Processing method and device of declaration mail, equipment, storage medium and program product

Abstract

The invention discloses a processing method, device, equipment, storage medium and program product, relates to the technical field of computers, and discloses a processing method of a declaration mail, which comprises the following steps: selecting candidate declaration mails according to content keywords and mail information of declaration mails for roaming fees; according to the context information of the candidate declaration e-mails, determining a target declaration e-mail; and extracting processing key information in the target announcement report mail, and selecting an announcement report processing solution according to the processing key information, the processing key information including at least one of a file name, an announcement report type and an announcement report solution. According to the invention, the processing efficiency of the declaration mail is improved.

Description

Technical Field

[0001] The present application relates to the field of computer technology, and in particular to a method, apparatus, device, storage medium and program product for processing declaration mails. Background Art

[0002] Complaint processing primarily involves sales personnel reading the email and determining the claim type. They then analyze and process the claim based on the claim type, and then email the results back to the recipient. With the rapid development of telecommunications services, international roaming data is growing exponentially. Faced with this massive influx of claims, sales personnel still manually analyze the issues, extract the analysis results and supporting data, and then email responses, resulting in extremely low efficiency. Summary of the Invention

[0003] The main purpose of this application is to provide a method, device, equipment, storage medium and program product for processing declaration mails, aiming to solve the technical problem of low efficiency in declaration processing.

[0004] To achieve the above-mentioned purpose, this application proposes a method for processing declaration mails, the method comprising:

[0005] Select candidate declaration emails based on the keywords and email information in the roaming fee declaration emails;

[0006] Determining a target declaration email based on the context information of the candidate declaration emails;

[0007] Extract key processing information from the target declaration email, and select a solution for declaration processing based on the key processing information, wherein the key processing information includes at least one of a file name, a declaration type, and a declaration request.

[0008] In one embodiment, after the step of selecting a solution for the claim processing according to the key processing information, the method further includes:

[0009] Generate SQL statements to be executed based on the solution of the declaration processing;

[0010] The SQL statement is executed in a preset solution database, and an execution result of the SQL statement is obtained. The solution database includes standard solutions.

[0011] In one embodiment, after the step of obtaining the execution result of the SQL statement, the method further includes:

[0012] generating a declaration processing result according to the execution result, the declaration processing result including at least one of document processing information, claim argumentation proof, conclusion, and additional explanation;

[0013] The report processing result is sent to the operator of the visited location.

[0014] In one embodiment, the step of selecting candidate declaration emails based on content keywords and email information of the roaming fee declaration email includes:

[0015] Establish a connection with the mail server and inspect the mailboxes of the mail server according to the mailbox name and inspection time;

[0016] During the inspection process, the declaration emails in different encoding formats are decoded and the contents of the historical reply emails to the declaration emails are obtained;

[0017] Determine the content keywords and the email information based on the decoded declaration email and the parsed content of the historical reply email;

[0018] Based on the content keywords and the email information, the candidate declaration email is selected.

[0019] In one embodiment, the step of extracting key processing information from the target declaration email further includes:

[0020] Constructing prompt words for a declaration expert model, wherein the prompt words include at least one of a goal, a context, an instruction, an expected output, and a constraint;

[0021] Constructing a roaming claim knowledge base of the claim expert model; the knowledge base includes at least one of proper noun descriptions, customer and roaming partner code sets, and call record types and their descriptions;

[0022] Based on the declaration expert model, the prompt words and the roaming declaration knowledge base, key information for processing the target declaration email is extracted.

[0023] In one embodiment, the method further comprises:

[0024] Get error type information of historical declaration emails;

[0025] Obtaining relevant information of the historical complaint email, the relevant information including at least one of the body text, a complaint description of the error type, and a complaint request;

[0026] Based on the error type information and the relevant information, historical declaration emails are selected as training texts, and the declaration expert model is iteratively trained based on the training texts.

[0027] In addition, to achieve the above-mentioned purpose, the present application also proposes a device for processing declaration mails, the device for processing declaration mails comprising:

[0028] A selection module is used to select candidate declaration emails based on content keywords and email information of the roaming fee declaration emails;

[0029] A determination module, configured to determine a target declaration email based on the context information of the candidate declaration emails;

[0030] The processing module is used to extract key processing information from the target declaration email and select a solution for declaration processing based on the key processing information. The key processing information includes at least one of the file name, declaration type and declaration request.

[0031] In addition, to achieve the above-mentioned purpose, the present application also proposes a declaration mail processing device, which includes: a memory, a processor, and a computer program stored on the memory and runnable on the processor, and the computer program is configured to implement the steps of the declaration mail processing method as described above.

[0032] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium. A computer program is stored on the storage medium, and when the computer program is executed by the processor, the steps of the method for processing declaration mails as described above are implemented.

[0033] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the method for processing declaration mails as described above.

[0034] One or more technical solutions proposed in this application have at least the following technical effects:

[0035] Quickly filter candidate emails using keywords, email information, and contextual information, reducing the workload of manual review and shortening response time. Automatically match pre-set solution libraries based on structured information like file name and application type, improving application processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0037] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0038] Figure 1This is a flowchart of the generation of the application declaration email;

[0039] Figure 2 A flowchart illustrating the first embodiment of the method for processing application declaration emails;

[0040] Figure 3 A flowchart of the second embodiment of the method for processing application declaration emails is provided;

[0041] Figure 4 This is a schematic diagram of the output processing information package in the method for processing the application declaration email;

[0042] Figure 5 This is a diagram of translating SQL statements in the method of processing emails for this application;

[0043] Figure 6 A flowchart of the third embodiment of the method for processing application declaration emails is provided;

[0044] Figure 7 This is a schematic diagram of the JSON format processing results of the application application email processing method;

[0045] Figure 8 This is a schematic diagram of the email reply text rules in the method for processing emails for this application;

[0046] Figure 9 This is a schematic diagram of the email reply process in the method for handling emails for this application;

[0047] Figure 10 A flowchart illustrating a fourth embodiment of the method for processing application declaration emails is provided;

[0048] Figure 11 This is a diagram of the IMAP-based mail inspection process in the method for processing mails submitted for this application;

[0049] Figure 12 This is a schematic diagram of the screening of application mails in the method for processing application mails for this application;

[0050] Figure 13 A flowchart of the fifth embodiment of the method for processing application declaration emails is provided;

[0051] Figure 14 This is a flowchart of the application expert model in the method for processing application emails;

[0052] Figure 15 This is a schematic diagram of the prompt for extracting declaration information in the method for processing the declaration email for this application;

[0053] Figure 16This is a schematic diagram of the JSON data set for the application information in the method for processing application emails;

[0054] Figure 17 This is a schematic diagram of extracting call record name and type information from emails in the method for processing emails for this application;

[0055] Figure 18 This is a processing sequence diagram for the method of processing emails for this application;

[0056] Figure 19 This is a diagram of the automatic processing architecture of international roaming data clearing declaration based on a large model in the method for processing the application declaration email;

[0057] Figure 20 This is a schematic diagram of the module structure of the device for processing declaration mails according to an embodiment of the present application;

[0058] Figure 21 This is a schematic diagram of the device structure of the hardware operating environment involved in the method for processing declaration emails in the embodiment of the present application.

[0059] The purpose, features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0060] It should be understood that the specific embodiments described herein are merely used to explain the technical solutions of the present application and are not intended to limit the present application.

[0061] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0062] International roaming, a key mobile communications application, allows users to use their mobile phones in other countries and regions. During international roaming, the Visited Place Mobile Network (VPMN) provides network services to the Home Public Mobile Network (HPMN), and the HPMN pays the VPMN accordingly. According to the Global System for Mobile Communications Association (GSMA), the VPMN generates a Transfer Account Procedure (TAP) file from the roaming user's communication records and periodically sends it to the HPMN. The HPMN then uses the TAP file to pay the VPMN and bill the user.

[0063] Reference Figure 1During the TAP file exchange process, the HPMN verifies the TAP file according to the GSMA standard. For problematic TAP files, it generates a Returned Account Procedure (RAP) file containing error information and sends the RAP file to the VPMN. RAP files are divided into four types: Missing (file skipping), Stop (no file uploaded for a long time), Fatal (file-level node error), and Severe (call record-level node error). According to GSMA regulations, if the VPMN has any objections to the returned RAP file, it must submit a complaint within a preset time period. After receiving the complaint, the HPMN must also process it within the preset time period and respond, either accepting or rejecting it. If rejecting, the HPMN must provide a reason for the rejection and follow up on the VPMN's subsequent response until the issue is resolved.

[0064] Currently, complaint processing is primarily done by business personnel who read complaint emails, determine the complaint type, analyze the problem and handle the complaint based on the complaint type, and then send a response to the other party via email. This cycle repeats until the problem is resolved.

[0065] The main solution of this embodiment is to select candidate claim emails based on the content keywords and email information of the roaming fee claim emails; determine the target claim email based on the context information of the candidate claim emails; and select a claim processing solution based on the key processing information in the target claim email, which includes the file name, claim type, and claim request. For ease of description, this embodiment is described below using the claim email processing device as the execution entity.

[0066] This application provides a solution that quickly filters candidate emails using content keywords, email information, and contextual information, reducing the workload of manual review and shortening response time. Based on structured information such as file name and application type, it automatically matches a pre-set solution library to improve the processing efficiency of application emails.

[0067] It should be noted that the execution subject of this embodiment may be a computing service device with data processing, network communication, and program execution capabilities, such as a tablet computer, personal computer, or mobile phone, or an electronic device capable of performing the aforementioned functions, or a device for processing mail declarations. This embodiment and the following embodiments will be described below using a device for processing mail declarations as an example.

[0068] Based on this, the present application embodiment provides a method for processing declaration mails, referring to Figure 2 , Figure 2 This is a flow chart of the first embodiment of the method for processing application declaration emails.

[0069] In this embodiment, the method for processing the declaration mail includes steps S10 to S30:

[0070] Step S10: Select candidate declaration emails based on the content keywords and email information of the roaming fee declaration emails.

[0071] Optionally, the keywords in the content of the roaming charge declaration email include roaming charges, roaming areas, error types, etc.

[0072] Optionally, the email information includes information such as the email subject, sender address, and specific email identifier of the declaration email.

[0073] In this embodiment, candidate emails are initially screened for those related to international roaming settlement applications. Emails are filtered by analyzing keywords and email information within the application. This ensures that the selected candidate emails are relevant to international roaming settlement applications, reducing complexity and effectively eliminating irrelevant data, reducing unnecessary data input, and thus improving processing efficiency and accuracy in subsequent steps.

[0074] Step S20: determining a target declaration email based on the context information of the candidate declaration emails.

[0075] Optionally, contextual information such as historical emails can be included to avoid isolating individual emails and identify duplicate claims or complex issues, such as multiple complaints about the same bill, improving problem location accuracy. Optionally, contextual information includes file name structure, bill type, RAP type, error type, and claim request type.

[0076] After selecting the candidate declaration emails, the context information of the candidate declaration emails is judged, and the candidate declaration emails are filtered twice to obtain the target declaration email. In this application, the target declaration email is the declaration email that needs to be processed and intervened.

[0077] Step S30: extract key processing information from the target declaration email, and select a solution for declaration processing based on the key processing information. The key processing information includes at least one of a file name, a declaration type, and a declaration request.

[0078] Extract key processing information from the target declaration email. Optionally, this key processing information includes the file name, declaration type, and declaration request. For example, the file name includes the TAP file name and the RAP file name, and the declaration type includes file number skipping, long period of no file upload, file-level node error, and call record-level node error.

[0079] Based on the key information of the target declaration email, the optimal declaration processing solution is selected. Optionally, the declaration processing solution includes information such as the file name, database table set, query conditions, and query order.

[0080] In this embodiment, candidate roaming fee claim emails are selected based on their content keywords and email information. A target claim email is determined based on the contextual information of the candidate emails. A claim processing solution is selected based on key processing information in the target email, including the file name, claim type, and claim request. This rapid screening of candidate emails based on content keywords, email information, and contextual information reduces the workload of manual review and shortens response time. Based on structured information such as the file name and claim type, a pre-set solution library is automatically matched, improving the efficiency of claim email processing.

[0081] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those in the above embodiment can be referred to the above introduction and will not be described in detail later. Figure 3 , after step S30, further comprising:

[0082] Step S40, generating an SQL statement to be executed based on the solution of the declaration processing;

[0083] Step S50: executing the SQL statement in a preset solution database and obtaining the execution result of the SQL statement, wherein the solution database includes standard solutions.

[0084] Optionally, update the declaration processing solution based on changes in operational business and scenarios to improve the needs of business scenarios.

[0085] Optionally, referring to the table below, different RAP types, error nodes, error codes, error types, and complaint requests correspond to different solution packages, that is, different complaint processing solutions.

[0086]

[0087]

[0088] In this embodiment, the full solution information for the application is translated into SQL (Structured Query Language) statements that can be executed in the solution database. Alternatively, a text2SQL module, also known as a Text-to-SQL model, can be used to translate the full solution information into SQL statements that can be executed in the database.

[0089] It should be noted that when choosing a Text-to-SQL model, the SQLCoder-7b model was selected. First, the SQLCoder-7b model has been specifically optimized and trained to perform better for Text-to-SQL tasks. This specialized capability enables higher accuracy and efficiency when generating complex SQL queries. Second, the SQLCoder-7b model may be more adaptable to various SQL query patterns and complex database architectures, which is particularly important for application scenarios requiring multi-table join queries and complex data extraction. Furthermore, the SQLCoder-7b model may provide faster query generation speeds in terms of performance, which is critical in scenarios requiring large-scale SQL query generation. Finally, from a cost-effectiveness perspective, the SQLCoder-7b model may have a greater advantage in call fees, especially in large-scale applications, which can effectively reduce overall costs.

[0090] The SQLCoder model categorizes each generated query into six categories: date, group_by, order_by, ratio, join, and where. Ranked second in overall performance, it consumes fewer resources than the top-ranked SQLCoder-7b model, making it suitable for resource-constrained hardware and offering faster response times. Furthermore, the queries in this system are relatively simple and low-complexity, making the SQLCoder-7b model fully capable.

[0091] After receiving the key processing information and solution package, the Text2SQL module uses its text-to-SQL capability to convert the key processing information and solution package into SQL that can be executed in the database. For example, the solution package is Figure 4 As shown, a solution package is obtained based on the declaration information, and a Severe RAP is identified. The error type of the declaration email is a rate error, and a text format information package is output to the Text2SQL module.

[0092] For example, the translated SQL statement is as follows Figure 5 As shown, for the execution of subsequent SQL statements, after receiving the information package, the Text2SQL module uses its large translation model capability to understand the key processing information and solution package of the information package and write SQL, and then sends the SQL to the declaration automatic processing module.

[0093] In this embodiment, the technical solution directly converts matched solutions into executable SQL operations, such as querying standard solutions, updating work order status, and invoking repair scripts. This enables end-to-end automation from problem analysis to implementation, reducing manual intervention. Automatically generating and executing SQL eliminates the delays associated with manually writing and reviewing SQL statements, enabling responses within seconds, especially for high-frequency, simple issues like rate cancellations.

[0094] Based on the first or second embodiment of the present application, in the third embodiment of the present application, the same or similar contents as those in the above embodiment can be referred to the above introduction and will not be described in detail later. Figure 6 , after step S50, further comprising:

[0095] Step S60: generating a declaration processing result according to the execution result, wherein the declaration processing result includes at least one of document processing information, claim argumentation proof, conclusion, and additional explanation;

[0096] Step S70: Send the declaration processing result to the operator of the visited location.

[0097] The SQL execution results for the extracted information list are returned in JSON format. Based on the results, conclusions are drawn and matched with the reply content. The reply content and information list are output in JSON format as the report processing results. The report processing results are sent to the email system of the operator in the visited location. Refer to the following table for SQL execution result analysis and email reply content matching rules:

[0098]

[0099]

[0100]

[0101] Figure 7 The execution result is in JSON format, including RAP file type, TAP type, and error type. Figure 8 Generates the email reply body logic for the execution results based on JSON format, which includes parameters: declaration type, reply content corresponding to the question, and call record name.

[0102] Optionally, refer to Figure 9Sending emails using IMAP (Internet Message Access Protocol) typically requires sending emails using SMTP (Simple Mail Transfer Protocol). Based on the execution results, select the email content, which includes the email subject, recipient, claim type, processing results, and call record details.

[0103] In this embodiment, the technical solution pushes processing results, such as document processing status and claim verification, directly to the carrier's system at the destination, eliminating the inefficiencies of manual email or phone calls and shortening the cross-network problem-solving cycle. Structured results unify the data interface between the two parties, reducing misunderstandings caused by differences in presentation.

[0104] Based on any one of the first to third embodiments of the present application, in the fourth embodiment of the present application, the same or similar contents as those in the above embodiments can be referred to the above introduction and will not be described in detail later. Figure 10 , step S10 includes:

[0105] Step S11, establishing a connection with the mail server, and inspecting the mailbox of the mail server according to the mailbox name and inspection time;

[0106] Step S12: During the inspection process, the declaration mails in different encoding formats are decoded and the contents of the historical reply mails to the declaration mails are obtained;

[0107] Step S13, determining the content keywords and the email information based on the decoded declaration email and the parsed content of the historical reply email;

[0108] Step S14: Select the candidate declaration email based on the content keywords and the email information.

[0109] It establishes a connection with the mail server through the IMAP protocol, allowing users to select the mailbox name that needs to be inspected according to custom options, and then performs a reverse traversal to inspect the emails in the mailbox and preliminarily screen out the target emails.

[0110] Reference Figure 11A fault tolerance mechanism has been added to address missed inspections due to time discrepancies between email servers or other factors. The fault tolerance mechanism sets an upper inspection threshold. During the inspection process, the inspection time is first determined to be within the preset time range. If it is outside this range, the fault tolerance mechanism determines whether the threshold has been exceeded. If the threshold is exceeded, the process is terminated and the inspection log is compiled and recorded. If it is within the threshold, the inspection continues to ensure inspection accuracy. When an anomaly is detected, the exception information, such as the email ID (identifier), is recorded, and the content of the abnormal email is compiled. For different types of email anomalies, emails with different encoding formats are decoded and the content of historical reply emails is parsed. Automatic decoding of emails of different formats addresses garbled characters caused by encoding discrepancies, such as when users send emails using non-standard clients, ensuring that the email body and attachments are fully extracted. Analysis of historical reply emails, such as multiple communication records in an email thread, can prevent misjudgments due to incomplete information in a single email, such as user supplementary comments in subsequent emails.

[0111] This multi-layered fault-tolerance mechanism not only enhances system stability but also dynamically adjusts for out-of-range or abnormal data based on actual conditions, preventing minor errors from disrupting the entire inspection process. This technical solution is highly scalable, adapting to various email formats and exception handling scenarios, ensuring comprehensive and accurate mailbox inspections.

[0112] Reference Figure 12 Based on the characteristics of international roaming declaration emails, a keyword filtering knowledge base has been added. Based on the information in the filtering base, multiple criteria such as the email subject, sender, recipient, CC, body, and attachments can be filtered to quickly locate specific emails. For example, emails containing certain characters in the title or body will be filtered, and emails received for the first time will continue to be processed. At the same time, users can perform inspections on different mailbox folders according to their needs, including the inbox, outbox, trash, drafts, and deleted items, improving the adaptability and flexibility of inspections.

[0113] In this embodiment, the technical solution automatically triggers scans based on mailbox name and inspection time, eliminating manual spot checks and ensuring that all reported emails are included, such as those that are expired or archived. Cross-period data integration supports batch processing by time range, adapting to different reporting scenarios, such as the need to review historical emails in the event of a sudden roaming failure.

[0114] Based on any one of the first to fourth embodiments of the present application, in the fifth embodiment of the present application, the same or similar contents as those in the above embodiments can be referred to the above introduction and will not be described in detail later. Figure 13 , step S30 further includes:

[0115] Step S31, constructing prompt words of the declaration expert model, wherein the prompt words include at least one of a target, a context, an instruction, an expected output, and a constraint;

[0116] Step S32: constructing a roaming claim knowledge base of the claim expert model; the knowledge base includes at least one of proper noun descriptions, customer and roaming partner code sets, and call record types and their descriptions;

[0117] Step S33: extracting key processing information of the target declaration email based on the declaration expert model, the prompt words and the roaming declaration knowledge base.

[0118] The model combines Qwen-7B with prompts designed specifically for international roaming services. Through corpus collection, Prompts is continuously optimized to continuously improve its accuracy so that it can better serve international roaming declaration services.

[0119] Compared with open source models of the same size, Qwen-7B (Tongyi Qianwen) has demonstrated outstanding performance on multiple evaluation data sets. Whether in the fields of natural language understanding and generation, mathematical problem solving, or code generation, Qwen-7B has demonstrated strong potential. At the same time, the use of a word segmenter with a larger vocabulary has advantages in word segmentation efficiency and supports a context length of up to 8K, which can meet the input of context information for multiple rounds of email replies. More importantly, specific optimizations have been made in plug-in calls. The model can effectively call plug-ins and upgrade to agents, which can better judge record types and dispatch corresponding processing tools for processing.

[0120] Reference Figure 14 , Figure 14 This is a flowchart of the data processing by the claim expert model. Specifically, by building an international roaming claim knowledge base and a prompt word project, combined with the Qwen-7B model, accurate information extraction from claim emails is achieved. The knowledge base covers terminology related to international roaming settlement claims, call record types, RAP types, and error types. The prompt word project uses targeted, context-rich prompt words to guide the model in extracting TAP and RAP file names, RAP types, error types, and claim content from emails. Ultimately, it outputs structured claim information in JSON (JavaScript Object Notation), improving processing efficiency and accuracy.

[0121] Build a comprehensive knowledge base for international roaming claims by collecting information related to international roaming settlement claims, including specialized terminology, claim email formats, and business processes. This includes descriptions of specific terms, customer and roaming partner code sets, call record types and their descriptions (voice caller: MOC; voice callee: MTC; SMS caller: SMS-MO; SMS callee: SMS-MT; data traffic: GPRS; call forwarding: SSE, etc.), RAP types (file skip: Missing; file delay: Stop; file-level error: Fatal; call record-level error: Severe), and error types (error node + error code, such as charge error: charge 200; call record expiration error: callEventStartTimeStamp 261).

[0122] Prompt words are text prompts that users enter when interacting with large models. They can guide the model to generate specific outputs and help users obtain the required information more efficiently. For example, prompt words such as Figure 15 As shown, among them, file name, RAP type, error type, appeal and request, etc. The following points need to be considered during the prompt word writing process: Goal: Make clear the goal that the prompt word wants to achieve, such as extracting important information from the email, including file name, RAP type, error type, appeal and request, etc. Context: Provide sufficient background information to help the big model understand the context and environment of the prompt word. For example, the composition of the file name, call record type, RAP type type, error type type and appeal type, etc. Instructions: Clear and specific instructions to guide the big model on how to respond to the prompt word. For example, extract the RAP file name for the appeal, extract the error type for the appeal, etc. Expected output: Clarify the answer expected from the big model and the format of the return. Constraints: Define the restrictions that the big model needs to abide by when responding to the prompt word, such as the word count limit.

[0123] In the key information extraction example, we obtain the body of the application email that has been screened twice in the early stage, and then generate the prompt word (prompt) for the complete application email information extraction by combining the prompt generation method. Then, we call the Qwen-7B model based on the knowledge base to obtain the JSON data of the application information extraction. For example, the input parameter includes the email body. The output parameter includes the application information JSON dataset, for example Figure 16 The JSON dataset shown includes the file name, RAP type, error type, and processing results.

[0124] In an optional embodiment, error type information of historical declaration emails is obtained; relevant information of the historical declaration emails is obtained, and the relevant information includes at least one of the body text, the error type complaint description, and the declaration request; based on the error type information and the relevant information, the historical declaration emails are selected as training texts, and the declaration expert model is iteratively trained based on the training texts.

[0125] To improve the information extraction and data analysis capabilities of the claim expert, a dedicated corpus collection module was built. This module is responsible for collecting and organizing corpus data related to roaming claims from various tasks. For example, it collects and organizes information about the types of errors reported to improve the model's understanding of these types of errors. It also collects email body text, along with complaint descriptions and request information related to the error type, and categorizes the emails by error type to improve the model's understanding of the email content, thereby iteratively training the claim expert model. Below are examples of the corpus collected for different tasks.

[0126] Collect and organize relevant information about international roaming settlement declarations, refer to the table below, to improve the model's understanding of declared call record information.

[0127]

[0128]

[0129] Reference Figure 17 , collect the email body text, and mark the RAP file name, TAP file name, RAP type, call error type, and complaint request in JSON format, and classify them according to the call error type. The complaints of the same type of errors are summarized and analyzed to improve the understanding ability of the large model in scenarios with different complaint descriptions of the same error type, thereby improving the model's ability to extract key complaint information based on the email content, and also enriching the knowledge base.

[0130] Training and evaluation combine multiple datasets to ensure the model accurately performs its tasks and produces satisfactory output. The training process includes data preprocessing, cleaning, and organization to ensure the data format is uniform and meets training requirements. These steps enable the design of an effective module that accurately processes and outputs declaration emails. This comprehensive training and evaluation approach significantly improves the model's overall performance and applicability.

[0131] In this embodiment, the technical solution uses prompts such as the target and context to define the model's analysis direction, such as identifying roaming rate dispute complaints. This avoids misjudgments often caused by general models due to task ambiguity, such as misclassifying network failures as rate issues. Utilizing proper noun descriptions and customer code sets, the system accurately parses technical terms and abbreviations in emails, reducing semantic ambiguity. The system also combines historical email context with customer codes in the knowledge base to identify complex complaints.

[0132] In one embodiment, an automatic inspection module for declaration mails is constructed based on the content and keyword characteristics of the declaration mails, and the declaration mails are screened out from the inbox. The declaration mails that have been preliminarily screened out are then input into the declaration processing large model, and its declaration expert module is used to identify the mails that need to be intervened and processed, and to extract the key information of the declaration processing for selecting the declaration processing solution. The declaration expert module outputs the selected solution to the text2SQL module, and uses its natural language and SQL conversion capabilities to convert the solution into SQL statements that can be executed in the database (such as specific queries and update statements), and outputs the SQL to the declaration automatic processing module for execution. Finally, the automatic declaration outputs the execution result to the mailbox system module, and the processing result is replied to the VPMN in the form of an email through the mail system. At the same time, the forecast collection module organizes the data into forecasts and feeds them back to the large model for training, continuously optimizes and improves the model, forming a virtuous circle, and finally trains a declaration processing model with expert level in the field of international roaming clearing.

[0133] For specific processing sequence diagram, refer to Figure 18 The user initiates a request and calls the main program, which then passes the request to the declaration email inspection module. During this step, the system filters emails by analyzing the email subject, sender address, email content keywords, and specific email identifiers. It then selects declaration emails and outputs them to the declaration expert model. This ensures that only emails related to international roaming settlement declarations are output to the declaration expert module, reducing the number of scenarios the declaration expert needs to consider and reducing complexity. It also effectively eliminates irrelevant data, reducing unnecessary data input, and improving the efficiency and accuracy of subsequent processing steps.

[0134] Specific system architecture diagram, such as Figure 19Figure 2 shows the architecture diagram for automated processing of international roaming data settlement claims based on a large model. After receiving a claim email, the claim expert model platform determines the email context and performs secondary filtering to further identify claims requiring intervention. It then extracts key information from these emails, including the file name (TAP and RAP file names), claim type (Missing, Stop, Fatal, Severe), and claim request. Based on this information, it selects the optimal solution, including the file name, database table set, query conditions, and query order, and sends the solution package to the text2SQL module.

[0135] After receiving the declaration solution information package, the text2SQL module translates the full solution information into SQL statements that can be executed in the database, and sends the SQL statements to the declaration automatic processing module.

[0136] After receiving the SQL statement, the automatic declaration processing module executes it in sequence in the corresponding database and returns the execution result to the declaration expert module.

[0137] After receiving the SQL execution results, the declaration expert module organizes the results into a declaration reply email information package, including file processing information, claim justification proof, conclusion and any necessary additional instructions, and transmits such information to the mailbox system through the interface.

[0138] After receiving the reply email information package, the mailbox system generates an email reply VPMN to ensure that the declaration is processed and responded to in a timely manner.

[0139] The corpus collection module actively collects evaluated evaluation forms, screens high-quality data sets, and performs iterative model training to improve the model's capabilities.

[0140] It should be noted that the above examples are only used to understand this application and do not constitute a limitation on the method of processing the declaration emails of this application. More simple transformations based on this technical concept are all within the scope of protection of this application.

[0141] This application also provides a processing device for declaration mail, please refer to Figure 20 , the processing device for the declaration mail includes:

[0142] The selection module 10 is used to select candidate declaration emails based on the content keywords and email information of the roaming fee declaration emails;

[0143] A determination module 20, configured to determine a target declaration email based on the context information of the candidate declaration emails;

[0144] The processing module 30 is used to extract key processing information from the target declaration email and select a solution for declaration processing based on the key processing information. The key processing information includes at least one of the file name, declaration type and declaration request.

[0145] Optionally, after the step of selecting a solution for the declaration processing according to the key processing information, the method further includes:

[0146] Generate SQL statements to be executed based on the solution of the declaration processing;

[0147] The SQL statement is executed in a preset solution database, and an execution result of the SQL statement is obtained. The solution database includes standard solutions.

[0148] Optionally, after the step of obtaining the execution result of the SQL statement, the method further includes:

[0149] generating a declaration processing result according to the execution result, the declaration processing result including at least one of document processing information, claim argumentation proof, conclusion, and additional explanation;

[0150] The report processing result is sent to the operator of the visited location.

[0151] Optionally, the step of selecting candidate declaration emails based on content keywords and email information of the roaming fee declaration email includes:

[0152] Establish a connection with the mail server and inspect the mailboxes of the mail server according to the mailbox name and inspection time;

[0153] During the inspection process, the declaration emails in different encoding formats are decoded and the contents of the historical reply emails to the declaration emails are obtained;

[0154] Determine the content keywords and the email information based on the decoded declaration email and the parsed content of the historical reply email;

[0155] Based on the content keywords and the email information, the candidate declaration email is selected.

[0156] Optionally, the step of extracting key processing information from the target declaration email further includes:

[0157] Constructing prompt words for a declaration expert model, wherein the prompt words include at least one of a goal, a context, an instruction, an expected output, and a constraint;

[0158] Constructing a roaming claim knowledge base of the claim expert model; the knowledge base includes at least one of proper noun descriptions, customer and roaming partner code sets, and call record types and their descriptions;

[0159] Based on the declaration expert model, the prompt words and the roaming declaration knowledge base, key information for processing the target declaration email is extracted.

[0160] Optionally, the method further includes:

[0161] Get error type information of historical declaration emails;

[0162] Obtaining relevant information of the historical complaint email, the relevant information including at least one of the body text, a complaint description of the error type, and a complaint request;

[0163] Based on the error type information and the relevant information, historical declaration emails are selected as training texts, and the declaration expert model is iteratively trained based on the training texts.

[0164] The declaration mail processing device provided in this application utilizes the declaration mail processing method of the aforementioned embodiment to address the technical issue of inefficient declaration processing. Compared to the prior art, the beneficial effects of the declaration mail processing device provided in this application are identical to those of the declaration mail processing method of the aforementioned embodiment. Other technical features of the declaration mail processing device are identical to those disclosed in the aforementioned embodiment and are not further elaborated here.

[0165] The present application provides a device for processing declaration mails, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the method for processing declaration mails in the above-mentioned embodiment one.

[0166] Reference below Figure 21 , which shows a schematic diagram of the structure of a declaration mail processing device suitable for implementing the embodiments of the present application. The declaration mail processing device in the embodiments of the present application may include, but is not limited to, mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, personal digital assistants (PDAs), tablet computers (PADs), portable multimedia players (PMPs), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., as well as fixed terminals such as digital TVs and desktop computers. Figure 21The processing device for declaration mail shown is merely an example and should not impose any limitations on the functions and scope of use of the embodiments of the present application.

[0167] like Figure 21 As shown, the processing device for declaring mail may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. Various programs and data required for the operation of the processing device for declaring mail are also stored in RAM 1004. The processing device 1001, ROM 1002, and RAM 1004 are connected to each other via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touch screen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, a magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 can allow the declaration mail processing device to communicate wirelessly or wired with other devices to exchange data. Although the figure shows a declaration mail processing device with various systems, it should be understood that it is not required to implement or have all of the systems shown. More or fewer systems may be implemented or have instead.

[0168] In particular, according to the embodiments disclosed in the present application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, the embodiments disclosed in the present application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program comprising program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0169] The mail processing device provided in this application utilizes the mail processing method described in the aforementioned embodiment to address the technical issue of inefficient mail processing. Compared to the prior art, the mail processing device provided in this application achieves the same beneficial effects as the mail processing method described in the aforementioned embodiment. Other technical features of the mail processing device are the same as those disclosed in the aforementioned embodiment and are not further elaborated upon here.

[0170] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0171] The above description is merely a specific embodiment of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of this application. Therefore, the scope of protection of this application should be based on the scope of protection of the claims.

[0172] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the method for processing declaration mails in the above-mentioned embodiment.

[0173] The computer-readable storage medium provided in this application can be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, systems or devices, or any combination thereof. More specific examples of computer-readable storage media can include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM, Erasable Programmable Read Only Memory or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM, CD-Read Only Memory), an optical storage device, a magnetic storage device, or any suitable combination thereof. In this embodiment, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium can be transmitted using any appropriate medium, including but not limited to: wires, optical cables, radio frequency (RF, Radio Frequency), etc., or any suitable combination thereof.

[0174] The above-mentioned computer-readable storage medium may be included in the processing device for declaring mails; or it may exist independently without being assembled into the processing device for declaring mails.

[0175] The computer-readable storage medium carries one or more programs. When executed by a device processing submitted emails, the device enables the device to: quickly screen candidate emails based on content keywords, email information, and contextual information, reducing the workload of manual review and shortening response time. It also automatically matches a library of pre-set solutions based on structured information such as file name and submission type, improving the efficiency of submitting emails.

[0176] The computer program code for performing the operations of the present application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer can be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or can be connected to an external computer (e.g., using an Internet service provider to connect via the Internet).

[0177] The flow charts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flow chart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flow chart, and the combination of the boxes in the block diagram and / or flow chart can be implemented by a dedicated hardware-based system that performs the specified function or operation, or can be implemented by a combination of dedicated hardware and computer instructions.

[0178] The modules described in the embodiments of the present application may be implemented in software or hardware, wherein the name of a module does not necessarily limit the unit itself.

[0179] The computer-readable storage medium provided in this application stores computer-readable program instructions (i.e., a computer program) for executing the aforementioned method for processing declaration mail, thereby resolving the technical issue of inefficient declaration processing. Compared to the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the method for processing declaration mail provided in the aforementioned embodiment, and are not further elaborated here.

[0180] The present application also provides a computer program product, including a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for processing declaration mails.

[0181] The computer program product provided in this application can solve the technical problem of low efficiency in declaration processing. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the method for processing declaration mails provided in the above embodiment, and will not be repeated here.

[0182] The above description is only part of the embodiments of the present application and does not limit the patent scope of the present application. All equivalent structural transformations made by using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect application in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A method for processing declaration mails, characterized in that: The method comprises: Select candidate declaration emails based on the keywords and email information in the roaming fee declaration emails; Determining a target declaration email based on the context information of the candidate declaration emails; Extract key processing information from the target declaration email, and select a solution for declaration processing based on the key processing information, wherein the key processing information includes at least one of a file name, a declaration type, and a declaration request.

2. The method according to claim 1, wherein After the step of selecting a solution for the declaration processing according to the key processing information, the method further includes: Generate SQL statements to be executed based on the solution of the declaration processing; The SQL statement is executed in a preset solution database, and an execution result of the SQL statement is obtained. The solution database includes standard solutions.

3. The method according to claim 2, wherein After the step of obtaining the execution result of the SQL statement, the method further includes: generating a declaration processing result according to the execution result, the declaration processing result including at least one of document processing information, claim argumentation proof, conclusion, and additional explanation; The report processing result is sent to the operator of the visited location.

4. The method according to claim 1, wherein The step of selecting candidate declaration emails based on the content keywords and email information of the roaming fee declaration email comprises: Establish a connection with the mail server and inspect the mailboxes of the mail server according to the mailbox name and inspection time; During the inspection process, the declaration emails in different encoding formats are decoded and the contents of the historical reply emails to the declaration emails are obtained; Determine the content keywords and the email information based on the decoded declaration email and the parsed content of the historical reply email; Based on the content keywords and the email information, the candidate declaration email is selected.

5. The method according to claim 1, wherein The step of extracting key processing information from the target declaration email also includes: Constructing prompt words for a declaration expert model, wherein the prompt words include at least one of a goal, a context, an instruction, an expected output, and a constraint; Constructing a roaming claim knowledge base of the claim expert model; the knowledge base includes at least one of proper noun descriptions, customer and roaming partner code sets, and call record types and their descriptions; Based on the declaration expert model, the prompt words and the roaming declaration knowledge base, key information for processing the target declaration email is extracted.

6. The method according to claim 5, wherein The method further comprises: Get error type information of historical declaration emails; Obtaining relevant information of the historical complaint email, the relevant information including at least one of the body text, a complaint description of the error type, and a complaint request; Based on the error type information and the relevant information, historical declaration emails are selected as training texts, and the declaration expert model is iteratively trained based on the training texts.

7. A device for processing declaration mails, characterized in that: The device comprises: A selection module is used to select candidate declaration emails based on content keywords and email information of the roaming fee declaration emails; A determination module, configured to determine a target declaration email based on the context information of the candidate declaration emails; The processing module is used to extract key processing information from the target declaration email and select a solution for declaration processing based on the key processing information. The key processing information includes at least one of the file name, declaration type and declaration request.

8. A device for processing declaration mails, characterized in that: The device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the method for processing declaration mails as described in any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the method for processing declaration mails as described in any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product includes a computer program, and when the computer program is executed by a processor, the steps of the method for processing declaration mails as described in any one of claims 1 to 6 are implemented.

Images