Business processing method and device based on large model, equipment and medium

Abstract

The invention relates to a business processing method and device based on a large model, equipment and a storage medium. The method comprises the steps of analyzing input content of a user through a large model, obtaining intention information and entity information of the input content, querying a knowledge graph based on the intention information and the entity information, determining a matching result with the input content and current information of the user, and determining the input content according to the matching result and the current information of the user. And generating a processing flow list and pre-filling information by using the service engine, and generating guidance information according to the processing flow list and the pre-filling information and feeding back the guidance information to the user. User input is accurately analyzed through a large model, intention recognition and entity extraction accuracy is improved, a foundation is laid for personalized business processing, dynamic association and real-time updating of policy information are achieved by querying a knowledge graph, business guidance accuracy and timeliness are ensured, a business engine automatically generates a process list and pre-filled information, and the business efficiency is improved. Repeated input and manual query of the user are reduced, and the business handling efficiency is improved.

Description

Technical Field

[0001] This application relates to the field of large model technology, and in particular to a business processing method, apparatus, device and storage medium based on large models. Background Technology

[0002] Currently, many existing digital systems in universities are information silos, with data fragmented between systems such as academic affairs, finance, and student affairs. When faculty and students handle cross-departmental matters, they must repeatedly submit information, resulting in cumbersome and error-prone operations. Furthermore, the systems lack sufficient natural language understanding capabilities, and applications such as intelligent customer service have low accuracy in extracting the intent and specific information of faculty and students' inquiries. Most inquiries still require human intervention, failing to achieve efficient automated responses. For example, when faculty and students inquire about scholarship applications, the system cannot accurately identify their specific intent regarding application materials or procedures, nor can it accurately extract key information such as the scholarship name, leading to low processing efficiency.

[0003] Therefore, how to improve the efficiency of business processing has become a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0004] In view of the above, this application provides a business processing method, apparatus, device and storage medium based on a large model, the purpose of which is to solve the above-mentioned technical problems.

[0005] Firstly, this application provides a business processing method based on a large model, the method comprising:

[0006] By analyzing user input through a large model, the intent and entity information of the input content can be obtained.

[0007] Based on the intent information and the entity information, a knowledge graph is queried to determine the matching result with the input content and the user's current information;

[0008] Based on the matching results and the user's current information, the business engine generates a processing flow list and pre-filled information.

[0009] Based on the processing flow list and the pre-filled information, guidance information is generated and fed back to the user.

[0010] Secondly, this application provides a large-scale model-based service processing apparatus, which includes:

[0011] Analysis module: Used to analyze user input through a large model to obtain intent and entity information of the input content;

[0012] Query module: used to query the knowledge graph based on the intent information and the entity information, and determine the matching result with the input content and the user's current information;

[0013] First generation module: used to generate a processing flow list and pre-filled information based on the matching result and the user's current information using the business engine;

[0014] The second generation module is used to generate guidance information and send it back to the user based on the processing flow list and the pre-filled information.

[0015] Thirdly, this application provides an electronic device, including a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0016] Memory, used to store computer programs;

[0017] When a processor executes a program stored in memory, it implements the steps of the business processing method based on a large model as described in any embodiment of the first aspect.

[0018] Fourthly, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps of the large-model-based business processing method as described in any embodiment of the first aspect.

[0019] The technical solutions provided in this application have the following advantages compared with the prior art:

[0020] This application analyzes user input using a large-scale model to obtain intent and entity information. Based on this intent and entity information, it queries a knowledge graph to determine matching results with the input and the user's current information. According to the matching results and the user's current information, a business engine generates a processing flow list and pre-filled information. Based on the processing flow list and pre-filled information, it generates guidance information and sends it back to the user. The large-scale model accurately analyzes user input, improving the accuracy of intent recognition and entity extraction, laying the foundation for personalized business processing. Querying the knowledge graph enables dynamic association and real-time updates of policy information, ensuring the accuracy and timeliness of business guidance. The business engine automatically generates process lists and pre-filled information, reducing repetitive user input and manual queries, and improving business processing efficiency. 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] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0023] Figure 1 This is a flowchart illustrating a preferred embodiment of the business processing method based on a large model in this application;

[0024] Figure 2 This is a schematic diagram of a preferred embodiment of the business processing device based on a large model in this application;

[0025] Figure 3 This is a schematic diagram of a preferred embodiment of the electronic device of this application;

[0026] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application. All other embodiments obtained by those skilled in the art based on the embodiments in this application without inventive effort are within the scope of protection of this application.

[0028] It should be noted that the use of terms such as "first" and "second" in this application is for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include at least one of those features. Furthermore, the technical solutions of the various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, such a combination of technical solutions should be considered non-existent and not within the scope of protection claimed in this application.

[0029] Reference Figure 1 The diagram shown is a flowchart illustrating an embodiment of the business processing method based on a large model according to this application. The method is executed by an electronic device, which can be implemented by a software system and / or a hardware system. The business processing method based on a large model includes:

[0030] Step S10: Analyze the user's input content using a large model to obtain the intent information and entity information of the input content;

[0031] Step S20: Based on the intent information and the entity information, query the knowledge graph to determine the matching result with the input content and the user's current information;

[0032] Step S30: Based on the matching result and the user's current information, use the business engine to generate a processing flow list and pre-filled information;

[0033] Step S40: Based on the processing flow list and the pre-filled information, generate guidance information and send it back to the user.

[0034] In university business processing scenarios, users (e.g., faculty and students) can use natural language to express their consultation needs in the university's intelligent service system, such as inquiring about scholarship applications, grade inquiries, and activity registrations. The user's natural language expression contains their specific intent (e.g., requesting a service or information) and key entity information (e.g., service name, timeframe, and personal identification details). Traditional systems have limited processing capabilities for this type of natural language, often leading to misunderstandings and low efficiency in business processing. However, by leveraging the natural language understanding capabilities of large-scale models, user input can be accurately parsed, key information extracted, and a foundation laid for accurate subsequent business processing.

[0035] Specifically, users can receive business inquiries or requests for services through the front-end interactive interface of the university's intelligent service system (such as mobile applications, web-based dialog boxes, etc.) in the form of voice or text. The system converts the input content into a unified text format and inputs the converted text into a large model. This large model has been pre-trained on a massive amount of university business-related corpus (including but not limited to policy documents, common business consultation dialogues, business processing documents, etc.) and is able to understand the specific semantics and contextual relationships in the university business field.

[0036] The large-scale model performs semantic analysis on the input text based on its internal attention mechanism and deep learning algorithms. On one hand, it identifies the user's overall intent, such as "scholarship application consultation" or "grade review application." On the other hand, it extracts key entity information from the text, such as the scholarship name, application year, student ID, and name. Taking "graduate scholarship application" as an example, if a user enters "I want to apply for this year's graduate academic scholarship, what materials do I need to prepare?", the large-scale model will identify the intent as "scholarship application materials consultation," and extract entity information including "graduate academic scholarship" (business name entity), "this year" (time point entity), and personal entities such as student ID and name associated with the user's identity information (obtained through front-end login status). Compared to traditional keyword matching or simple template-based natural language processing methods, the analysis based on the large-scale model can more accurately understand the user's complex natural language expressions, enabling a smoother business processing flow.

[0037] Because the professional knowledge of universities is scattered across documents, regulations, and historical business data in various departments, traditional information systems struggle to integrate and utilize this fragmented knowledge. By constructing a dynamically updated knowledge graph of rules and regulations, various types of professional knowledge can be linked together to form a structured knowledge network. By querying and matching user input intent and entity information with the knowledge graph, the most relevant professional knowledge to the user's needs can be accurately located. Simultaneously, by combining the user's current information (such as identity / role, existing business processing status, etc.), a precise basis is provided for generating subsequent processing flows tailored to the user's actual situation, avoiding feedback that does not conform to the user's current status. The knowledge graph construction process is as follows:

[0038] Data is collected periodically (e.g., every 5 minutes) from various university business systems (academic affairs system, financial system, student affairs system, etc.) and policy document databases. The collected data is cleaned and categorized to extract business entities (such as various scholarship programs, course names, activity names, etc.), relationships (such as the correlation between scholarship application requirements and course grades, and the correlation between activity participation and user permissions), and attributes (such as scholarship amount, application deadline, course credits, etc.). A graph database is used to store business entities, relationships, and attributes, constructing a knowledge graph. Through pre-defined semantic parsing algorithms and data fusion rules, new policy document content and changes in business system data are updated to the knowledge graph in real time, ensuring the timeliness and accuracy of the knowledge graph. For example, when the financial system updates scholarship distribution standards, the attribute information of the corresponding scholarship entity in the knowledge graph is updated synchronously.

[0039] The system obtains user intent information (such as "scholarship application material consultation") and entity information (such as scholarship name, time frame, etc.) from the large model analysis module. Using the intent information as the primary direction of the query, it retrieves relevant business knowledge nodes and their associated paths from the knowledge graph. Simultaneously, entity information is used as a filtering condition to further narrow the query scope and locate precisely matching business knowledge. For example, in the scenario of "postgraduate academic scholarship application material consultation," the system first finds all business knowledge nodes related to "scholarship application" in the knowledge graph, and then filters out the corresponding application material requirements nodes for specific scholarship programs based on the entity information "postgraduate academic scholarship."

[0040] Next, the system retrieves the user's current information from the university's unified identity authentication system and various business systems, such as identity and role, existing business processing records (e.g., whether the initial review has been passed, whether there are any outstanding fees), and the user's department and major. This current user information is then cross-referenced with the retrieved business knowledge to determine the final matching result. For example, if the user is a graduate student belonging to a specific department, the system will combine the specific application requirements for that scholarship for graduate students in that department from the knowledge graph to determine the matching result.

[0041] Because university business processing procedures typically involve multiple stages and departments, and the procedures vary significantly between different services, users need to understand and organize these procedures themselves, filling in a large amount of repetitive basic information. This not only increases the burden on users but also easily leads to delays due to errors. Utilizing a business engine to generate processing flow lists and pre-filled information based on matching results and user information enables automated customization of business processes and pre-filling of information, thereby improving the accuracy and efficiency of business processing.

[0042] Specifically, a process template library can be pre-built. Based on the handling standards and processes of various common business operations in universities, different process templates for different business operations can be pre-configured in the business engine. Each process template defines in detail the order of business processing steps, the forms involved, the list of required materials, and the approval links of each department. For example, for the "graduate scholarship application" business, the process template includes student application, supervisor recommendation, departmental preliminary review, school review, and result announcement, as well as the corresponding application form, recommendation form, transcript, and other material requirements for each step. A data mapping relationship is established between business knowledge nodes in the knowledge graph and process templates in the business engine. Simultaneously, the rules for filling user information with form fields in the process template are set to ensure that user information can be accurately pre-filled into the corresponding forms. For example, the scholarship application condition nodes in the knowledge graph are mapped to the application form fields in the process template. When a user meets certain conditions, the corresponding options are automatically checked in the form.

[0043] The system receives matching results (including business knowledge, process requirements, etc.) from the knowledge graph query module and the user's current information from the business engine. Based on the matching results, it retrieves the corresponding business process template from the process template library. The process template is then customized in conjunction with the user's current information. For example, if the user has already passed certain prerequisite steps (such as passing a tutor recommendation), that step is skipped or marked as completed in the generated processing flow list.

[0044] Based on data mapping and population rules, the system automatically populates the user's current information into the forms involved in the business process. For example, in the "Graduate Scholarship Application" form, it automatically fills in the user's basic information such as student ID, name, department, and major, and pre-fills some grade fields based on the user's existing academic performance data, reducing the need for manual input by the user. Through the automated processing of the business engine, business processing efficiency can be improved.

[0045] When handling university-related matters, users need intuitive guidance to understand the entire process and specific operational requirements. Besides providing users with a process checklist and pre-filled forms, this information needs to be integrated into easily understandable guidance to ensure users can complete the process accordingly. Specifically, the process checklist should be converted into concise and clear text descriptions in logical order, including the steps, precautions, and estimated processing time for each stage. Key fields in pre-filled forms should be highlighted to remind users to pay close attention and verify them. For example, the guidance for "Graduate Scholarship Application" should state, "Step 1: Log in to the scholarship application system and fill out the application form (your basic information has been pre-filled and can be modified if necessary), paying special attention to verifying the accuracy of the grade fields; Step 2: After submitting the application, wait for the supervisor's recommendation...". This guidance information should be pushed to users through the front-end interface of the university's intelligent service system (mobile application push notifications, web pop-ups, etc.). Users can also view and download the guidance content at any time within the system for convenient reference in different scenarios. The generated guidance information meets users' various information needs during the process, effectively improving the success rate of business processing and user satisfaction.

[0046] In one embodiment, the step of analyzing user input using a large model to obtain intent and entity information of the input includes:

[0047] If the input content includes voice information, the voice information is converted into text information;

[0048] Perform preprocessing operations on the text information to obtain preprocessed text;

[0049] The intent classification module of the large model is used to perform semantic analysis on the preprocessed text to obtain the intent information of the input content;

[0050] The named entity recognition module of the large model is used to perform entity analysis on the preprocessed text to obtain the entity information of the input content.

[0051] Accurately converting voice information into text information enables subsequent large-scale model analysis modules to better understand and process user input, ensuring the smooth initiation of business processes and improving system usability and user experience. Specifically, the system receives user requests in voice form through the front-end voice interaction module of the university's intelligent service system (such as voice input buttons on mobile applications and voice input functions on web pages). The system performs preliminary processing on the voice information, standardizing its format, including unifying parameters such as sampling rate and quantization bits, to ensure that the voice data meets conversion requirements. The voice recognition engine uses acoustic models, language models, and other technologies to map the voice signal into corresponding text content.

[0052] Raw text information may contain various noises and formatting issues, such as extra spaces, special characters, colloquial expressions, and spelling errors. These factors can affect the accuracy of large models' understanding and analysis of the text. By performing preprocessing operations on the text, noise can be removed, formatting can be standardized, and expressions can be normalized, making the text cleaner and more standardized, thus improving the semantic analysis performance of large models.

[0053] User-input text may contain various potential intents, such as inquiring about business processing procedures, checking business status, or submitting business applications. By performing semantic analysis on the preprocessed text through the intent classification module of the large model, the underlying business intent categories can be deeply uncovered, enabling the system to respond accordingly. The preprocessed text is input into the intent classification module, which encodes the text and extracts semantic feature vectors. For example, in the "graduate scholarship application" example, the preprocessed text "I want to apply for this year's graduate academic scholarship, what materials do I need to prepare?" will be encoded into a specific semantic feature vector, which represents the text's position and characteristics in the semantic space. Based on the extracted semantic feature vectors, the intent classification module uses an internal classifier to calculate the probability distribution of the text belonging to different business intent categories, selecting the category with the highest probability as the text's intent information output. For example, in the above example, the model will calculate that the text has the highest probability of belonging to the intent category of "scholarship application materials consultation," thus determining this intent information.

[0054] By performing entity analysis on preprocessed text using the named entity recognition module of the large model, various entity information contained in the text can be accurately extracted, providing detailed and accurate data support for subsequent business processing. Specifically, the preprocessed text is input into the named entity recognition module, which scans and analyzes each word or phrase in the text to determine whether it belongs to a predefined entity category (such as business name, time point, name, student ID, etc.). For example, in the example text "I want to apply for this year's graduate academic scholarship, what materials do I need to prepare?", the model will identify "graduate academic scholarship" as a business name entity and "this year" as a time point entity. The identified entities and their corresponding category information are integrated to form a structured entity information output. For example, the output is in a format like "Business Name: Graduate Academic Scholarship, Time Point: This Year", which facilitates the direct calling and utilization of this entity information by subsequent business processing modules.

[0055] In one embodiment, querying the knowledge graph based on the intent information and the entity information to determine the matching result with the input content and the user's current information includes:

[0056] Based on the intent information and the entity information, standardized query parameters are generated;

[0057] Based on the query parameters, perform multi-level matching query operations in the business knowledge graph to obtain policy rule nodes related to the query parameters, and output a policy matching list based on the policy rule nodes;

[0058] Based on the user entity information in the query parameters, the user data interface is called to obtain the user's current information;

[0059] Based on the user's current information, policy items that match the user's current status are selected from the policy matching list as the matching result.

[0060] Based on the structure and query requirements of the business knowledge graph, the format of query parameters is predefined. Query parameters include an intent category field and multiple entity information fields, each with a clearly defined name, data type, and value range. User intent and entity information are extracted from the current session records of the business processing system. Intent information is converted into predefined intent category values; for example, "scholarship application material consultation" is converted to "intent_type = application_material_inquiry". For entity information, conversion is performed according to predefined entity types and format requirements; for example, "graduate academic scholarship" in the entity information is converted to "business_name = postgraduate_scholarship".

[0061] By performing multi-level matching queries, policy rule nodes related to user needs can be progressively explored within the knowledge graph, providing users with accurate and comprehensive policy matching results. Based on the intent category and business name entity information in the query parameters, a preliminary query is performed in the business knowledge graph to find policy rule nodes directly related to the intent category and business name, yielding initial matching results. For example, based on the query parameters, basic policy nodes related to graduate scholarship application materials can be found in the knowledge graph.

[0062] Based on the initial matching results, further queries are performed to identify other policy and rule nodes associated with these nodes. By traversing the edge relationships in the knowledge graph, higher-level policy nodes (such as the school's general rules for scholarship management), lower-level policy nodes (such as detailed regulations for specific scholarship programs), and peer-level policy nodes (such as application material requirements for other related scholarship programs) connected to the basic policy nodes are found. For example, it was discovered that the postgraduate academic scholarship application materials node is associated with the school's scholarship management regulations node, and is also connected to specific scholarship application materials nodes of other colleges.

[0063] Based on the time nodes and other entity information in the query parameters, the policy rule nodes obtained from the extended query are filtered, retaining nodes that meet the time node and other conditions, and removing nodes that do not meet the conditions. For example, expired scholarship application policy nodes from previous years are filtered out, and only valid policy nodes for this year are retained. The policy rule nodes after multi-level matching queries and filtering are organized into a policy matching list. Each item in the list contains information such as the name, description, related business processes, and requirements of the policy rule.

[0064] A user's current status (such as identity information, academic performance, and existing business transaction records) is crucial for determining the applicable policies and rules. By accessing the user's current information through the user data interface, the system can accurately match the user's actual situation with the policy rules. User entity information is extracted from the query parameters, including the user's unique identifier (such as student ID or employee ID) and identity type (such as undergraduate, graduate, faculty, etc.). Based on the user entity information, the system calls the university's user data interface to extract the user's current status information from the returned data. This information includes the user's identity information (such as college, major, and year), academic performance (such as GPA), existing business transaction records (such as awards that have passed the initial review), and incomplete business processes. By understanding the user's current information, the system can filter out the business processing solutions that best match the user's needs and status, avoiding the recommendation of inapplicable business processes.

[0065] After obtaining the user's current information, the policy items in the policy matching list need to be further filtered to determine which policy items match the user's current state. Specifically, the obtained user's current information is compared with the applicable conditions of each policy item in the policy matching list. For each policy item, it is checked whether the user's current state meets its applicable conditions. Policy items that meet the applicable conditions are filtered out and output as the final matching result. By filtering based on the user's current information, it can be ensured that the matching results are highly relevant to the user's actual situation.

[0066] In one embodiment, the step of performing a multi-level matching query operation in the business knowledge graph based on the query parameters to obtain policy rule nodes related to the query parameters includes:

[0067] The query parameters are broken down into intent dimension parameters and constraint dimension parameters;

[0068] Based on the intent dimension parameter, matching is performed in the policy category node of the business knowledge graph, and the initially matched policy category node is output.

[0069] Based on the constraint dimension parameters, the policy rule nodes related to the query parameters are obtained by matching the policy category nodes under the initially matched policy category nodes.

[0070] Query parameters contain various types of information. Intent-level parameters reflect the core category of the user's business needs, while constraint-level parameters specify the concrete conditions and scope of those needs. Breaking down query parameters allows for clearer identification of relevant nodes in the knowledge graph, improving query accuracy and efficiency. Intent-level parameters typically represent the user's business intent category, such as "business consultation" or "business application." Constraint-level parameters include specific business names, timeframes, and other entity information, used to limit the scope and conditions of the business. By splitting query parameters into intent-level and constraint-level parameters, the scope and conditions of the query can be defined more clearly, providing explicit input for subsequent multi-level matching queries and ensuring the relevance and accuracy of the query process.

[0071] The intent category value in the intent dimension parameter is matched with the name or attribute of the policy category node. For example, the intent category corresponding to "application_material_inquiry" in the intent dimension parameter is "business application consultation". Nodes in the policy category node whose names or attributes contain keywords such as "application" or "consultation" are searched. Based on the initial matching results, the query is expanded to include child and sibling nodes associated with the matched node, ensuring that no potentially related policy category nodes are missed. For example, if the initial match is a "scholarship application" node, its child nodes "graduate scholarship application", "undergraduate scholarship application", etc., are further queried. The matched policy category nodes and their associated nodes are then compiled into a preliminary list of matched policy category nodes.

[0072] The constraint dimension parameters provide specific business conditions and scope information. Using these parameters, detailed matching can be performed on the subordinate nodes of the initially matched policy category nodes to obtain policy rule nodes that precisely match the user's needs. Within the subordinate nodes of the initially matched policy category nodes, each node is matched one by one based on the specific values ​​of the constraint dimension parameters. For example, within the subordinate nodes of the "Postgraduate Scholarship Application" policy category node, nodes whose names or attributes contain "postgraduate_scholarship" and "current_year" are searched. The matched nodes are then validated to ensure they meet the requirements of all constraint dimension parameters. For example, it is checked whether a node simultaneously satisfies the conditions of a business name of "Postgraduate Scholarship" and a timeframe of "this year". The validated policy rule nodes are output as the final matching results. Through precise matching based on constraint dimension parameters, policy rule nodes that precisely match the user's needs can be obtained, ensuring the accuracy and usability of the query results.

[0073] In one embodiment, generating a processing flow list and pre-filled information using a business engine based on the matching result and the user's current information includes:

[0074] Extract the structured descriptive information representing the application conditions and material requirements from the matching results;

[0075] Based on the user's current information, determine the user's satisfied conditions and missing information;

[0076] Based on the structured description information, the satisfied conditions, and the missing information, a processing flow information list is generated;

[0077] Based on the processing flow list and the user's current information, data fields for automatic filling are extracted as the pre-filled information.

[0078] The policy and rule nodes in the matching results may contain a wealth of information. Application conditions and material requirements are key elements for users processing their applications. Extracting this structured descriptive information provides foundational data support for generating accurate processing lists and pre-filled information, ensuring users clearly understand the conditions and materials required for processing. Specifically, within the policy and rule nodes of the matching results, relevant fields or attributes describing application conditions and material requirements are identified. The values ​​of these identified fields are extracted and converted into structured descriptive information. For example, the extracted application conditions might be "graduate student, GPA ≥ 3.5," and the material requirements might be "personal statement, transcripts, recommendation letters," etc. The extracted structured descriptive information is then organized according to a specific format for easier subsequent processing. For example, it can be organized into a JSON object containing two attributes: "application conditions" and "material requirements," storing the corresponding extracted content.

[0079] A comprehensive analysis of the acquired user information is conducted to extract relevant eligibility information for business processing. This information includes, but is not limited to, the user's identity information, academic performance, and existing business processing records. For example, the user's current information might include a graduate student status, a GPA of 3.8, and a previous scholarship application. The extracted user eligibility information is then compared one by one with the extracted application requirements. For each requirement, it is determined whether the user already meets it. For instance, comparing the user's GPA of 3.8 with the requirement of "GPA ≥ 3.5" confirms that the user meets this requirement. Based on the comparison results, any unmet requirements and missing information are identified. For example, if the application requires a letter of recommendation, but the user's current information does not mention any submission record, then the letter of recommendation is identified as missing information. By identifying the user's met requirements and missing information, the system can provide users with quick and accurate guidance for business processing, helping them understand their progress and any necessary supplementary information.

[0080] Generating a processing flow information list provides users with specific and detailed business processing steps and requirements, ensuring that users can smoothly complete business processes by following the list's guidance. Specifically, a template for the processing flow information list is pre-designed based on the business type and common business processing procedures. The template defines the basic structure and content framework of the list, including fields such as step number, step description, and required materials. The extracted structured description information, the user's met conditions, and missing information are then filled into the list template. The list content is personalized according to the user's actual situation. For example, steps that have met the conditions are marked as "completed" or "no action required" in the list; for steps with missing information, detailed explanations of the required supplementary materials and operating methods are provided. Based on business logic and user convenience, the steps in the processing flow list are optimized and sorted to ensure that the order of steps in the list is reasonable and conforms to the user's actual processing flow. The generated processing flow information list is output in a prescribed format for subsequent use. For example, it can be output as an HTML page, a PDF document, or directly displayed to the user on the business processing system interface.

[0081] By extracting relevant data fields from the user's current information, automatic form filling can be achieved, greatly simplifying user operations, reducing repetitive input, and improving business processing efficiency and user experience. Specifically, a mapping relationship is pre-established between the form fields involved in each step of the processing flow list and the data fields in the user's current information. For example, the "Name" field in the application form corresponds to the "name" field in the user's current information, and the "Student ID" field corresponds to the "student_id" field, etc. Based on the mapping relationship, the corresponding data fields are extracted from the user's current information. The extracted data fields undergo necessary format conversion and validation to ensure that the data meets the form's filling requirements. For example, the value "Zhang San" in the user's current information's name field is extracted, and its format is verified to meet the requirements of the name field in the form. The extracted and converted data fields are used as pre-filled information and filled into the corresponding form fields. For example, in a scholarship application form, the user's name, student ID, identity type, and other known information are automatically filled in, eliminating the need for the user to manually enter this information. The generated pre-filled information is output along with the processing flow list for the user to use during the business processing. For example, pre-filled forms can be displayed on the interface of a business processing system, allowing users to submit them directly or make minor additions and modifications. By extracting data fields for automatic filling and generating pre-filled information, the amount of input required from users during business processing can be significantly reduced, the error rate can be lowered, and the efficiency and accuracy of business processing can be improved.

[0082] In one embodiment, before generating guidance information and feeding it back to the user based on the processing flow list and the pre-filled information, the method further includes:

[0083] Based on the user's current information, the processing flow list and pre-filled information are verified.

[0084] Because business processing procedures involve numerous details and compliance requirements, errors or incomplete processing lists or pre-filled information can lead to user failures or delays, or even submissions that do not comply with business rules. Therefore, it is necessary to validate the processing list and pre-filled information. Specifically, based on university business policies and processing standards, detailed validation rules are formulated. These rules cover the compliance, logical order, and material completeness of each step in the processing list, as well as the format, accuracy, and completeness of pre-filled information. For example, for postgraduate scholarship applications, the application process must include key steps such as "submitting the application form," "faculty recommendation," and "departmental preliminary review," and the order of these steps must not be reversed. The validation rules are stored in configuration files or a database so that the business engine can call and execute the validation operations.

[0085] According to the verification rules, each step in the processing flow list is compared one by one to check whether it is consistent with the preset business process specifications. For example, it checks whether the process steps are complete, whether the order is correct, and whether the required materials match the business requirements. At the same time, it verifies whether the value of each field in the pre-filled information is consistent with the corresponding value in the user's current information, and whether the format meets the requirements. For example, it verifies whether the pre-filled student ID is consistent with the student ID in the user's current information, and whether the format is 10 digits; whether the name matches the user's identity information, and whether it contains only Chinese characters, etc.

[0086] During the validation process, if any discrepancies are found between the processing flow list and pre-filled information and the validation rules, the specific error information is recorded, including the erroneous step, field name, error type (e.g., formatting error, inconsistent content, missing information, etc.), and the corresponding validation rule description. The recorded error information is then fed back to the business processing system for subsequent correction. Simultaneously, based on the severity and scope of the error, it is determined whether the current business processing flow needs to be interrupted or only a warning is issued. If the validation fails, the processing flow list and pre-filled information are corrected according to the error information. The corrected information needs to be validated again until it passes. Through these rigorous validation steps, potential errors and incompleteness in the processing flow list and pre-filled information can be effectively identified and corrected, ensuring that the information fed back to the user is accurate, complete, and compliant.

[0087] In one embodiment, the method further includes:

[0088] The user's current information, the processing flow list, the pre-filled information, and the guidance information are uploaded to the blockchain.

[0089] By leveraging the immutability of blockchain, the security of user information, processing lists, pre-filled information, and guidance information during storage and transmission is ensured, preventing malicious tampering or leakage of data. For example, in the graduate scholarship application process, information such as user academic records and application materials uploaded to the blockchain cannot be modified by unauthorized personnel, guaranteeing the authenticity and credibility of user data. The traceability of blockchain ensures that all relevant information during the business process is verifiable, supporting business review, traceability, and supervision. For instance, when the scholarship review committee reviews an application, it can quickly retrieve the applicant's original application information and processing flow through the blockchain, ensuring transparency in the review process.

[0090] Reference Figure 2 The diagram shown is a functional module schematic of the business processing device 100 based on a large model in this application.

[0091] The large-model-based business processing device 100 described in this application is installed in an electronic device. Depending on the functions implemented, the large-model-based business processing device 100 includes an analysis module 110, a query module 120, a first generation module 130, and a second generation module 140. These modules can also be referred to as units, which are a series of computer program segments that can be executed by the processor of an electronic device and perform a fixed function, and are stored in the memory of the electronic device.

[0092] In this embodiment, the functions of each module / unit are as follows:

[0093] Analysis module 110: used to analyze the user's input content through a large model to obtain the intent information and entity information of the input content;

[0094] Query module 120: used to query the knowledge graph based on the intent information and the entity information, and determine the matching result with the input content and the user's current information;

[0095] First generation module 130: used to generate a processing flow list and pre-filled information based on the matching result and the user's current information using a business engine;

[0096] The second generation module 140 is used to generate guidance information and feed it back to the user based on the processing flow list and the pre-filled information.

[0097] The specific implementation of the large-model-based business processing device in this application is largely the same as the specific implementation of the large-model-based business processing method described above, and will not be repeated here.

[0098] Reference Figure 3 The diagram shown is a schematic representation of a preferred embodiment of the electronic device of this application.

[0099] The electronic device includes a processor 111, a communication interface 112, a memory 113, and a communication bus 114, wherein the processor 111, the communication interface 112, and the memory 113 communicate with each other through the communication bus 114.

[0100] Memory 113 is used to store computer programs, such as business processing programs based on large models;

[0101] In some embodiments, the processor 111 may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 111 is typically used to control the overall operation of the electronic device, such as performing data interaction or communication-related control and processing. In this embodiment, the processor 111 is used to run program code stored in the memory 113 or process data.

[0102] The communication interface 112 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface). The communication interface 112 may also be used to establish a communication connection between the electronic device and other electronic devices.

[0103] The memory 113 includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 113 may be an internal storage unit of the electronic device, such as the hard disk or memory of the electronic device. In other embodiments, the memory 113 may also be an external storage device of the electronic device, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc. of the electronic device. Of course, the memory 113 may include both internal storage units and external storage devices of the electronic device. In this embodiment, the memory 113 is typically used to store the operating system and various computer programs installed on the electronic device, such as program code for business processing programs based on large models. In addition, the memory 113 may also be used to temporarily store various types of data that have been output or will be output.

[0104] Figure 3Only an electronic device with components 111-114 is shown; however, it should be understood that it is not required to implement all of the components shown, and more or fewer components may be implemented instead.

[0105] In one embodiment of this application, when the processor 111 executes the program stored in the memory 113, it implements the business processing method based on a large model provided in any of the foregoing method embodiments, including:

[0106] By analyzing user input through a large model, the intent and entity information of the input content can be obtained.

[0107] Based on the intent information and the entity information, a knowledge graph is queried to determine the matching result with the input content and the user's current information;

[0108] Based on the matching results and the user's current information, the business engine generates a processing flow list and pre-filled information.

[0109] Based on the processing flow list and the pre-filled information, guidance information is generated and fed back to the user.

[0110] For a detailed explanation of the above steps, please refer to the above. Figure 1 A flowchart illustrating an embodiment of a business processing method based on a large model.

[0111] Furthermore, this application also proposes a computer-readable storage medium that is both non-volatile and volatile. This computer-readable storage medium is any one or any combination of several of the following: hard disk, multimedia card, SD card, flash memory card, SMC, read-only memory (ROM), erasable programmable read-only memory (EPROM), portable compact disc read-only memory (CD-ROM), USB memory, etc. The computer-readable storage medium includes a data storage area and a program storage area. The program storage area stores a business processing program based on a large model. When executed by a processor, the business processing program based on the large model performs the following operations:

[0112] By analyzing user input through a large model, the intent and entity information of the input content can be obtained.

[0113] Based on the intent information and the entity information, a knowledge graph is queried to determine the matching result with the input content and the user's current information;

[0114] Based on the matching results and the user's current information, the business engine generates a processing flow list and pre-filled information.

[0115] Based on the processing flow list and the pre-filled information, guidance information is generated and fed back to the user.

[0116] The specific implementation of the computer-readable storage medium in this application is largely the same as the specific implementation of the business processing method based on the large model described above, and will not be repeated here.

[0117] It should be noted that the sequence numbers of the embodiments in this application are for descriptive purposes only and do not represent the superiority or inferiority of the embodiments. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, apparatus, article, or method. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0118] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware simulation platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk) as described above, and includes several instructions to cause a terminal device to execute the methods described in the various embodiments of this application.

[0119] The above are merely preferred embodiments of this application and do not limit the patent scope of this application. Any equivalent structural or procedural transformations made using the content of this application's specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of this application.

Claims

1. A business processing method based on a large model, characterized in that, The method includes: By analyzing user input through a large model, the intent and entity information of the input content can be obtained. Based on the intent information and the entity information, a knowledge graph is queried to determine the matching result with the input content and the user's current information; Based on the matching results and the user's current information, the business engine generates a processing flow list and pre-filled information. Based on the processing flow list and the pre-filled information, guidance information is generated and fed back to the user.

2. The business processing method based on a large model as described in claim 1, characterized in that, The step of analyzing user input using a large model to obtain intent and entity information from the input includes: If the input content includes voice information, the voice information is converted into text information; Perform preprocessing operations on the text information to obtain preprocessed text; The intent classification module of the large model is used to perform semantic analysis on the preprocessed text to obtain the intent information of the input content; The named entity recognition module of the large model is used to perform entity analysis on the preprocessed text to obtain the entity information of the input content.

3. The business processing method based on a large model as described in claim 1, characterized in that, The step of querying the knowledge graph based on the intent information and the entity information to determine the matching result with the input content and the user's current information includes: Based on the intent information and the entity information, standardized query parameters are generated; Based on the query parameters, perform multi-level matching query operations in the business knowledge graph to obtain policy rule nodes related to the query parameters, and output a policy matching list based on the policy rule nodes; Based on the user entity information in the query parameters, the user data interface is called to obtain the user's current information; Based on the user's current information, policy items that match the user's current status are selected from the policy matching list as the matching result.

4. The business processing method based on a large model as described in claim 3, characterized in that, The step of performing a multi-level matching query operation in the business knowledge graph based on the query parameters to obtain policy rule nodes related to the query parameters includes: The query parameters are broken down into intent dimension parameters and constraint dimension parameters; Based on the intent dimension parameter, matching is performed in the policy category node of the business knowledge graph, and the initially matched policy category node is output. Based on the constraint dimension parameters, the policy rule nodes related to the query parameters are obtained by matching the policy category nodes under the initially matched policy category nodes.

5. The business processing method based on a large model as described in claim 1, characterized in that, The step of generating a processing flow list and pre-filled information using the business engine based on the matching result and the user's current information includes: Extract the structured descriptive information representing the application conditions and material requirements from the matching results; Based on the user's current information, determine the user's satisfied conditions and missing information; Based on the structured description information, the satisfied conditions, and the missing information, a processing flow information list is generated; Based on the processing flow list and the user's current information, data fields for automatic filling are extracted as the pre-filled information.

6. The business processing method based on a large model as described in claim 1, characterized in that, Before generating guidance information and sending it back to the user based on the processing flow list and the pre-filled information, the method further includes: Based on the user's current information, the processing flow list and pre-filled information are verified.

7. The business processing method based on a large model as described in claim 1, characterized in that, The method further includes: The user's current information, the processing flow list, the pre-filled information, and the guidance information are uploaded to the blockchain.

8. A business processing device based on a large model, characterized in that, The device includes: Analysis module: Used to analyze user input through a large model to obtain intent and entity information of the input content; Query module: used to query the knowledge graph based on the intent information and the entity information, and determine the matching result with the input content and the user's current information; First generation module: used to generate a processing flow list and pre-filled information based on the matching result and the user's current information using the business engine; The second generation module is used to generate guidance information and send it back to the user based on the processing flow list and the pre-filled information.

9. An electronic device, characterized in that, It includes a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; Memory, used to store computer programs; When executing a program stored in memory, the processor implements the business processing method based on a large model as described in any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the business processing method based on the large model as described in any one of claims 1 to 7.

Images