Intelligent e-commerce customer service after-sales processing method, system and device and medium

Abstract

The invention provides an intelligent e-commerce customer service after-sales processing method, system and device and a medium, and relates to the technical field of intelligent customer service, and the method comprises the steps: obtaining text data, image data, dialogue historical data and a compensation mechanism associated with an after-sales question; performing feature processing on the text data and the image data to obtain text features and image features; according to the dialogue historical data, obtaining an importance proportion between the text features and the image features; performing interactive feature enhancement processing and fusion processing on the text features and the image features in combination with the importance proportion to obtain a problem classification result and a problem classification confidence coefficient; according to the question classification result, the question classification confidence coefficient and the compensation mechanism, generating processing information corresponding to the after-sales question; and processing the after-sales problem according to the problem classification result corresponding to the after-sales problem, the problem classification confidence and the processing information. By comprehensively analyzing the text and the image, the after-sales problem classification accuracy is improved, so that the consumer experience is optimized.

Description

Technical Field

[0001] This application relates to the technical field of intelligent customer service, and in particular, to an intelligent e-commerce customer service after-sales processing method, system, device, and medium. Background Art

[0002] With the rapid development of e-commerce, consumers' demand for after-sales service is increasing continuously. The existing intelligent customer service after-sales processing technology mainly relies on text processing and simple question-and-answer modes, and it is difficult to handle complex after-sales problems. Especially when consumers upload pictures to show damaged, stained, or missing parts of the goods, the existing technology often cannot accurately analyze the picture content and cannot provide effective solutions. In addition, most of the existing intelligent customer service after-sales processing technologies cannot effectively integrate picture and text information, resulting in low accuracy of problem recognition, increasing the processing complexity and time consumption, and reducing the consumer experience. Summary of the Invention

[0003] This application provides an intelligent e-commerce customer service after-sales processing method, system, device, and medium to solve one or more technical problems existing in the prior art and at least provide beneficial choices or create conditions.

[0004] On the one hand, this application provides an intelligent e-commerce customer service after-sales processing method, including the following steps: Obtain text data, image data, conversation history data, and compensation mechanisms associated with the after-sales problem; Perform feature processing on the text data and the image data to obtain text features and image features; According to the conversation history data, obtain the importance ratio between the text features and the image features; Combine the importance ratio to perform interactive feature enhancement processing and fusion processing on the text features and the image features to obtain a problem classification result and a problem classification confidence level; Generate processing information corresponding to the after-sales problem according to the problem classification result, the problem classification confidence level, and the compensation mechanism; Process the after-sales problem according to the problem classification result, the problem classification confidence level, and the processing information corresponding to the after-sales problem.

[0005] Further, the performing feature processing on the text data and the image data to obtain text features and image features includes: Use a preset text processing model to perform feature processing on the text data to obtain the text features; Use a preset image processing model to perform feature processing on the image data to obtain the image features.

[0006] Further, obtaining the importance ratio between the text feature and the image feature according to the conversation history data includes: Performing feature processing on the conversation history data by using a preset text processing model to obtain a conversation history vector; Obtaining the importance ratio between the text feature and the image feature according to the conversation history vector.

[0007] Further, combining the importance ratio to perform interactive feature enhancement processing and fusion processing on the text feature and the image feature to obtain a problem classification result and a problem classification confidence includes: Combining the importance ratio to perform linear transformation on the text feature and the image feature to obtain a weighted text feature and a weighted image feature in the same dimension; Using a deep learning model to perform interactive feature enhancement processing and fusion processing on the weighted text feature and the weighted image feature to obtain a problem classification result and a problem classification confidence.

[0008] Further, using a deep learning model to perform interactive feature enhancement processing and fusion processing on the weighted text feature and the weighted image feature to obtain a problem classification result and a problem classification confidence includes: According to the weighted text feature, using the self-attention mechanism of the deep learning model to obtain a self-enhanced text feature; According to the weighted image feature, using the self-attention mechanism of the deep learning model to obtain a self-enhanced image feature; According to the weighted text feature and the weighted image feature, using the cross-modal attention mechanism of the deep learning model to obtain a cross-modal enhanced text feature and a cross-modal enhanced image feature; According to the self-enhanced text feature, the self-enhanced image feature, the cross-modal enhanced text feature, and the cross-modal enhanced image feature, using a multi-layer perceptron model to obtain the problem classification result and the problem classification confidence.

[0009] Further, processing the after-sales problem according to the problem classification result, the problem classification confidence, and the processing information corresponding to the after-sales problem includes: If the problem classification confidence corresponding to the after-sales problem exceeds a preset confidence threshold, automatically processing the after-sales problem according to the problem classification result and the processing information corresponding to the after-sales problem; Otherwise, sending the problem classification result and the processing information corresponding to the after-sales problem to a human customer service for the human customer service to process the after-sales problem.

[0010] On the other hand, the present application provides an intelligent e-commerce customer service after-sales processing system, including a data acquisition module, a feature processing module, a weight adjustment module, a feature interaction module, and an after-sales processing module; The data acquisition module is used to acquire text data, image data, conversation history data, and compensation mechanisms associated with after-sales problems; The feature processing module is used to perform feature processing on the text data and the image data to obtain text features and image features; The weight adjustment module is used to obtain the importance ratio between the text features and the image features according to the conversation history data; The feature interaction module is used to perform interactive feature enhancement processing and fusion processing on the text features and the image features in combination with the importance ratio to obtain a problem classification result and a problem classification confidence level; The after-sales processing module is used to generate processing information corresponding to the after-sales problem according to the problem classification result, the problem classification confidence level, and the compensation mechanism; and process the after-sales problem according to the problem classification result, the problem classification confidence level, and the processing information corresponding to the after-sales problem.

[0011] Further, the after-sales processing module includes an automatic processing unit and a manual processing unit; The automatic processing unit is used to automatically process the after-sales problem according to the problem classification result and the processing information corresponding to the after-sales problem when the problem classification confidence level corresponding to the after-sales problem exceeds a preset confidence level threshold; The manual processing unit is used to send the problem classification result and the processing information corresponding to the after-sales problem to a human customer service when the problem classification confidence level corresponding to the after-sales problem does not exceed the preset confidence level threshold, so as to facilitate the human customer service to process the after-sales problem.

[0012] On the other hand, the present application provides an intelligent e-commerce customer service after-sales processing device, including a processor and a memory, where the memory is used to store a program; when the program is executed by the processor, the processor implements the foregoing intelligent e-commerce customer service after-sales processing method.

[0013] On the other hand, the present application provides a computer-readable storage medium, in which a program executable by a processor is stored, and the program executable by the processor is used to implement the foregoing intelligent e-commerce customer service after-sales processing method when executed by the processor.

[0014] The beneficial effects of the present application are as follows: The present application provides an intelligent e-commerce customer service after-sales processing method, including: obtaining text data, image data, conversation history data, and compensation mechanisms associated with after-sales problems; performing feature processing on the text data and image data to obtain text features and image features; obtaining the importance ratio between the text features and image features according to the conversation history data; combining the importance ratio to perform interactive feature enhancement processing and fusion processing on the text features and image features to obtain a problem classification result and a problem classification confidence level; generating processing information corresponding to the after-sales problem according to the problem classification result, the problem classification confidence level, and the compensation mechanism; and processing the after-sales problem according to the problem classification result, the problem classification confidence level, and the processing information corresponding to the after-sales problem. By comprehensively analyzing text, image, and conversation history data, the present application improves the accuracy of after-sales problem classification, thereby optimizing the consumer experience. The present application also provides corresponding devices, systems, and media. The beneficial effects of the devices, systems, and media are similar to those of the method and will not be repeated here.

[0015] Other features and advantages of the present application will be described in the following specification, and, in part, will be obvious from the specification, or will be understood by implementing the present application. The objectives and other advantages of the present application can be realized and obtained by the structures specifically pointed out in the specification, the claims, and the drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] The drawings are used to provide a further understanding of the technical solutions of the present invention, and constitute a part of the specification. They are used together with the embodiments of the present invention to explain the technical solutions of the present invention, and do not constitute a limitation to the technical solutions of the present invention.

[0017] Figure 1 is a flowchart of an intelligent e-commerce customer service after-sales processing method provided by the present application; Figure 2 is a schematic diagram of determining a problem classification result and a problem classification confidence level provided by the present application; Figure 3 is a structural diagram of an intelligent e-commerce customer service after-sales processing system provided by the present application; Figure 4 is a structural diagram of an intelligent e-commerce customer service after-sales processing device provided by the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0018] In order to make the objectives, technical solutions, and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0019] The present application will be further described below in conjunction with the accompanying drawings of the specification and specific embodiments. The described embodiments should not be construed as limiting the present application. All other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the scope of protection of the present application.

[0020] In the following description, reference is made to "some embodiments", which describe a subset of all possible embodiments. However, it can be understood that "some embodiments" can be the same subset or different subsets of all possible embodiments, and can be combined with each other without conflict.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the art belonging to the technical field of the present application. The terms used herein are only for the purpose of describing the embodiments of the present application and are not intended to limit the present application.

[0022] The booming development of e-commerce has increased the demand for after-sales service and also raised consumers' requirements for shopping experience. Traditional manual customer service is inefficient in handling after-sales issues. Facing repetitive inquiries, it is difficult to respond quickly, consuming a large amount of time and energy, and is prone to causing customer dissatisfaction due to errors or omissions, especially when dealing with routine requests such as refunds and exchanges, the process is cumbersome. Although existing intelligent customer service systems have reduced the workload to a certain extent, their rigid responses based on the Q&A mode are limited in helping with complex problems, and they lack the ability to integrate and analyze image and text information. When consumers upload pictures of product problems along with descriptions, most existing intelligent customer service systems are unable to effectively associate these two types of information, resulting in an inability to accurately understand customer needs, increasing the complexity and time cost of communication, and making the consumer experience poor.

[0023] In view of the problems and deficiencies in the related art, the embodiments of the present application provide an intelligent e-commerce customer service after-sales processing method, system, device and medium. The embodiments of the present application obtain text data, image data, conversation history data and compensation mechanisms associated with after-sales problems, comprehensively analyze text and image features, and perform interactive feature enhancement and fusion processing in combination with the importance ratio of the conversation history, thereby improving the accuracy and confidence of problem classification. For problems with high confidence, the system automatically processes and generates detailed processing information to be sent to the customer; while for problems with low confidence, preliminary processing suggestions are sent to the manual customer service for further review. This method not only improves the accuracy of problem recognition, enhances the processing efficiency, optimizes the user experience, but also reduces the operation cost, effectively solves the deficiencies of existing intelligent customer service systems in multi-modal information integration analysis and complex problem processing, and thus improves the quality and efficiency of e-commerce after-sales service.

[0024] First, the implementation steps of the intelligent e-commerce customer service after-sales processing method provided by the embodiments of the present application will be elaborated in detail with reference to the accompanying drawings.

[0025] The intelligent e-commerce customer service after-sales processing method proposed by the embodiments of the present application can be applied to a terminal, or to a server, or can also be software running on a terminal or a server, etc. The terminal can be a tablet computer, a notebook computer, a desktop computer, etc., but is not limited thereto. The server can be an independent physical server, or a server cluster or a distributed system composed of multiple physical servers, or can also be a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks, and big data and artificial intelligence platforms.

[0026] Refer to Figure 1 , the implementation process of the intelligent e-commerce customer service after-sales processing method provided by the embodiments of the present application can include but is not limited to the following steps.

[0027] Step 101, obtain text data, image data, conversation history data, and compensation mechanisms associated with after-sales problems.

[0028] In step 101, collect all necessary information related to after-sales problems. By obtaining text data (such as customer descriptions), image data (such as photos of product problems), conversation history data (such as previous interaction records), and compensation mechanisms (such as company policies), the system can obtain a complete context view, ensuring that subsequent processing is based on a sufficient information basis. This not only helps improve the accuracy of problem classification but also provides a basis for generating reasonable processing information, thereby enhancing the overall service quality.

[0029] Step 102, perform feature processing on the text data and image data to obtain text features and image features.

[0030] In step 102, perform feature processing on the text data and image data to obtain text features and image features. Among them, text features can capture the customer's description intention and after-sales needs, while image features can reveal the specific condition of the product (such as the degree of damage, the location of stains, etc.). By performing feature processing on these two modalities of data, the system can obtain a more structured and interpretable information representation, laying a solid foundation for subsequent feature fusion and problem classification.

[0031] Step 103, obtain the importance ratio between the text features and the image features according to the conversation history data.

[0032] In step 103, by analyzing the dialogue history data, the relative importance ratio between the text features and the image features is calculated. The dialogue history data provides rich context information, which can help the system understand which information is more crucial for the current problem. For example, in some cases, the image may more intuitively reflect the essence of the problem than the text description; while in other cases, detailed text explanations may be the key to solving the problem. Determining the importance ratio helps to more accurately weigh the impacts of different modal information in the subsequent steps, improving the accuracy of problem classification.

[0033] Step 104, combining the importance ratio, performs interactive feature enhancement processing and fusion processing on the text features and the image features to obtain the problem classification result and the problem classification confidence.

[0034] In step 104, combining the importance ratio obtained in the previous step, interactive feature enhancement and fusion processing are performed on the text features and the image features. This processing method can not only strengthen the feature representations of their respective modalities, but also establish the association between the two, forming a richer and more comprehensive problem representation. Finally, the system outputs the problem classification result and its confidence score, ensuring that subsequent processing decisions are based on reliable data support, improving the accuracy and credibility of automated processing.

[0035] Step 105, according to the problem classification result, the problem classification confidence, and the compensation mechanism, generates the processing information corresponding to the after-sales problem.

[0036] In step 105, based on the problem classification result and the confidence score, combined with the company's compensation mechanism, detailed after-sales processing information is automatically generated. This includes specific processing information (such as refund, exchange, repair, etc.), operation guidelines (such as return process, preparing product packaging, etc.), time frame (such as estimated processing time), and other additional information (such as the way to contact customer service). The generated processing information not only complies with the company's policies and regulations, but also meets the needs of customers, ensuring that the processing process is transparent and efficient, enhancing the customer's trust and satisfaction.

[0037] Step 106, according to the problem classification result, the problem classification confidence, and the processing information corresponding to the after-sales problem, processes the after-sales problem.

[0038] In step 106, according to the problem classification result, the problem classification confidence, and the processing information corresponding to the after-sales problem, the generated processing information is applied to the actual problem processing. This method not only improves the processing efficiency, but also reduces the customer waiting time, thereby enhancing the quality of after-sales service and the customer experience.

[0039] Through the above steps, the intelligent e-commerce customer service after-sales processing method provided by the embodiments of the present application realizes the full-process automation and intelligence from data acquisition to problem solving, effectively improves the accuracy and efficiency of after-sales service, optimizes the user experience, and reduces the operation cost.

[0040] In some embodiments of the present application, referring to Figure 2 , in step 102, the implementation process of performing feature processing on the text data and the image data to obtain text features and image features may include but is not limited to the following steps.

[0041] Step 201, use a preset text processing model to perform feature processing on the text data to obtain text features.

[0042] In step 201, a preset text processing model is used to extract features from the text data provided by the customer. Through deep learning technology, the semantic information and context relationships in the text can be captured, and high-dimensional text feature vectors can be generated. The text features not only contain information at the lexical level but also reflect the overall semantic structure of sentences or paragraphs, enabling the system to more accurately understand the customer's problem description and providing strong support for subsequent problem classification.

[0043] In some embodiments of the present application, the text processing model includes the RoBERTa model.

[0044] The RoBERTa model is an improved version of the BERT model. Through a series of optimization measures, such as a larger training data set, a dynamic masking strategy, etc., the performance and robustness of the model are effectively improved. Through large-scale pre-training, the RoBERTa model can capture the deep semantic information and context relationships in the text. When processing after-sales problems, the system uses the RoBERTa model to extract features from the text data provided by the customer and generate high-dimensional text feature vectors. These features not only contain information at the lexical level but also reflect the overall semantic structure of sentences or paragraphs, enabling the system to more accurately understand the customer's description intention and sentiment tendency, thereby improving the accuracy of problem classification.

[0045] Moreover, the RoBERTa model has powerful natural language processing capabilities, especially excellent performance in processing complex and diverse text data. Through the text features extracted by the RoBERTa model, the system can more accurately perform problem classification in subsequent steps and generate reliable problem classification confidence scores. This high-precision classification result helps the automated processing module quickly respond to simple problems, while ensuring that complex problems can be correctly identified and transferred to the manual customer service for review, improving the overall processing reliability and efficiency.

[0046] In some embodiments of the present application, the implementation process of using a preset RoBERTa model to process text data to obtain text features may include but is not limited to the following steps.

[0047] First, perform necessary preprocessing operations on the input text data, such as word segmentation, removing stop words, converting to lowercase, etc., to ensure that the text sequence input into the RoBERTa model is clean and formatted. For example, use the Jieba word segmentation tool to segment the text into words or subwords.

[0048] Secondly, the RoBERTa model receives the preprocessed text sequence as input. Each word or subword (token) is mapped to an embedding vector. Suppose the text sequence contains tokens, then the input embedding can be represented as , where represents the embedding vector of the -th token, and the embedding dimension is . These embedding vectors integrate lexical, positional, and segment information, enabling the model to understand the context of the text.

[0049] Furthermore, the RoBERTa model processes the input embedding through multiple layers of Transformer encoders. Each layer of the encoder includes a self-attention mechanism for capturing the relationships between different parts within the text. The self-attention mechanism satisfies the following formula (1): (1); In formula (1), represents the query matrix, ; represents the key matrix, ; represents the value matrix, ; , and are learnable weight matrices, and is the dimension of the key matrix , used to scale the dot product to stabilize the gradient. Through the self-attention mechanism, the model can capture long-range dependencies and context information in the text.

[0050] Then, each layer of the Transformer encoder in the RoBERTa model also includes a layer normalization function and a feed-forward neural network layer. The layer normalization function helps to accelerate training and improve the model performance, while the feed-forward neural network layer further enhances the feature expression ability. After passing through multiple layers of encoders, the finally output hidden state is represented as , where represents The corresponding hidden state contains rich semantic information.

[0051] Finally, aggregate and output the text features. To obtain the feature representation of the entire text, the system usually performs some form of aggregation on the hidden states of all tokens. Common methods include taking the hidden state of the last token (i.e., the [CLS] token), mean pooling, or max pooling. Assuming mean pooling is selected, the text features Satisfy the following formula (2): (2).

[0052] Through the above steps, the embodiment of the present application uses a preset RoBERTa model to process the text data, generating text features that can accurately capture the text semantics and context relationships, providing a solid foundation for subsequent interactive feature enhancement and fusion processing. This method not only improves the accuracy of problem classification but also enhances the system's understanding and processing capabilities.

[0053] Step 202: Use a preset image processing model to process the image data to obtain image features.

[0054] In step 202, a preset image processing model is used to extract features from the product images uploaded by consumers. These models can automatically learn the key visual features in the images, such as shape, color, texture, etc., and convert them into low-dimensional feature vectors. The image features can help the system intuitively identify the status of the product (such as damage, stains, missing parts, etc.) and combine with the text features to form a more comprehensive problem representation. In this way, the system can make more accurate judgments based on multimodal data, effectively improving the accuracy and reliability of problem classification.

[0055] In some embodiments of the present application, the image processing model includes a ResNet model and combines a Mask R-CNN model on the basis of the ResNet model. The ResNet model ensures that the system can accurately capture the key information in the product images through its powerful image feature extraction ability; while the Mask R-CNN model further enhances the object detection and segmentation accuracy, enabling the system to more carefully understand the specific situation of the problem.

[0056] Specifically, by introducing residual connections, the ResNet model solves the problem of vanishing gradients in deep neural networks, enabling the network to be deeper and more effectively trained. The ResNet model has strong image feature extraction capabilities and can capture low-level features in images (such as edges and textures) and high-level semantic information (such as object shapes and color distributions, etc.). In the embodiments of the present application, the ResNet model is used to perform preliminary feature extraction on the uploaded product images to generate high-dimensional image feature vectors.

[0057] In addition, the Mask R-CNN model is an improvement based on the Faster R-CNN model. It can not only perform object detection (identifying objects in the image and their positions), but also generate pixel-level segmentation masks (masks) for each detected object simultaneously. This ability is particularly important for handling complex after-sales problems. For example, when a customer uploads a photo of a damaged product, the Mask R-CNN model can help the system accurately locate and segment the damaged area, thus better understanding the specific situation of the problem.

[0058] In some embodiments of the present application, the implementation process of using the ResNet model combined with the Mask R-CNN model to process image data to obtain image features may include but is not limited to the following steps.

[0059] First, preprocess the image data to obtain the input image. Specifically, perform operations such as resizing, cropping, and normalizing the image data to obtain an input image that meets the model input requirements.

[0060] Second, perform feature extraction on the input image. Use the ResNet model to extract high-level semantic features from the input image. The ResNet model solves the problem of vanishing gradients in deep neural networks by introducing residual blocks and allows for the construction of deeper network structures. After passing through the ResNet model, the output of the input image is a feature map rich in semantic information.

[0061] Furthermore, use the region proposal network in the Mask R-CNN model to generate a series of candidate regions based on the feature map generated by ResNet. These candidate regions are the basis for subsequent processing and define the positions where objects of interest may be contained. To obtain a fixed-size feature map from each candidate region, the Mask R-CNN model performs candidate region alignment, retaining more spatial information.

[0062] Then, for each candidate region, the Mask R-CNN model parallelly performs three tasks: predicting the bounding box position, classifying the category, and predicting the mask. Among them, the classification task is used to determine the category of the object within the region, the task of predicting the bounding box position is used to fine-tune the position of the candidate region to more accurately match the position of the real object, and the mask prediction task is used to generate a binary mask that matches the size of the candidate region to represent the precise contour of the object.

[0063] Finally, combine the classification results, bounding boxes, and masks output by the Mask R-CNN model to obtain the precise positioning and classification information of each instance in the image as image features.

[0064] In some embodiments of the present application, referring to Figure 2 , in step 103, the implementation process of obtaining the importance ratio between the text feature and the image feature according to the conversation history data may include but is not limited to the following steps.

[0065] Step 301, use a preset text processing model to perform feature processing on the conversation history data to obtain a conversation history vector.

[0066] In step 301, use a preset text processing model to extract features from the previous conversation history data to generate a conversation history vector. These models can capture semantic information, sentiment tendencies, and context relationships in the conversation and generate high-dimensional conversation history feature vectors. The conversation history vector not only contains information at the lexical level but also reflects the interaction pattern and key topics between the customer and the customer service during the entire conversation process. This helps the system better understand the background of the current problem and provides a basis for determining the importance ratio of text and image features subsequently.

[0067] Step 302, obtain the importance ratio between the text feature and the image feature according to the conversation history vector.

[0068] In step 302, based on the generated conversation history vector, evaluate the relative importance of the text feature and the image feature in the current problem. Specifically, calculate the importance parameter corresponding to the text feature and the importance parameter corresponding to the image feature respectively, and then obtain the importance ratio between the text feature and the image feature. By analyzing the keywords, high-frequency topics, and customer concerns in the conversation history, the system can identify which information is more critical for solving the problem. For example, if the customer mentions the specific damage situation of the product multiple times and uploads relevant pictures, the image feature may be more important than the text description; conversely, if the customer details the service experience or process problem, the text feature may be more important. In this way, the system can dynamically adjust the weights of the text and image features to ensure that the most relevant information is fully utilized in subsequent processing, improving the accuracy of problem classification and the effectiveness of information processing.

[0069] In some embodiments of the present application, the importance parameter corresponding to the text feature satisfies the following formula (3): (3); In formula (3), represents the importance parameter corresponding to the text feature, represents the dialogue history vector, represents the random weight matrix corresponding to the text feature, represents the random weight matrix corresponding to the image feature; the importance parameter corresponding to the image feature is represented as: .

[0070] In some embodiments of the present application, referring to Figure 2 , in step 104, in combination with the importance ratio, the process of performing interactive feature enhancement processing and fusion processing on the text feature and the image feature to obtain the problem classification result and the problem classification confidence can include but is not limited to the following steps.

[0071] Step 401, in combination with the importance ratio, perform a linear transformation on the text feature and the image feature to obtain weighted text features and weighted image features of the same dimension.

[0072] In step 401, according to the previously calculated relative importance ratio between the text feature and the image feature, perform a linear transformation on these two features to make them have the same dimension. Specifically, the system multiplies the text feature and the image feature by the corresponding weight coefficients to obtain the weighted text feature and the weighted image feature. This linear transformation ensures that data of different modalities can be effectively compared and fused in the same space, providing a unified basis for subsequent interactive feature enhancement processing. In this way, the system can more accurately capture the importance degree of each modality information, thereby improving the accuracy of the overall feature representation.

[0073] Step 402, use a deep learning model to perform interactive feature enhancement processing and fusion processing on the weighted text feature and the weighted image feature to obtain the problem classification result and the problem classification confidence.

[0074] In step 402, use a deep learning model to perform interactive feature enhancement processing and fusion processing on the weighted text feature and the image feature, and output the problem classification result and its corresponding confidence score. This method not only improves the accuracy of problem classification, but also provides a reliable confidence evaluation to help the system decide whether manual review is required. In this way, the system can make a more accurate judgment based on multi-modal data, effectively improving the quality and efficiency of after-sales service.

[0075] Through the above steps, the embodiments of the present application achieve efficient interactive enhancement and fusion processing of text features and image features, ensuring that the system can comprehensively understand after-sales problems from different perspectives and generate accurate problem classification results and confidence scores. This not only enhances the decision-making ability of the system but also provides a solid foundation for subsequent automated processing or manual review, ensuring that customer problems are solved in a timely and effective manner.

[0076] In some embodiments of the present application, referring to Figure 2 , in step 402, the implementation process of using a deep learning model to perform interactive feature enhancement processing and fusion processing on weighted text features and weighted image features to obtain a problem classification result and a problem classification confidence level may include but is not limited to the following steps.

[0077] Step 501, according to the weighted text features, use the self-attention mechanism of the deep learning model to obtain self-enhanced text features.

[0078] In step 501, the self-attention mechanism in the deep learning model is used to process the weighted text features. The self-attention mechanism can capture the relationships between different parts within the text, emphasizing those words or sentence fragments that are more important for the current problem. In this way, the system can generate richer and more representative self-enhanced text features. These features not only contain the original text information but also strengthen the understanding of key content, thereby improving the accuracy of problem classification.

[0079] Step 502, according to the weighted image features, use the self-attention mechanism of the deep learning model to obtain self-enhanced image features.

[0080] In step 502, the self-attention mechanism of the deep learning model is also used to process the weighted image features. The self-attention mechanism can capture the correlations between different regions within the image, emphasizing those visual elements (such as damaged parts, stain positions, etc.) that are more important for the current problem. In this way, the system can generate more detailed and representative self-enhanced image features. These features not only retain the key visual information of the original image but also enhance the understanding of problem-related regions, further improving the accuracy of problem classification.

[0081] Step 503, according to the weighted text features and the weighted image features, use the cross-modal attention mechanism of the deep learning model to obtain cross-modal enhanced text features and cross-modal enhanced image features.

[0082] In step 503, the cross-modal attention mechanism of the deep learning model is used to combine the weighted text features and the weighted image features. The cross-modal attention mechanism can establish associations between the two modalities, enabling the system to understand the mutual influence between text and images. For example, if the text describes the damage situation in a specific area and the image exactly shows the details of that area, the cross-modal attention mechanism can help the system better integrate these two pieces of information. In this way, the cross-modal enhanced text features and cross-modal enhanced image features generated by the system not only more comprehensively reflect the overall picture of the problem but also improve the accuracy and confidence of problem classification.

[0083] Step 504: Based on the self-enhanced text features, self-enhanced image features, cross-modal enhanced text features, and cross-modal enhanced image features, use a multi-layer perceptron model to obtain the problem classification result and the problem classification confidence.

[0084] In step 504, the above four enhanced features (self-enhanced text features, self-enhanced image features, cross-modal enhanced text features, and cross-modal enhanced image features) are input into the multi-layer perceptron model for the final problem classification. Through multiple layers of non-linear transformations, the multi-layer perceptron model can effectively extract the information that best characterizes the problem from the high-dimensional feature space and output the problem classification result and its corresponding confidence score. This method not only improves the accuracy of problem classification but also provides a reliable confidence assessment to help the system decide whether manual review is required. In this way, the system can make more accurate judgments based on multi-modal data, effectively improving the quality and efficiency of after-sales service.

[0085] In some embodiments of the present application, the deep learning model includes a Transformer model.

[0086] In some embodiments of the present application, the implementation process of using the Transformer model to perform interactive feature enhancement processing and fusion processing on the weighted text features and the weighted image features to obtain the problem classification result and the problem classification confidence may include but is not limited to the following steps.

[0087] First, assume that the weighted text feature is , and the weighted image feature is , and their dimension is . According to the weighted text feature , use the self-attention mechanism of the Transformer model to obtain the self-enhanced text feature . According to the weighted image feature , use the self-attention mechanism of the Transformer model to obtain the self-enhanced image feature 。The core formula of the self-attention mechanism of the Transformer model refers to formula (1).

[0088] Secondly, based on the weighted text features and the weighted image features , using the cross-modal attention mechanism of the Transformer model, cross-modal enhanced text features and cross-modal enhanced image features are obtained. Among them, the cross-modal enhanced text features satisfy the following formula (4): (4); In formula (4), represents 's learning rate parameter, controlling 's update amplitude, represents the attention score of the weighted text features based on the weighted image features, ; the cross-modal enhanced image features satisfy the following formula (5): (5); In formula (5), represents 's learning rate parameter, controlling 's update amplitude, represents the attention score of the weighted image features based on the weighted text features, .

[0089] Then, based on the self-enhanced text features , the self-enhanced image features , the cross-modal enhanced text features and the cross-modal enhanced image features , interactive iterative updates are performed to obtain the final enhanced text features and . According to and , using the multi-layer perceptron model, the problem classification probability distribution satisfies the following formula (6): (6); In formula (6), represents the multi-layer perceptron model, represents the learnable weight parameter related to , represents the learnable bias parameter related to .

[0090] Finally, based on the problem classification probability distribution , the problem classification result and the confidence level of the problem classification are obtained.

[0091] Through the above steps, the embodiments of the present application achieve efficient interactive enhancement and fusion processing of weighted text features and weighted image features, ensuring that the system can comprehensively understand after-sales problems from different perspectives, and generate accurate problem classification results and confidence scores. This not only enhances the decision-making ability of the system, but also provides a solid foundation for subsequent automated processing or manual review, ensuring that customer problems are solved promptly and effectively.

[0092] In some embodiments of the present application, in step 106, the implementation process of processing the after-sales problem according to the problem classification result, the confidence level of the problem classification, and the processing information corresponding to the after-sales problem may include but are not limited to the following steps.

[0093] If the confidence level of the problem classification corresponding to the after-sales problem exceeds the preset confidence threshold, the after-sales problem is automatically processed according to the problem classification result and the processing information corresponding to the after-sales problem. Otherwise, the problem classification result and the processing information corresponding to the after-sales problem are sent to the artificial customer service for the artificial customer service to process the after-sales problem.

[0094] In this step, if the confidence level of the problem classification corresponding to the after-sales problem exceeds the preset confidence threshold, the system will directly process the after-sales problem through the automated processing module according to the problem classification result and the processing information. This method can quickly generate a detailed processing plan in the case of high confidence and send it to the customer in the form of a formal notice, reducing the waiting time, improving the service efficiency and customer satisfaction. Automated processing not only improves the response speed, but also reduces the need for manual intervention and lowers the operating cost.

[0095] Otherwise, when the confidence level of the problem classification is lower than the preset threshold, the system will send the problem classification result and its initially generated processing information to the artificial customer service team for further review and processing. Based on this information, the artificial customer service can ensure that complex or uncertain problems receive the attention of professionals and are properly solved, avoiding misjudgment or improper handling caused by system limitations, providing more personalized and accurate services, and thus comprehensively improving the quality of after-sales service and the customer experience.

[0096] In summary, the intelligent e-commerce customer service after-sales processing method proposed by the embodiments of the present application can provide the following technical effects.

[0097] In the embodiments of the present application, by comprehensively analyzing text, image, and conversation history data, and performing interactive feature enhancement and fusion processing in combination with the importance ratio, the accuracy and confidence of problem classification are effectively improved, and misjudgment is reduced. The automated processing module can quickly generate a detailed processing plan and directly send it to the customer with high confidence, reducing the need for manual intervention, accelerating the response speed, improving service efficiency and customer satisfaction. For complex or low-confidence problems, the system sends preliminary processing suggestions to the human customer service for further review to ensure that the problems receive the attention of professionals and are properly resolved, avoiding customer dissatisfaction caused by system limitations. This method not only optimizes the user experience, reduces operating costs, but also improves resource utilization efficiency, effectively enhancing the quality and efficiency of after-sales service, bringing effective technical effects and economic benefits to enterprises and consumers.

[0098] Secondly, referring to Figure 3 , the embodiments of the present application provide an intelligent e-commerce customer service after-sales processing system, including a data acquisition module, a feature processing module, a weight adjustment module, a feature interaction module, and an after-sales processing module.

[0099] The data acquisition module is used to acquire text data, image data, conversation history data, and compensation mechanisms associated with after-sales problems.

[0100] The feature processing module is used to perform feature processing on the text data and image data to obtain text features and image features.

[0101] The weight adjustment module is used to obtain the importance ratio between the text features and the image features according to the conversation history data.

[0102] The feature interaction module is used to perform interactive feature enhancement processing and fusion processing on the text features and the image features in combination with the importance ratio to obtain a problem classification result and a problem classification confidence level.

[0103] The after-sales processing module is used to generate processing information corresponding to the after-sales problem according to the problem classification result, the problem classification confidence level, and the compensation mechanism; and process the after-sales problem according to the problem classification result, the problem classification confidence level, and the processing information corresponding to the after-sales problem.

[0104] In some embodiments of the present application, the after-sales processing module includes an automated processing unit and a manual processing unit.

[0105] The automated processing unit is used to automatically process the after-sales problem according to the problem classification result and the processing information corresponding to the after-sales problem when the problem classification confidence level corresponding to the after-sales problem exceeds a preset confidence threshold.

[0106] The manual processing unit is configured to send the problem classification result and processing information corresponding to the after-sales problem to the human customer service when the confidence level of the problem classification corresponding to the after-sales problem does not exceed a preset confidence threshold, so as to facilitate the human customer service to process the after-sales problem.

[0107] Furthermore, referring to Figure 4 , an intelligent e-commerce customer service after-sales processing device provided by an embodiment of the present application includes a processor and a memory, and the memory is used to store programs. When the program is executed by the processor, the processor implements the foregoing intelligent e-commerce customer service after-sales processing method.

[0108] In addition, an embodiment of the present application provides a computer-readable storage medium, in which a program executable by a processor is stored, and the program executable by the processor is used to implement the foregoing intelligent e-commerce customer service after-sales processing method when executed by the processor.

[0109] Similarly, the content in the foregoing method embodiments is applicable to the system embodiments, device embodiments, and storage medium embodiments of the present application. The functions specifically implemented by the system embodiments, device embodiments, and storage medium embodiments of the present application are the same as those of the foregoing method embodiments, and the beneficial effects achieved are also the same as those of the foregoing method embodiments.

[0110] In some alternative embodiments, the functions / operations mentioned in the block diagram may not occur in the order mentioned in the operation diagram. For example, depending on the functions / operations involved, two consecutive blocks shown may actually be executed substantially simultaneously or the blocks can sometimes be executed in the reverse order. In addition, the embodiments presented and described in the flowcharts of the present invention are provided by way of example for the purpose of providing a more comprehensive understanding of the technology. The disclosed method is not limited to the operations and logical flows presented herein. Alternative embodiments are contemplated in which the order of various operations is changed and sub-operations described as part of a larger operation are executed independently. In addition, although the present invention has been described in the context of functional modules, it should be understood that, unless otherwise stated to the contrary, one or more of the functions and / or features may be integrated in a single physical device and / or software module, or one or more functions and / or features may be implemented in separate physical devices or software modules. It should also be understood that a detailed discussion of the actual implementation of each module is not necessary for an understanding of the present invention. Rather, given the attributes, functions, and internal relationships of the various functional modules in the devices disclosed herein, the actual implementation of the modules will be understood within the ordinary skills of an engineer. Thus, those skilled in the art can implement the present invention as set forth in the claims without undue experimentation. It should also be understood that the specific concepts disclosed are merely illustrative and are not intended to limit the scope of the present invention, which is determined by the full scope of the appended claims and their equivalents. If the described function is implemented in the form of a software functional unit and sold or used as an independent product, it may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, may be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several programs for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes: various media such as USB flash drives, mobile hard disks, read-only memories (ROMs), random access memories (RAMs), magnetic disks, or optical discs that can store program codes. The logic and / or steps represented in the flowchart or otherwise described herein, for example, may be considered as a sequenced list of executable programs for implementing logical functions, and may be specifically implemented in any computer-readable medium for use by a program execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can retrieve and execute programs from a program execution system, apparatus, or device), or in conjunction with these program execution systems, apparatuses, or devices. For the purposes of this specification, a "computer-readable medium" may be any device that can contain, store, communicate, propagate, or transport a program for use by or in conjunction with a program execution system, apparatus, or device. More specific examples (nonexhaustive list) of computer-readable media include the following: an electrical connection (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, as the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing as appropriate, and then stored in a computer memory. It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable program execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like. In the above description of this specification, the description with reference to the terms "one embodiment / example", "another embodiment / example", or "certain embodiments / examples", etc. means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in the embodiments or examples of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples. Although the embodiments of the present invention have been shown and described, those of ordinary skill in the art can understand that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the claims and their equivalents. The above has specifically described the preferred embodiments of the present invention, but the present invention is not limited to the described embodiments. Those skilled in the art can also make various equivalent deformations or substitutions without departing from the spirit of the present invention, and these equivalent deformations or substitutions are all included within the scope defined by the claims of the present invention.

Claims

1. An intelligent e-commerce customer service after-sales processing method, characterized in that: The steps include: Obtain text data, image data, conversation history data, and compensation mechanisms associated with after-sales issues; Performing feature processing on the text data and the image data to obtain text features and image features; Obtaining, according to the conversation history data, an importance ratio between the text feature and the image feature; In combination with the importance ratio, interactive feature enhancement processing and fusion processing are performed on the text features and the image features to obtain a question classification result and a question classification confidence; Generate processing information corresponding to the after-sales problem according to the problem classification result, the problem classification confidence and the compensation mechanism; The after-sales problem is processed according to the problem classification result, problem classification confidence and processing information corresponding to the after-sales problem.

2. The intelligent e-commerce customer service after-sales processing method according to claim 1 is characterized in that: The performing feature processing on the text data and the image data to obtain text features and image features includes: Performing feature processing on the text data using a preset text processing model to obtain the text features; The image data is subjected to feature processing using a preset image processing model to obtain the image features.

3. The intelligent e-commerce customer service after-sales processing method according to claim 1 is characterized in that: The obtaining, according to the conversation history data, the importance ratio between the text feature and the image feature comprises: Performing feature processing on the conversation history data using a preset text processing model to obtain a conversation history vector; According to the conversation history vector, an importance ratio between the text feature and the image feature is obtained.

4. The intelligent e-commerce customer service after-sales processing method according to claim 1 is characterized in that: The interactive feature enhancement processing and fusion processing are performed on the text features and the image features in combination with the importance ratio to obtain the question classification result and the question classification confidence, including: In combination with the importance ratio, linear transformation is performed on the text feature and the image feature to obtain weighted text features and weighted image features of the same dimension; By using a deep learning model, interactive feature enhancement processing and fusion processing are performed on the weighted text features and the weighted image features to obtain a question classification result and a question classification confidence.

5. The intelligent e-commerce customer service after-sales processing method according to claim 4 is characterized in that: The method of using the deep learning model to perform interactive feature enhancement processing and fusion processing on the weighted text features and the weighted image features to obtain a question classification result and a question classification confidence includes: According to the weighted text features, a self-attention mechanism of a deep learning model is used to obtain a self-enhanced text feature; According to the weighted image features, a self-attention mechanism of a deep learning model is used to obtain self-enhanced image features; According to the weighted text features and the weighted image features, using the cross-modal attention mechanism of the deep learning model, a cross-modal enhanced text feature and a cross-modal enhanced image feature are obtained; According to the self-enhanced text features, the self-enhanced image features, the cross-modal enhanced text features and the cross-modal enhanced image features, a multi-layer perceptron model is used to obtain the question classification result and the question classification confidence.

6. The intelligent e-commerce customer service after-sales processing method according to claim 1 is characterized in that: The processing of the after-sales problem according to the problem classification result, the problem classification confidence and the processing information corresponding to the after-sales problem includes: If the problem classification confidence level corresponding to the after-sales problem exceeds a preset confidence threshold, automatically process the after-sales problem according to the problem classification result and processing information corresponding to the after-sales problem; Otherwise, the problem classification result and processing information corresponding to the after-sales problem are sent to the manual customer service to facilitate the manual customer service to handle the after-sales problem.

7. An intelligent e-commerce customer service after-sales processing system, characterized in that: It includes data acquisition module, feature processing module, weight adjustment module, feature interaction module and after-sales processing module; The data acquisition module is used to acquire text data, image data, conversation history data and compensation mechanism associated with after-sales issues; The feature processing module is used to perform feature processing on the text data and the image data to obtain text features and image features; The weight adjustment module is used to obtain the importance ratio between the text feature and the image feature according to the conversation history data; The feature interaction module is used to perform interactive feature enhancement processing and fusion processing on the text features and the image features in combination with the importance ratio to obtain a question classification result and a question classification confidence; The after-sales processing module is used to generate processing information corresponding to the after-sales problem according to the problem classification result, the problem classification confidence and the compensation mechanism; The after-sales problem is processed according to the problem classification result, problem classification confidence and processing information corresponding to the after-sales problem.

8. The intelligent e-commerce customer service after-sales processing system according to claim 6 is characterized in that: The after-sales processing module includes an automated processing unit and a manual processing unit; The automated processing unit is used to automatically process the after-sales problem according to the problem classification result and processing information corresponding to the after-sales problem when the problem classification confidence corresponding to the after-sales problem exceeds a preset confidence threshold; The manual processing unit is used to send the problem classification result and processing information corresponding to the after-sales problem to the manual customer service when the problem classification confidence corresponding to the after-sales problem does not exceed a preset confidence threshold, so as to facilitate the manual customer service to handle the after-sales problem.

9. An intelligent e-commerce customer service after-sales processing device, characterized in that: It includes a processor and a memory, the memory is used to store a program; when the program is executed by the processor, the processor implements the intelligent e-commerce customer service after-sales processing method as described in any one of claims 1 to 6.

10. A computer-readable storage medium storing a program executable by a processor, characterized in that: The program executable by the processor is used to implement the intelligent e-commerce customer service after-sales processing method as described in any one of claims 1 to 6 when executed by the processor.

Images