Mail processing method and device, electronic equipment, storage medium and program product

Abstract

The invention provides a mail processing method and device, electronic equipment, a storage medium and a program product, and relates to the field of financial science and technology or other related fields. The method comprises the steps of obtaining text content of a mail and metadata related to a sender, processing the text content through a classification model to generate first probability distribution, and generating second probability distribution based on the metadata and historical data of the sender, and the category of the mail can be determined by fusing the first probability distribution and the second probability distribution. Through two-channel processing of email text semantics and metadata, the limitation of single semantic features can be eliminated, the ambiguity and wrong classification problems of semantically similar emails are solved, the email classification accuracy is improved, and the subsequent processing flow can be conveniently carried out based on the email classification result.

Description

Technical Field

[0001] This application relates to the field of financial technology or other related fields, and in particular to an email processing method, apparatus, electronic device, storage medium and program product. Background Technology

[0002] Email is an essential tool for the daily operations of businesses and organizations. Due to complex organizational structures and diverse business types, email system management is quite challenging. This is especially true in the banking and financial sectors, where a large number of cross-departmental emails of different types, such as business requests, approval processes, and internal notifications, require accurate categorization before targeted processing.

[0003] Currently, most companies use keyword-based automatic email categorization. This method is not very accurate and often misclassifies emails with similar semantic meanings, affecting the efficiency and reliability of subsequent processing. Summary of the Invention

[0004] This application provides an email processing method, apparatus, electronic device, storage medium, and program product to solve the technical problem of inaccurate email classification.

[0005] Firstly, this application provides an email processing method, including:

[0006] Retrieve the text content of the email and metadata related to the sender;

[0007] The text content is processed using a classification model to generate a first probability distribution;

[0008] Based on metadata and sender's historical data, a second probability distribution is generated. Both the first and second probability distributions represent the probability that an email belongs to each of the preset multiple email categories.

[0009] The first probability distribution and the second probability distribution are merged, and the category of the email is determined based on the fusion result.

[0010] Secondly, this application provides an email processing apparatus, comprising:

[0011] The acquisition module is used to acquire the text content of the email and metadata related to the sender;

[0012] The first processing module is used to process the text content through a classification model and generate a first probability distribution;

[0013] The second processing module is used to generate a second probability distribution based on metadata and the sender's historical data. Both the first and second probability distributions represent the probability that an email belongs to each of the preset multiple email categories.

[0014] The classification module is used to fuse the first probability distribution and the second probability distribution, and determine the category of the email based on the fusion result.

[0015] Thirdly, this application provides an electronic device, including: a processor, and a memory communicatively connected to the processor;

[0016] The memory stores the instructions that the computer executes;

[0017] The processor executes computer execution instructions stored in memory to implement a mail processing method as described in any of the first aspects.

[0018] Fourthly, this application provides a computer-readable storage medium, comprising: computer-executable instructions stored in the computer-readable storage medium, which, when executed by a processor, are used to implement an email processing method as described in any of the first aspects.

[0019] Fifthly, this application provides a computer program product, including a computer program that, when executed by a processor, implements an email processing method as described in any of the first aspects.

[0020] The email processing method, apparatus, electronic device, storage medium, and program products provided in this application acquire the text content of the email and metadata related to the sender. A first probability distribution is generated by processing the text content through a classification model, and a second probability distribution is generated based on the metadata and the sender's historical data. The email category is then determined by fusing the first and second probability distributions. This dual-channel processing of email text semantics and metadata eliminates the limitations of single semantic features, resolves ambiguity and misclassification issues of semantically similar emails, and improves email classification accuracy, enabling subsequent processing based on the email classification results. Attached Figure Description

[0021] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0022] Figure 1 This is a schematic diagram of an application scenario provided by an embodiment of this application;

[0023] Figure 2 A flowchart illustrating an email processing method provided in an embodiment of this application;

[0024] Figure 3 This is a schematic diagram of the processing flow of an intelligent classification layer provided in an embodiment of this application;

[0025] Figure 4 A schematic diagram of the processing flow of the verification and evidence storage layer provided in this application embodiment;

[0026] Figure 5 This application provides a schematic diagram of the entire email processing workflow.

[0027] Figure 6 This is a schematic diagram of the structure of an email processing device provided in an embodiment of this application;

[0028] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.

[0029] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0030] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0031] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of the relevant data all comply with the relevant laws, regulations, and standards of the relevant countries and regions, have taken necessary confidentiality measures, do not violate public order and good morals, and provide corresponding operation access points for users to choose to authorize or refuse.

[0032] Furthermore, the technical solution involved in this application, which involves big data analysis of user information (including but not limited to personal biometrics, identity data, consumption data, asset data, electronic terminal operation data, etc.) and the use of artificial intelligence technology for automated decision-making, and makes decisions that have a significant impact on personal rights based on the results of automated decision-making, provides users with corresponding operation entry points for users to choose to agree to or reject the results of automated decision-making; if the user chooses to reject, the process will proceed to the expert decision-making process.

[0033] It should be noted that the email processing method, apparatus, electronic device, storage medium and program product provided in this application can be used in the field of financial technology, or in any field other than financial technology. The application field of the email processing method, apparatus, electronic device, storage medium and program product in this application is not limited.

[0034] In this application, it should be understood that the terms used have the following meanings:

[0035] BERT: A bidirectional language model based on a Transformer encoder, which utilizes a masked language model and a next-sentence prediction task for large-scale pre-training.

[0036] ERP: Enterprise Resource Planning, a comprehensive management system that integrates a company's business and information flows.

[0037] Blockchain: Blockchain is a distributed database ledger system that encrypts and hashes data and its operation records using cryptographic methods, creating a transparent, immutable, and traceable database. This technology can be widely used in payment clearing, evidence preservation and retrieval, value transfer, supply chain finance, user credit investigation, and regulatory auditing. Blockchain records all transactions since the genesis block; once the block size reaches a certain level, historical transaction records will not be altered.

[0038] Closed-loop verification center: refers to a security module that uses blockchain to store key operations and synchronously verify the consistency between email tasks and business system status.

[0039] In enterprises and financial institutions, email is a core medium for cross-departmental collaboration and business processes. Due to complex organizational structures and diverse business types, there are a large number of emails involving business requests, approval processes, and internal notifications. These emails need to be accurately categorized before they can be processed in a targeted manner to achieve efficient collaboration and compliance management.

[0040] Currently, most companies use keyword matching-based automatic email classification. However, traditional rule engines struggle to distinguish between semantically similar email types (such as equipment procurement requests and supply chain optimization suggestions). They also have low accuracy in identifying email categories with similar wording or multiple meanings, which can easily lead to misclassification and affect the efficiency and reliability of subsequent business processes.

[0041] To address the aforementioned issue of inaccurate email classification, this application proposes a technical concept: a dual-channel processing architecture based on text semantic feature channels and metadata feature channels for email classification. The text semantic feature channel identifies text semantics and enhances classification accuracy by incorporating industry knowledge bases, while the metadata feature channel identifies and processes information such as the sender's identity and department. The processing results from both channels are combined to determine the email category. This dual-channel fusion mechanism effectively distinguishes semantically similar email types (such as equipment procurement requests versus supply chain optimization suggestions), reduces the need for manual intervention, resolves ambiguity and misclassification issues related to semantically similar emails, and improves email classification accuracy, enabling subsequent processing based on the email classification results.

[0042] Figure 1 This is a schematic diagram illustrating an application scenario provided by an embodiment of this application. For example... Figure 1 As shown, terminal 102 communicates with server 101 via a network. A data storage system can store the data that server 101 needs to process. The data storage system can be integrated onto server 101 or located in the cloud or on other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc. Server 101 can be implemented using a standalone server or a server cluster consisting of multiple servers.

[0043] The email processing method, apparatus, electronic device, storage medium, and program products provided in this application can be applied to, for example... Figure 1 In the application environment shown, users can edit the subject, body, and recipient description of an email through terminal 102. Server 101 can obtain the relevant content of the email to be processed (such as email text and sender metadata) by communicating with terminal 102. Then, server 101 can classify the email according to this content and return the classification result to terminal 102 or use the classification result for a preset subsequent processing flow.

[0044] The email processing methods, apparatus, electronic devices, storage media, and program products provided in this application are intended to solve the above-mentioned technical problems of the prior art.

[0045] The technical solution of this application and how the technical solution of this application solves the above-mentioned technical problems are described in detail below with specific embodiments. These specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will now be described with reference to the accompanying drawings.

[0046] Figure 2 This is a flowchart illustrating an email processing method provided in an embodiment of this application. Figure 2 As shown, the method includes:

[0047] Step S201: Obtain the text content of the email and metadata related to the sender.

[0048] The text content of the email can include the subject and the text content in the body. Metadata can include the sender's identity information (such as employee ID), department, time zone, and other attribute information.

[0049] The email processing method of this application embodiment can be applied to an email management system, which can obtain emails to be processed from the recipient's mailbox or directly obtain emails that have not yet been sent from the sender's email client.

[0050] Step S202: Process the text content using a classification model to generate a first probability distribution.

[0051] The classification model can be trained on top of BERT using emails with category labels. The first probability distribution represents the probability that an email belongs to each of several predefined email categories.

[0052] For example, a classification model may include multiple convolutional layers and at least one fully connected layer. The convolutional layers can be used to extract and process features from the text content. The processing results of the convolutional layers can be input into the fully connected layer, which can generate a probability distribution representing the email category through an activation function (such as the softmax function).

[0053] Step S203: Generate a second probability distribution based on metadata and the sender's historical data.

[0054] The second probability distribution represents the probability that an email belongs to each of a set of preset email categories.

[0055] For example, in the metadata feature channel, historical data such as the sender's historical habits and behavioral characteristics can be obtained based on the sender's identity. These data, combined with metadata features such as the sender's historical habits, high-frequency words in historical emails, and the business attributes of the sender's department, are used to perform feature calculations and generate a probability distribution representing the email category. Taking high-frequency words in historical emails as an example, emails exchanged between the sender and recipient departments can be modeled to extract high-frequency keywords from historical emails. For instance, if a code testing department frequently sends test scripts to a development department via email, then emails sent by the code testing department will have a higher feature score in the test task category. This feature score can be mapped to the probability that the email belongs to that category.

[0056] Step S204: Fuse the first probability distribution and the second probability distribution, and determine the category of the email based on the fusion result.

[0057] For example, the system can preset multiple email categories, including category A, category B, category C, category D, and category E. A first probability distribution is [category A: 0.1, category B: 0.4, category C: 0.3, category D: 0.2, category E: 0], and a second probability distribution is [category A: 0.2, category B: 0.1, category C: 0.3, category D: 0, category E: 0.4]. If the two probability distributions are merged with a 1:1 weight ratio, the merging process can be represented as [category A: 0.1 + 0.2 = 0.3, category B: 0.4 + 0.1 = 0.5, category C: 0.3 + 0.3 = 0.6, category D: 0.2 + 0 = 0.2, category E: 0 + 0.4 = 0.4]. It can be seen that category C has a probability of 0.6 in the merged probability distribution, which is the highest probability after merging. Therefore, category C can be used as the category of this email.

[0058] In some possible implementations, different weights can be assigned to the two probability distributions when fusing the first and second probability distributions. This process can include:

[0059] The probability of each class in the first probability distribution is summed in a weighted manner with the corresponding probability of the class in the second probability distribution. In the process of calculating the weighted sum, the weight of the class probability in the first probability distribution is greater than the weight of the class probability in the second probability distribution.

[0060] The first probability distribution is derived from the semantics of the email text, while the second probability distribution is derived from the email's metadata. A higher weight can be assigned to the first probability distribution to make the classification result more aligned with the semantics of the text. For example, if category A has probabilities of 0.1 and 0.2 in the first and second probability distributions respectively, and the first and second probability distributions are merged with a weight ratio of 0.7:0.3, the probability fusion process for category A is 0.1×0.7+0.2×0.3=0.13. This step establishes a weight allocation principle in the fusion decision that prioritizes the semantics of the email text and supplements it with the sender's metadata. This ensures that while retaining the reference value of historical behavior, the classification result more closely matches the actual content of the current email. By assigning higher decision weight to the semantics of the text, potential classification bias caused by the solidification or deviation of the sender's historical behavior patterns can be effectively suppressed. When the email content is clear and unambiguous, this mechanism can strengthen semantic-driven classification judgment; when the content is ambiguous, it can use metadata weights for reasonable calibration. This clear-cut fusion strategy can further improve the accuracy and adaptability of the classification system in recognizing the immediate intent of emails.

[0061] The email processing method in this embodiment obtains the text content of the email and metadata related to the sender. A classification model is used to process the text content to generate a first probability distribution, and based on the metadata and the sender's historical data, a second probability distribution is generated. The first and second probability distributions are then fused to determine the email category. This dual-channel processing of email text semantics and metadata eliminates the limitations of single semantic features, resolves ambiguity and misclassification issues between semantically similar emails, and improves email classification accuracy, enabling subsequent processing based on the email classification results.

[0062] In one embodiment, a second probability distribution is generated based on metadata and the sender's historical data, including:

[0063] Obtain the sender's historical email data; analyze the historical email data to extract the sender's behavioral statistical features across multiple email categories; and generate a second probability distribution based on the behavioral statistical features.

[0064] Among these, behavioral statistical features may include: the quantity distribution of the sender's historical emails under each email category; and the high-frequency keywords of the sender's historical emails corresponding to each email category.

[0065] In this embodiment, historical email data of the sender is obtained, and their behavioral statistical features under various categories, such as quantity distribution or high-frequency keywords, are analyzed and extracted. A second probability distribution is then generated based on these features, quantifying the sender's email tendencies from their historical behavior and providing supplementary probabilistic basis for classification independent of the current email text. By introducing historical sender behavior statistics, stable email sending patterns and content preferences can be identified, thus providing classification correction based on the sender's behavioral context when the email text is semantically ambiguous or contains ambiguities. This effectively reduces misjudgments caused by relying solely on text semantics, further improving the accuracy and reliability of overall email classification.

[0066] In one embodiment, after determining the category of the email, the following steps are also included:

[0067] Extract information from the text content of the email and perform subsequent processing based on the extracted information, which corresponds to the category.

[0068] In enterprise email management processes, emails often require manual processing even after being categorized. This is particularly problematic in cross-departmental collaboration scenarios, where manual processing presents significant efficiency bottlenecks: emails must undergo multiple levels of manual identification, forwarding, or entry, leading to delays in information delivery and increasing the risk of missed deliveries, misinterpretations, or duplicate submissions. This human-driven approach is not only slow and time-consuming but also fails to guarantee the accuracy and consistency of task flow between departments, becoming a major obstacle to automating email-driven business processes.

[0069] In this embodiment, based on automatic email classification, a classification-driven automated processing mechanism is further designed. Specifically, the system can automatically extract key information from the email text content according to the determined email category and trigger a preset processing flow bound to that category. For example, if the email belongs to the "business request" category, the request parameters (such as amount, date, and number) are extracted and the interface is called to synchronize to the business system; if it belongs to the "decision consultation" category, the department or role requiring opinions is parsed, an optimized routing path is automatically generated, and forwarded; if it belongs to the "department notification" category, the notification recipient and urgency are identified, and multiple channels such as email, instant messaging, or SMS are adaptively selected for sending. This solution achieves an end-to-end automated closed loop from "identification and classification" to "execution of operations." By deeply coupling the email classification results with subsequent business processing flows, the number of manual intervention steps can be reduced, information delays and transmission distortions in cross-departmental workflows can be avoided, and the accuracy and consistency of task processing can be ensured through structured information extraction and automatic process triggering, significantly improving the overall efficiency and automation level of email management.

[0070] For example, information is extracted from the text content of the email, and subsequent processing flows corresponding to the category are executed based on the extracted information, including:

[0071] From the text content, the structured field extraction engine identifies and extracts target parameters related to the business request; the target parameters are then synchronized in real time to the business system corresponding to the business request to trigger and generate the corresponding business work order or business processing flow.

[0072] This example can be used for business request emails. Once an email is categorized as a business request, the system automatically identifies and extracts key business parameters from the email body using a pre-defined structured field extraction engine. For instance, for a server resource request email, the engine will extract the system type, configuration specifications, and usage period; for a procurement approval email, it will extract the supplier number, amount, and delivery date. These extracted parameters can be synchronized in real-time and in a structured manner to the corresponding business system (such as ERP) via an API gateway, thereby automatically triggering and generating standard business work orders or initiating downstream processing flows.

[0073] In the above example, by converting unstructured email text into structured parameters that the system can recognize and achieving seamless integration with the backend business system, the manual reading, manual entry, and cross-system submission steps in the traditional process can be eliminated, greatly improving processing efficiency. Automated synchronization avoids data errors or omissions that may be caused by manual transcription, ensuring the consistency and accuracy of business data in the process of flow, and providing key support for the agility and reliability of the enterprise's core business processes.

[0074] For example, information is extracted from the text content of the email, and subsequent processing flows corresponding to the category are executed based on the extracted information, including:

[0075] Extract the event fingerprint representing the email business event from the text content; match the associated historical processing path in the historical task library based on the event fingerprint; generate an optimized decision routing path based on the current organizational structure information and the matched historical processing path.

[0076] The decision routing path is used to distribute emails to recipients who are relevant to decisions regarding email business matters.

[0077] This example can be used for automated routing of emails requiring decision-making input. The system extracts a unique fingerprint from the email text that identifies the core characteristics of the business matter (e.g., "Munich factory expansion = DE-MUC022"). This fingerprint is then used to search the historical task database, matching processing paths and node information for similar historical tasks. Combined with the current real-time organizational structure, the system compresses and optimizes the path using intelligent routing optimization mechanisms (such as loading the organizational topology and running pruning algorithms to eliminate redundant intermediate nodes), thereby generating a final distribution path directly to the core decision-making node. Furthermore, this process can include mechanisms to prevent duplicate task distribution, such as semantic similarity comparison, task timeliness overlap analysis, and departmental load balancing, to ensure accurate and efficient routing.

[0078] In the above example, by combining historical processing experience with real-time organizational status, intelligent and direct routing of decision-making emails can be achieved. This changes the inefficient traditional model of relying on manual judgment and layer-by-layer forwarding, automatically bypassing unnecessary intermediate levels and significantly shortening the transmission chain and turnaround time for opinion solicitation. Simultaneously, anti-duplicate mechanisms effectively prevent tasks from being sent repeatedly or missed, ensuring that key decision-making information reaches the responsible parties quickly and accurately, greatly improving the response speed and processing quality of cross-organizational and cross-regional collaborative decision-making.

[0079] For example, information is extracted from the text content of the email, and subsequent processing flows corresponding to the category are executed based on the extracted information, including:

[0080] The coordinates of the target paragraph representing the email's intended message are extracted from the text content. A data probe is then embedded into the email based on the extracted coordinates before it is sent to the recipient. The data probe is used to obtain the recipient's reading feedback on the target paragraph.

[0081] This example can be used for emails that need to notify specific individuals or departments. The system can locate key paragraphs from the email text and embed lightweight data probes (such as JS probes) that can track the recipient's reading interactions, such as scrolling, clicking, etc., on specific paragraphs.

[0082] In some possible implementations, the system can also parse the email text content to determine the notification's level and nature, and then execute intelligent distribution via a hierarchical delivery engine. Distribution can include three methods: main channel, emergency channel, and source tracing channel. The main channel achieves precise delivery at the departmental level according to the organizational structure tree; the emergency channel initiates a three-terminal linkage of email, communication tools, and SMS for red-level notifications; and the source tracing channel mandates the addition of digital signatures and timestamps for compliance notifications.

[0083] After sending an email, the system can integrate reading data and feedback tasks to generate a multi-dimensional dashboard: a heatmap showing departmental reading coverage; a feedback funnel chart showing completion rates at each stage; and automatic identification of unread users and push notifications.

[0084] In the above example, by using technical means to transform traditionally scattered and hidden email reading feedback into structured and analyzable data, it is possible to achieve quantitative management of the entire process of notification delivery effectiveness. This approach enables trusted reading verification, unified management of notifications and feedback, solves the problem of information discontinuity, and significantly improves the reliability, compliance, and efficiency of important notifications delivered within the organization.

[0085] In one embodiment, the email processing method further includes:

[0086] Evidence data generated from actions triggered by emails is written to the blockchain, and anomaly monitoring is performed based on the data in the blockchain.

[0087] In email management scenarios, there are also issues such as the reliance on manual operation to synchronize email tasks and business system status, which may lead to data inconsistencies, and the reliance on centralized log archiving for email operations, which may result in compliance issues that fail to meet audit requirements.

[0088] This embodiment constructs a blockchain-based email operation auditing and monitoring system. Specifically, when a critical operation (such as approval, rejection, or read confirmation) is triggered for an email, the system generates composite data including an operation timestamp, operator identity credentials, and semantic fingerprint of the email content, calculates its hash value, and writes it to the blockchain, thus forming a time-sequential, tamper-proof operation evidence chain. Simultaneously, the system monitors the entire process in real time through a visual dashboard and blockchain data. Once an anomaly in the evidence chain or inconsistency in the business process status is detected, a pre-set multi-level alarm mechanism is automatically triggered.

[0089] In the above embodiments, by leveraging the immutability of blockchain, not only can email operations be recorded on the chain to solve the system state synchronization problem, but also trusted evidence can be provided for critical email operations. This addresses the issues of traditional centralized logs being easily tampered with and failing to meet audit requirements, thereby enhancing the transparency, traceability, and security of the email processing workflow.

[0090] The email management system of this application embodiment can adopt a three-layer intelligent processing architecture, which may specifically include an intelligent classification layer, a typological processing layer, and a verification and evidence storage layer.

[0091] Figure 3 This is a schematic diagram illustrating the processing flow of an intelligent classification layer provided in an embodiment of this application. Figure 3 As shown, the intelligent classification layer deploys an enterprise-grade multimodal classification engine, achieving accurate recognition through dual-channel dynamic fusion:

[0092] The text semantic channel uses a pre-trained BERT language model to parse email content and embeds an industry knowledge base to map professional terms. For example, "SAP E105 fault code" is associated with the IT service ticket class, "AP invoice verification failure" is associated with the financial anomaly ticket class, and items containing keywords such as "notification," "instructions," and "announcement" but without feedback requirements are associated with the notification class.

[0093] The meta-feature channel can comprehensively calculate features such as departmental business attributes, sender's historical behavior patterns, and cross-time zone sensitivity. Among them, cross-time zone sensitivity can characterize the degree of matching between the sender's actual sending time and the sender's normal working hours in their time zone.

[0094] The type-based processing layer can implement differentiated processes according to email categories, such as extracting target parameters for business application emails and synchronizing them to the business system. For details, please refer to the above embodiments, which will not be repeated here.

[0095] Figure 4This is a schematic diagram illustrating the processing flow of a verification and evidence storage layer provided in an embodiment of this application. The verification and evidence storage layer can operate in a closed-loop verification center and may include a state synchronizer and a blockchain evidence storage module. The state synchronizer can continuously compare the email task status with ERP system records. The blockchain evidence storage module can generate a composite hash value containing a timestamp, operator certificate, and semantic fingerprint for key operations on the email (approval / rejection / read confirmation) and write it to the blockchain, forming an immutable evidence storage chain.

[0096] Figure 5 This is a schematic diagram illustrating the entire process of email processing provided in an embodiment of this application. Figure 5 As shown, senders can submit emails through the email management system client. The system can categorize the emails using a classification engine, extract the email content to generate a task fingerprint, and query the fingerprint database to return a code. The system can check if similar historical tasks exist. If they do, it returns a similar task number to the sender's client and terminates the process, allowing the sender to handle the task according to the historical task. If no similar tasks exist, it is considered a new task, and the topology engine is requested to run a pruning algorithm to optimize the path, determine the directly connected target responsible person as the recipient, send the email, obtain the recipient's feedback, write it to the blockchain record, and update the task status on the management dashboard. The email management system monitors the entire process in real time through the management dashboard, and abnormal events trigger a multi-level alarm mechanism.

[0097] Applying the above email management methods offers multiple benefits, such as: Improved classification accuracy: The dual-channel fusion mechanism effectively resolves ambiguity issues in emails with mixed semantics. Significantly improved collaboration efficiency: The topology pruning algorithm significantly compresses cross-departmental collaboration paths. Optimized resource consumption: The tiered outreach mechanism completely eliminates interference from invalid information on professional roles. Enhanced risk control system: Blockchain-based evidence storage enables full lifecycle traceability of critical operations.

[0098] It should be noted that, for the sake of simplicity, the foregoing method embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are all optional embodiments, and the actions and modules involved are not necessarily essential to this application.

[0099] It should be further noted that although the steps in the flowchart are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0100] Figure 6 This is a schematic diagram of the structure of an email processing device provided in an embodiment of this application, as shown below. Figure 6 As shown, the email processing device 600 includes:

[0101] The acquisition module 601 is used to acquire the text content of the email and metadata related to the sender;

[0102] The first processing module 602 is used to process the text content through a classification model and generate a first probability distribution;

[0103] The second processing module 603 is used to generate a second probability distribution based on metadata and the sender's historical data. Both the first probability distribution and the second probability distribution represent the probability that the email belongs to each of the preset multiple email categories.

[0104] The classification module 604 is used to fuse the first probability distribution and the second probability distribution, and determine the category of the email based on the fusion result.

[0105] In some possible implementations, the second processing module 603 can also be used to: obtain the sender's historical email data; analyze the historical email data and extract the sender's behavioral statistical features across multiple email categories; and generate a second probability distribution based on the behavioral statistical features.

[0106] In some possible implementations, the classification module 604 can also be used to: perform a weighted summation of each category probability in the first probability distribution and the corresponding category probability in the second probability distribution; in the process of calculating the weighted summation, the weight of the category probability in the first probability distribution is greater than the weight of the category probability in the second probability distribution.

[0107] In some possible implementations, the classification module 604 can also be used to: extract information from the text content of the email and perform subsequent processing procedures corresponding to the category based on the extracted information.

[0108] In some possible implementations, the classification module 604 can also be used to: identify and extract target parameters related to the business request from the text content through a structured field extraction engine; and to trigger and generate the corresponding business work order or business processing flow by synchronizing the target parameters to the business system corresponding to the business request in real time.

[0109] In some possible implementations, the classification module 604 can also be used to: extract a matter fingerprint representing the email business matter from the text content; match the associated historical processing path in the historical task library based on the matter fingerprint; and generate an optimized decision routing path based on the current organizational structure information and the matched historical processing path, the decision routing path being used to distribute the email to recipients related to the decision of the email business matter.

[0110] In some possible implementations, the classification module 604 can also be used to: extract the coordinates of the target paragraph representing the intention of the email from the text content, embed a data probe into the email based on the extracted coordinates, and send it to the recipient; obtain the recipient's reading feedback on the target paragraph through the data probe.

[0111] In some possible implementations, the classification module 604 can also be used to: generate evidence data for operations triggered by emails and write it into the blockchain, and perform anomaly monitoring based on the data in the blockchain.

[0112] The email processing device provided in this embodiment is used to execute the technical solutions in any of the foregoing method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0113] It should be understood that the above-described device embodiments are merely illustrative, and the device of this application can also be implemented in other ways. For example, the division of units / modules in the above embodiments is only a logical functional division, and there may be other division methods in actual implementation. For example, multiple units, modules, or components may be combined, or integrated into another system, or some features may be ignored or not executed.

[0114] Furthermore, unless otherwise specified, the functional units / modules in the various embodiments of this application can be integrated into one unit / module, or each unit / module can exist physically separately, or two or more units / modules can be integrated together. The integrated units / modules described above can be implemented in hardware or as software program modules.

[0115] Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 7 As shown, the electronic device 70 includes:

[0116] Processor 71, memory 72, and communication interface 73;

[0117] The memory 72 is used to store the executable instructions of the processor 71; the executable instructions can be instructions that the computer can execute.

[0118] The processor 71 is configured to execute the technical solutions in any of the foregoing method embodiments by executing executable instructions.

[0119] Optionally, the memory 72 can be either standalone or integrated with the processor 71.

[0120] Optionally, when the memory 72 is a device independent of the processor 71, the electronic device 70 may further include:

[0121] Bus 74, memory 72 and communication interface 73 are connected to processor 71 through bus 74 and complete communication with each other. Communication interface 73 is used to communicate with other devices.

[0122] Optionally, the communication interface 73 can be implemented using a transceiver. The communication interface is used to enable communication between the database access device and other devices (e.g., clients, read-write databases, and read-only databases). The memory may include random access memory (RAM) and may also include non-volatile memory, such as at least one disk drive.

[0123] Bus 74 can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of representation, only one line is used in the diagram, but this does not imply that there is only one bus or one type of bus.

[0124] The processors mentioned above can be general-purpose processors, including central processing units (CPUs), network processors (NPs), etc.; they can also be digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components.

[0125] The electronic device is used to execute the technical solutions in any of the foregoing method embodiments. Its implementation principle and technical effect are similar, and will not be described again here.

[0126] This application also provides a readable storage medium, which can be a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the technical solution provided in any of the foregoing method embodiments.

[0127] This application also provides a computer program product, including a computer program, which, when executed by a processor, is used to implement the technical solutions provided in any of the foregoing method embodiments.

[0128] If the integrated unit / module is implemented as a software program module and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard drive, magnetic disk, or optical disk.

[0129] In the above embodiments, the descriptions of each embodiment have their own emphasis. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions of other embodiments. The technical features of the above embodiments can be combined arbitrarily. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as the combination of these technical features does not contradict each other, it should be considered within the scope of this specification.

[0130] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the following claims.

[0131] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. An email processing method, characterized in that, include: Retrieve the text content of the email and metadata related to the sender; The text content is processed using a classification model to generate a first probability distribution; Based on the metadata and the sender's historical data, a second probability distribution is generated. Both the first probability distribution and the second probability distribution represent the probability that the email belongs to each of a set of preset email categories. The first probability distribution and the second probability distribution are fused together, and the category of the email is determined based on the fusion result.

2. The method according to claim 1, characterized in that, The step of generating a second probability distribution based on the metadata and the sender's historical data includes: Obtain the sender's historical email data; The historical email data is analyzed to extract the sender's behavioral statistical characteristics across multiple email categories; The second probability distribution is generated based on the aforementioned behavioral statistical features.

3. The method according to claim 2, characterized in that, The behavioral statistical characteristics include at least one of the following: The distribution of the number of historical emails of the sender under each email category; The sender's historical emails correspond to high-frequency keywords under each email category.

4. The method according to any one of claims 1 to 3, characterized in that, The fusion of the first probability distribution and the second probability distribution includes: The probability of each category in the first probability distribution and the corresponding category probability in the second probability distribution are weighted and summed. In the weighted summation calculation process, the weight of the class probability in the first probability distribution is greater than the weight of the class probability in the second probability distribution.

5. The method according to any one of claims 1 to 3, characterized in that, After determining the category of the email, the process also includes: Information is extracted from the text content of the email, and subsequent processing procedures corresponding to the category are executed based on the extracted information.

6. The method according to claim 5, characterized in that, The step of extracting information from the text content of the email and performing subsequent processing procedures corresponding to the category based on the extracted information includes: From the text content, the target parameters related to the business request are identified and extracted using a structured field extraction engine; The target parameters are synchronized in real time to the business system corresponding to the business request to trigger and generate the corresponding business work order or business processing flow.

7. The method according to claim 5, characterized in that, The step of extracting information from the text content of the email and performing subsequent processing procedures corresponding to the category based on the extracted information includes: Extract the event fingerprint representing the email business matter from the text content; Based on the fingerprint of the event, match the associated historical processing path in the historical task library; Based on the current organizational structure information and the matched historical processing paths, an optimized decision routing path is generated. The decision routing path is used to distribute the email to recipients who are related to the decision-making of the email business matter.

8. The method according to claim 5, characterized in that, The step of extracting information from the text content of the email and performing subsequent processing procedures corresponding to the category based on the extracted information includes: The coordinates of the target paragraph representing the message's intent are extracted from the text content, and a data probe is embedded into the email based on the extracted coordinates before it is sent to the recipient. The data probe is used to obtain the recipient's reading feedback on the target paragraph.

9. The method according to any one of claims 1 to 3, characterized in that, Also includes: The operation triggered by the email will generate evidence data, which will be written into the blockchain, and anomaly monitoring will be performed based on the data in the blockchain.

10. An email processing device, characterized in that, include: The acquisition module is used to acquire the text content of the email and metadata related to the sender; The first processing module is used to process the text content through a classification model to generate a first probability distribution; The second processing module is used to generate a second probability distribution based on the metadata and the sender's historical data. Both the first probability distribution and the second probability distribution represent the probability that the email belongs to each of a set of multiple preset email categories. The classification module is used to fuse the first probability distribution and the second probability distribution, and determine the category of the email based on the fusion result.

11. An electronic device, characterized in that, include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 9.

12. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 9.

13. A computer program product, characterized in that, Includes a computer program that, when executed by a processor, implements the method of any one of claims 1 to 9.

Images