine configures the processing flow according to dynamic business needs and return policies. For the quality inspection of returned goods, computer vision technology is used for image recognition to automatically detect damage or defects in goods, and RFID tags and sensor technology are combined to perform product scanning and status recording.

[0016] Preferably, the refund and compensation module further includes defining and executing refund calculation logic through a rule engine, automatically calculating the refund amount according to the reason for return, product status and return policy, and flexibly handling different return scenarios, integrating with the payment platform, automatically initiating a refund request, and selecting a refund channel according to the payment method, automatically calculating the compensation amount according to the compensation policy, and supporting multiple compensation methods.

[0017] Preferably, the intelligent analysis module further includes analyzing massive return data through a big data processing framework using parallel computing and distributed technology, applying association rule mining, regression analysis and machine learning algorithms to identify potential patterns and trends, predicting return rates and abnormal situations, and using time series analysis to predict trends in return data;

[0018] Combining NLP technology and sentiment analysis, sentiment analysis is performed on return instructions and consumer feedback. The calculation formula is:

[0019]

[0020] Where S is the sentiment score, f(x i ) is the sentiment analysis score of the i-th feedback text, and N is the number of feedback texts.

[0021] Identify the reasons for returns and customer sentiment tendencies, segment customers through clustering algorithms, and analyze the return behavior patterns of different customer groups.

[0022] Preferably, the feedback module further includes integrating with major sales platforms, social media, email, customer service hotlines, mobile applications and other channels to uniformly collect consumer feedback from different channels, and aggregate it into the platform through an API interface, and automatically convert and analyze customer service hotline and chat feedback in combination with speech recognition and natural language processing technology to identify consumer emotions and feedback content;

[0023] Use machine learning to automatically categorize and prioritize feedback, prioritizing high-priority issues based on feedback urgency, severity, and sentiment analysis results, calculated as:

[0024] P=w 1 *Urgency+w 2 *Severity+w 3 *SentimentScore

[0025] Where P is the final feedback priority score, Urgency is the urgency score of the feedback, Severity is the severity score of the feedback, SentimentScore is the sentiment analysis score of the feedback, and w 1 , w 2 , w3 They are the weight of urgency, the weight of severity, and the weight of sentiment analysis results.

[0026] (III) Beneficial effects

[0027] Compared with the prior art, the present invention provides a cross-channel return management system for omni-channel retail enterprises based on a large language model, which has the following beneficial effects:

[0028] The present invention allows consumers to easily initiate return requests on multiple sales channels through the return application module, and the system automatically approves qualified return applications, greatly reducing the time and workload of manual review. The automated return processing module automatically generates corresponding return processes based on different return types, thereby further shortening the return processing cycle. The return information integration module ensures the consistency and accuracy of information by uniformly integrating return data from different sales channels. Based on the comprehensive analysis of these data, the system can identify return trends and potential problems in real time, help enterprises adjust product and service strategies in a timely manner, and improve customer satisfaction. In the refund and compensation module, the system automatically calculates and processes refunds and compensations according to the status of the returned goods and the return policy of the enterprise, reducing manual intervention, reducing the risk of operational errors, and improving the efficiency of the refund process and customer experience. The intelligent analysis module can discover the root causes and potential trends of returns through in-depth analysis of return data, identify high-risk goods and customers, and help enterprises take preventive measures to reduce return rates and optimize inventory management and product quality. Through the feedback module, the system collects consumer feedback during the return process and, combined with sentiment analysis technology, promptly identifies areas of customer dissatisfaction and proposes improvement plans, providing companies with improvement directions, thereby continuously optimizing the return process and user experience and enhancing customer loyalty. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the structure of the present invention. DETAILED DESCRIPTION

[0030] 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.

[0031] The present invention provides a technical solution: a cross-channel return management system for omni-channel retail enterprises based on a large language model, including a return application module, a return information integration module, an automated return processing module, a refund and compensation module, an intelligent analysis module and a feedback module;

[0032] The return application module is used to initiate return requests in different sales channels, and the system automatically approves eligible return requests;

[0033] By using API to connect with different sales channels and automatically receive return applications, the system can access return requests from different channels through a unified interface and integrate them into the return application module. For offline store channels, by scanning the return order or voucher, the return information is extracted using OCR technology and automatically imported into the system to avoid manual input errors. Consumers can initiate return requests from any channel, and the system can automatically identify and integrate data from all channels to achieve seamless return applications across all channels. Through API and OCR technology, data collection and entry can be automated, reducing manual operations and improving application efficiency.

[0034] Realize dynamic management and automated approval of return policies. Return conditions are configured through the rule engine. The system automatically determines whether the return conditions are met based on the set rules. By analyzing historical return data, the system uses machine learning models to learn the approval rules for different types of return requests. For example, the system can automatically predict whether the return should be approved based on information such as product type and return reason. For the return application text submitted by consumers, the system can use NLP technology to perform text analysis and automatically identify whether it meets the return conditions, such as whether the return reason is compliant. The system can automatically approve most eligible return requests based on the set rules and historical data, reducing the time and cost of manual approval. Through machine learning and NLP, the system can improve the accuracy of return approval, and can deal with complex return reasons and avoid errors in manual judgment.

[0035] The return information integration module is used to integrate return data from different sales channels;

[0036] Through API and middleware technology, the return data from different sales channels are integrated into a unified data platform. Real-time or scheduled synchronization is performed through standardized data interfaces to ensure that the data of each channel is updated in a timely manner. ETL technology is used to extract, clean, convert and load data from different systems into a unified database. Through ETL tools, a large amount of heterogeneous data can be processed and pre-processed to ensure data quality. Through the use of API and middleware, the system can collect and integrate return data from different sales channels in real time and stably, realize cross-platform and cross-channel data synchronization, and uniformly clean and standardize data from different channels through the ETL process to ensure data consistency and facilitate subsequent processing and analysis.

[0037] Message queue technology is used to achieve asynchronous data transmission. For some return data that need to be processed in real time, stream data processing technology is used to synchronize and process data in real time. For some return data that does not need to be processed in real time, regular synchronization and data cleaning are performed through scheduled tasks or batch processing to ensure the timeliness and integrity of the data. By combining real-time data streams and scheduled batch processing, it is ensured that different types of return data can reach the integration module in a timely manner to meet the needs of real-time processing and batch processing. Real-time data synchronization allows the system to respond quickly to return requests and integrate return information into the system, providing instant feedback to relevant departments and consumers.

[0038] The automated return processing module automatically generates the corresponding return process according to the return type;

[0039] Use rule engines, such as Droo ls and Apache Came l, to manage and execute the return process. The rule engine can dynamically generate specific processes for return processing based on different return conditions. For example, if the reason for the return is "damaged goods", a product inspection may be required; if it is "unsuitable size", a direct exchange may be processed. Use a workflow engine to design and execute the return process. The system automatically triggers the corresponding workflow steps based on the return type and ensures that each processing link is executed according to the predetermined process, such as product inspection, warehouse acceptance, logistics arrangement, and refund processing. Through the combination of the rule engine and the workflow engine, the return processing process can be automatically generated and executed according to different return situations, greatly reducing manual intervention. The rule engine allows the system to be configured according to changing business needs and return policies, ensuring that the return process can quickly respond to new business rules.

[0040] For the quality inspection of returned goods, the system can use computer vision technology to analyze the appearance of returned goods through image recognition to determine whether the goods meet the return standards. For example, use image recognition to determine whether the goods are damaged, defective or have other quality problems that affect secondary sales. When processing offline returns, use sensors and RFID tag technology to scan and inspect the goods, automatically record the status of the goods, and generate a test report. Computer vision and RFID technology can automatically identify product problems, reduce the time and error rate of manual inspections, and improve the efficiency and accuracy of product quality audits. The automated inspection process can ensure that each returned product undergoes consistent quality inspections, reduce human differences, and improve the reliability of the return process.

[0041] The refund and compensation module automatically calculates and processes refunds based on the status of the returned goods and the return policy;

[0042] Through the rule engine, the refund calculation logic is defined and executed. The system can automatically calculate the refund amount according to different return reasons, product status, and return policies. For example, if the product is intact and meets the return policy, the system will refund the full amount at the original price; if the product is defective, a certain amount of compensation may be deducted. Enterprises are allowed to define flexible refund calculation rules according to different return scenarios to ensure that the system can cope with various complex return situations. The system can automatically calculate the accurate refund amount based on factors such as product status, return reason, policy configuration, etc., reduce manual intervention, and ensure the accuracy and efficiency of refund processing. Through the configuration of the rule engine, enterprises can adjust return and refund policies according to different business needs and flexibly respond to refund requirements in different scenarios.

[0043] The system is integrated with major payment platforms and automatically initiates refund requests directly through the payment gateway. The system automatically selects the corresponding refund method based on the refund amount and payment method, and triggers the refund process. For complex refund processes or refund scenarios that require transparency, smart contracts and blockchain technology are introduced. Smart contracts can automatically execute refunds when specific conditions are met, ensuring that each refund has a basis and cannot be tampered with, improving the transparency and trust of the refund process. The integration of the payment gateway automates the refund process, allowing consumers to receive a refund within a short period of time after the return is confirmed, improving consumer satisfaction, and ensuring the transparency and non-tamperability of the refund process through blockchain and smart contracts, enhancing consumer trust, and preventing fraud and misoperation.

[0044] The system automatically calculates the amount of compensation through the rule engine and compensation policy configuration. For example, if the product returned by the consumer cannot be used due to quality problems, the system will automatically calculate the amount of compensation based on the compensation policy, which may include part of the value of the product, transportation costs, etc. Depending on the refund method and user preferences, the system can support compensation payment through multiple channels, such as refund to the original payment account, cash back, points reward or coupon issuance, etc. Through the rule engine, the system can accurately calculate the amount of compensation to be paid based on the product status, reason for return and compensation policy, reduce manual errors, improve the transparency and consistency of compensation, support multiple compensation methods to meet the needs of different consumers, such as cash back, account points, coupons and other multi-channel compensation methods.

[0045] The intelligent analysis module is used to analyze return data and identify return trends and potential problems;

[0046] Use the big data processing framework to conduct large-scale data analysis and identify potential patterns and trends in return data. Through parallel computing and distributed processing technology, analyze massive return data, extract valuable information, and use association rule mining technology to analyze the correlation between different products, return reasons, user characteristics, etc. For example, whether a certain type of product has a high return rate due to a certain defect, or the return trend of certain products in a specific time period. Through big data analysis, potential trends and anomalies in return data can be identified, providing deeper business insights. Through pattern recognition, it can be found which products or channels have a high return rate, predict possible problems and provide a basis for improvement measures.

[0047] Use statistical methods such as regression analysis to build a return prediction model and predict future return trends based on historical data. Machine learning algorithms learn and predict which products may have higher return rates based on historical return data, so as to give early warnings. Use time series analysis methods to predict trends in return data, identify trends in return rates in different time periods, and predict possible return fluctuations during holidays, promotional seasons, etc. Through machine learning and regression analysis, the system can predict which products or regions may face higher return rates, so as to take measures in advance to reduce return risks. Using time series analysis, it can predict return trends in the next few months or quarters, helping companies better adjust inventory and marketing strategies, etc.

[0048] Perform text analysis on the reasons for returns, use NLP technology to process the return instructions submitted by consumers, and identify common reasons for returns. Sentiment analysis can help the system extract sentiment tendencies from consumer reviews and feedback, analyze customer dissatisfaction with the product, and thus reveal potential quality problems. Use clustering algorithms to segment customers and analyze the return behavior patterns of different customer groups. For example, customers who frequently return products may have different reasons and patterns for returns than those who occasionally return products, and the system can take targeted measures. NLP and sentiment analysis can help the system accurately identify the specific reasons for consumer returns, providing reference for subsequent product improvements or sales strategy optimization. Customer segmentation and behavioral analysis help companies identify the return patterns of different customer groups and provide data support for personalized marketing and customer management.

[0049] The feedback module is used to collect feedback and optimize the user experience.

[0050] Through the integration with major sales platforms, social media, email, customer service hotline, mobile applications and other channels, feedback information provided by consumers in various ways is collected. Feedback from different channels is aggregated into the system through the API interface to ensure feedback collection from all channels. After the return process is completed, the system sends a feedback request to the consumer through email or pop-up window. Through a simple and easy-to-use online feedback form, consumers' ratings and opinions on return process, product quality, customer service experience, etc. are collected. For feedback from customer service calls and customer service chats, the system can combine voice recognition and natural language processing technology to convert voice into text, and perform sentiment analysis and keyword extraction to automatically identify consumer emotions and feedback content. By integrating multi-channel feedback, it is ensured that consumer feedback can be seamlessly entered into the system. Whether it is in the pre-sales, sales, or after-sales stage, it can be collected in a timely manner. Feedback information from different channels can be unified and integrated to avoid information fragmentation, helping companies to fully grasp the real thoughts and problems of consumers.

[0051] Feedback is automatically classified through machine learning, divided into different categories, and sorted according to the urgency and severity of the feedback. For example, if the feedback mentions "damaged goods" or "refund not received", the system will automatically mark it as a high-priority issue. Combined with multi-dimensional data such as customer historical return records, purchase amounts, and sentiment analysis results of feedback content, the system automatically evaluates the priority of each feedback and prioritizes high-priority feedback to ensure that the most urgent and important issues can be resolved in a timely manner. Feedback can be automatically classified and sorted according to type and priority, helping companies save manual processing time and improve the efficiency of feedback processing. Through priority sorting, companies can focus on solving the most urgent feedback issues, improve customer experience, and avoid negative emotions caused by long-term unresolved issues.

[0052] The priority quantization process is as follows:

[0053] Urgency: 5, system crash, extremely urgent;

[0054] Severity: 4, affecting transactions of some users;

[0055] Sentiment analysis: -0.8, user sentiment is very negative.

[0056] The weight of each factor is: 1 =0.5, w 2 =0.3, w 3 =0.2;

[0057] The priority score is calculated according to the formula:

[0058] P=(0.5*5)+(0.3*4)+(0.2*(0.8))=3.54

[0059] Set a priority score range to determine the order in which feedback is addressed:

[0060] High priority: P≥4.0, urgent issues that seriously affect a large number of users;

[0061] Medium priority: 2.0≤P<4.0, issues with less impact but need to be dealt with as soon as possible;

[0062] Low priority: P<2.0, advisory issues or issues with little impact on users.

[0063] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0064] 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 cross-channel return management system for omni-channel retail enterprises based on a large language model, characterized by: It includes return application module, return information integration module, automated return processing module, refund and compensation module, intelligent analysis module and feedback module; The return application module is used to initiate return requests in different sales channels, and the system automatically approves eligible return requests; The return information integration module is used to integrate return data from different sales channels; The automated return processing module automatically generates a corresponding return process according to the return type; The refund and compensation module automatically calculates and processes refunds based on the status of the returned goods and the return policy; The intelligent analysis module is used to analyze return data and identify return trends and potential problems; The feedback module is used to collect feedback and optimize user experience.

2. The cross-channel return management system for omni-channel retail enterprises based on a large language model according to claim 1, characterized in that: The return application module further includes integrating return data from different sales channels into a unified platform through API and middleware technology, and realizing real-time or scheduled data synchronization through standardized data interfaces, using ETL technology to extract, clean, convert and load data, combining message queues to realize asynchronous data transmission, and using stream data processing technology to process return data that requires immediate feedback in real time, while non-real-time data is synchronized and cleaned through scheduled tasks or batch processing.

3. The cross-channel return management system for omni-channel retail enterprises based on a large language model according to claim 1, characterized in that: The return information integration module further includes connecting with different sales channels through API, automatically receiving and integrating return applications, and using OCR technology to scan return orders or vouchers from offline store channels, automatically extracting return information and importing it into the system, dynamically managing return policies through a rule engine, automatically determining return conditions, and combining machine learning models to analyze historical data to predict the approval results of return requests, using NLP technology to analyze the return application text, and automatically identifying whether the reason for the return is compliant.

4. The cross-channel return management system for omni-channel retail enterprises based on a large language model according to claim 1, characterized in that: The automated return processing module further includes automatically generating and executing the return processing flow according to different return conditions through the combination of the rule engine and the workflow engine. The rule engine configures the processing flow according to dynamic business needs and return policies. For the quality inspection of returned goods, computer vision technology is used for image recognition to automatically detect damage or defects in goods, and RFID tags and sensor technology are combined to scan goods and record their status.

5. The cross-channel return management system for omni-channel retail enterprises based on a large language model according to claim 1, characterized in that: The refund and compensation module further includes defining and executing refund calculation logic through a rule engine, automatically calculating the refund amount based on the reason for return, product status and return policy, and flexibly handling different return scenarios, integrating with the payment platform, automatically initiating a refund request, and selecting a refund channel based on the payment method, automatically calculating the compensation amount based on the compensation policy, and supporting multiple compensation methods.

6. The cross-channel return management system for omni-channel retail enterprises based on a large language model according to claim 1, characterized in that: The intelligent analysis module further includes analyzing massive return data through a big data processing framework using parallel computing and distributed technology, applying association rule mining, regression analysis and machine learning algorithms to identify potential patterns and trends, predict return rates and anomalies, and use time series analysis to predict trends in return data; Combining NLP technology and sentiment analysis, sentiment analysis is performed on return instructions and consumer feedback. The calculation formula is: Where S is the sentiment score, f(x i ) is the sentiment analysis score of the i-th feedback text, and N is the number of feedback texts. Identify the reasons for returns and customer sentiment tendencies, segment customers through clustering algorithms, and analyze the return behavior patterns of different customer groups.

7. The cross-channel return management system for omni-channel retail enterprises based on a large language model according to claim 1, characterized in that: The feedback module further includes integrating with major sales platforms, social media, email, customer service hotlines, mobile applications and other channels to uniformly collect consumer feedback from different channels, aggregate it into the platform through an API interface, and automatically convert and analyze customer service hotline and chat feedback in combination with speech recognition and natural language processing technology to identify consumer emotions and feedback content; Use machine learning to automatically categorize and prioritize feedback, prioritizing high-priority issues based on feedback urgency, severity, and sentiment analysis results, calculated as: P=w1*Urgency+w2*Severity+w3*SentimentScore Where P is the final feedback priority score, Urgency is the urgency score of the feedback, Severity is the severity score of the feedback, SentimentScore is the sentiment analysis score of the feedback, and w1, w2, and w3 are the weights of urgency, severity, and sentiment analysis results, respectively.

Images