Service information processing method and device, computer equipment and readable storage medium

Abstract

The invention relates to a service information processing method and device, computer equipment and a readable storage medium. The method comprises the following steps: performing feature extraction on original service information input by a user to obtain initial service features; selecting a target model corresponding to a business scene of the original business information according to the initial business characteristics, and generating business form field data according to the original business information and the target model; and adding the business form field data to a corresponding business form field of a business form. The method can dynamically adapt to the business process difference of different financial institutions, and improves the flexibility.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence technology, and in particular to a business information processing method, apparatus, computer equipment, and readable storage medium. Background Technology

[0002] In the financial and banking sectors, user information entry is a crucial step in core business processes such as account opening, account management, and risk assessment. Traditional form-filling methods suffer from the following pain points: Complex user operation: Financial forms have numerous fields, requiring users to provide large amounts of structured and unstructured data, leading to high user churn rates; Insufficient data accuracy: Users may cause data errors due to formatting errors, missing information, or ambiguous expressions, requiring manual review and increasing operational costs; High compliance risks: Financial businesses must strictly adhere to regulatory requirements, and traditional rule engines struggle to cover complex scenarios; High system development costs: To adapt to diverse business needs, frequent updates to the rule base or model are required, resulting in long system iteration cycles and high maintenance costs.

[0003] Among related technologies, OCR (Optical Character Recognition), speech recognition, or a single NLP (Natural Language Processing) model can solve some of the problems of traditional form filling methods, but there are still problems such as the inability to dynamically adapt to the differences in business processes of different financial institutions and poor flexibility. Summary of the Invention

[0004] Therefore, it is necessary to provide a business information processing method, apparatus, computer equipment, and readable storage medium that can dynamically adapt to the differences in business processes of different financial institutions and improve flexibility in response to the above-mentioned technical problems.

[0005] Firstly, this application provides a business information processing method, including:

[0006] Feature extraction is performed on the raw business information input by the user to obtain initial business features;

[0007] Select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model;

[0008] Add the data from the business form fields to the corresponding business form fields.

[0009] In one embodiment, selecting a target model corresponding to the business scenario based on the initial business characteristics and the original business information includes:

[0010] The initial business characteristics are analyzed for field types and language complexity to obtain the feature analysis results;

[0011] Based on the feature analysis results, select the target model that corresponds to the business scenario of the original business information.

[0012] In one embodiment, the target model includes a target business domain model and a general large language model; selecting the target model corresponding to the business scenario of the original business information based on the feature analysis results includes:

[0013] When the feature analysis results include structured information and / or unstructured information with low linguistic complexity, select the target business domain model from multiple candidate business domain models that corresponds to the business scenario of the original business information; and

[0014] When the feature analysis results also include unstructured information with high linguistic complexity, a general large language model is selected.

[0015] In one embodiment, generating business form field data based on the original business information and the target model includes:

[0016] The structured information from the original business information, and / or the unstructured information with low language complexity, are input into the target business domain model for classification processing to obtain the classification results.

[0017] The unstructured information with high linguistic complexity in the original business information is input into a general large language model for intent recognition, and the intent recognition result is obtained.

[0018] The classification results and intent recognition results are combined to obtain the combined result, and business form field data is generated based on the combined result.

[0019] In one embodiment, feature extraction is performed on the original business information input by the user to obtain initial business features, including:

[0020] Keyword and contextual features are extracted from the original business information input by the user to obtain initial business features.

[0021] In one embodiment, the method further includes:

[0022] Perform compliance validation on the data in the business form fields;

[0023] After passing the compliance verification, the data from the business form fields will be added to the corresponding business form fields.

[0024] In one embodiment, the method further includes:

[0025] Generate dialog prompts for missing, ambiguous, or conflicting fields in business forms;

[0026] Receive response information to dialogue prompts and optimize business forms based on the response information.

[0027] Secondly, this application also provides a business information processing apparatus, comprising:

[0028] The feature extraction module is used to extract features from the raw business information input by the user to obtain initial business features;

[0029] The field generation module is used to select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model.

[0030] The form filling module is used to add data from business form fields to the corresponding business form fields.

[0031] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:

[0032] Feature extraction is performed on the raw business information input by the user to obtain initial business features;

[0033] Select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model;

[0034] Add the data from the business form fields to the corresponding business form fields.

[0035] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:

[0036] Feature extraction is performed on the raw business information input by the user to obtain initial business features;

[0037] Select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model;

[0038] The business form field data is added to the corresponding business form field of the business form. Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:

[0039] Feature extraction is performed on the raw business information input by the user to obtain initial business features;

[0040] Select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model;

[0041] Add the data from the business form fields to the corresponding business form fields.

[0042] The aforementioned business information processing method, apparatus, computer equipment, and readable storage medium, based on existing business information processing technologies, further introduce a dynamic model selection mechanism. By selecting a target model corresponding to the business scenario of the original business information based on the initial business characteristics of the original business information, business form field data is automatically generated to achieve automatic filling of business forms. This significantly improves the adaptability, accuracy, and scalability of business form filling in complex business scenarios, and solves the problem that traditional technologies cannot dynamically adapt to the differences in business processes of different financial institutions and have poor flexibility. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart illustrating a business information processing method in one embodiment;

[0045] Figure 2 This is a flowchart illustrating the process of selecting a target model corresponding to a business scenario based on initial business characteristics and original business information in one embodiment.

[0046] Figure 3 This is a flowchart illustrating the process of selecting a target model corresponding to a business scenario based on feature analysis results, as described in one embodiment.

[0047] Figure 4 This is a flowchart illustrating the dynamic model combination process in one embodiment;

[0048] Figure 5 This is a flowchart illustrating compliance verification and multi-round dialogue guidance in one embodiment;

[0049] Figure 6 This is a schematic diagram of the structure of a business information processing system in one embodiment;

[0050] Figure 7 This is a structural block diagram of a business information processing device in one embodiment;

[0051] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0052] 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.

[0053] It should be noted that the terms "comprising" and "having," and any variations thereof, as used in this application, are intended to cover non-exclusive inclusion. The term "multiple" as used in this application refers to two or more. The term "and / or" as used in this application refers to one of the solutions, or any combination of multiple solutions.

[0054] In one exemplary embodiment, such as Figure 1 As shown, a business information processing method is provided. This embodiment illustrates the method applied to a terminal, but it is understood that the method can also be applied to a server, or to a system including both a terminal and a server, and implemented through interaction between the terminal and the server. In this embodiment, the method includes the following steps:

[0055] Step 102: Extract features from the original business information input by the user to obtain initial business features.

[0056] Here, "raw business information" refers to the natural language information provided by the user for generating business forms, which can include various forms such as text, voice, and images. "Business form" refers to the standardized set of structured fields that ultimately need to be generated or filled in. "Initial business features" refers to the key elements automatically extracted from the raw business information input by the user, used for dynamically selecting the model.

[0057] In some implementations, the user-inputted raw business information is obtained. This raw business information may include text, voice, or image information, such as "I need to apply for a resource loan of 1 million, with real estate as collateral." The raw business information is then preprocessed. Preprocessing methods may include: extracting identity information, such as ID card information or bank card information, from the raw business information using OCR; converting voice business information into text information; cleaning the image business information, such as denoising or cropping; and extracting features from the processed information to obtain initial business features.

[0058] In an exemplary embodiment, feature extraction of the user-inputted raw business information to obtain initial business features includes: extracting keywords and contextual features from the user-inputted raw business information to obtain initial business features.

[0059] The processed business information is then subjected to keyword and contextual feature extraction to obtain initial business features. For example, for the business information ("I need to apply for a resource loan of 1 million, with real estate as collateral"), keywords ("resource loan amount", "collateral") and contextual features ("real estate" → collateral type) are extracted. Or, for the business information ("I want to apply for a consumer card"), keywords ("apply", "card") and contextual features ("apply" → card) are extracted.

[0060] By extracting keywords and contextual features from the raw business information input by users, the core business elements and their semantic relationships in the input content can be identified in advance, thereby forming initial business features and providing an accurate basis for subsequent model selection.

[0061] Step 104: Select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model.

[0062] Step 106: Add the business form field data to the corresponding business form field of the business form.

[0063] The target model refers to the model used to generate business form field data. It can be a single business domain model or a combination of a business domain model and a general language model. Business form field data refers to the field values ​​corresponding to the fields in the business form.

[0064] In some implementations, feature analysis is performed on the initial business characteristics to obtain feature analysis results. Based on the feature analysis results, a target model corresponding to the business scenario of the original business information is selected. Based on the original business information and the target model, business form field data is automatically generated. For example, based on the original business information "My monthly income is 5000 yuan," the "monthly income" field is automatically extracted. Then, the business form field data is added to the corresponding business form fields to complete the automatic form filling.

[0065] The aforementioned business information processing method, based on the original business information processing technology, further introduces a dynamic model selection mechanism. By selecting the target model corresponding to the business scenario of the original business information based on the initial business characteristics of the original business information, business form field data is automatically generated to achieve automatic filling of business forms. This significantly improves the adaptability, accuracy, and scalability of business form filling in complex business scenarios, and solves the problem that traditional technologies cannot dynamically adapt to the differences in business processes of different financial institutions and have poor flexibility.

[0066] In one exemplary embodiment, such as Figure 2 As shown, the target model selected based on the initial business characteristics and the business scenario corresponding to the original business information includes:

[0067] Step 202: Perform field type and language complexity analysis on the initial business features to obtain feature analysis results.

[0068] Step 204: Select the target model corresponding to the business scenario of the original business information based on the feature analysis results.

[0069] In some implementations, during the dynamic model selection process, the initial business features are analyzed for field type and language complexity to obtain feature analysis results. Field type analysis determines whether the initial business features are structured information. Language complexity analysis assesses the level of language complexity of the initial business features. Based on the feature analysis results, it is then determined whether to select only a single business domain model (i.e., a domain-specific small model) or a combination of models (i.e., a combination of a business domain model and a general language model). Finally, a target model corresponding to the business scenario of the original business information is selected. Domain-specific small models refer to models customized for the financial domain, such as credit scoring models, KYC (Know Your Customer) compliance models, and financial terminology parsing models. General language models can be LLM (Large Language Model).

[0070] In this embodiment, by analyzing the field type and language complexity of the initial business features, feature analysis results that reflect the structural form, semantic complexity, and contextual relationships of the input text are obtained. Subsequently, based on these feature analysis results, a target model corresponding to the business scenario of the original business information is automatically selected, including a general large model, a domain small model, or a combination thereof. Through this feature-driven model selection mechanism, the model combination can be adjusted in real time according to different user inputs, field requirements of different financial institutions, and different business scenarios, thereby achieving flexible semantic parsing capabilities and improving the parsing accuracy and adaptability for complex business inputs.

[0071] In one exemplary embodiment, such as Figure 3 As shown, the target model includes a target business domain model and a general large language model; step 204 involves selecting the target model corresponding to the business scenario of the original business information based on the feature analysis results, including steps 302 to 304. Wherein:

[0072] Step 302: If the feature analysis results include structured information and / or unstructured information with low language complexity, select the target business domain model from multiple candidate business domain models that corresponds to the business scenario of the original business information.

[0073] Step 304: If the feature analysis results also include unstructured information with high linguistic complexity, select a general large language model.

[0074] In some implementations, when the feature analysis results include structured information (such as "Name: Zhang San, Phone: XXX") and / or unstructured information with low linguistic complexity, it is necessary to select a business domain model. Specifically, a target business domain model corresponding to the business scenario of the original business information can be selected from multiple candidate business domain models. The target business domain model can understand the professional terms of the corresponding business scenario or domain, such as collateral valuation and credit scoring.

[0075] When the feature analysis results also include unstructured information with high linguistic complexity, a general-purpose large language model is selected. This unstructured information can include fuzzy text, such as "I need to apply for resource transfer; the collateral is real estate." The general-purpose large language model covers financial terms (such as "credit score" and "collateral ratio") and multilingual scenarios (such as mixed Chinese and English input). It is used to handle complex semantics, multilingual parsing, and understanding of fuzzy expressions, as well as to generate thought logic and a financial corpus for training smaller models through intent recognition.

[0076] Furthermore, a model hierarchical invocation strategy can be adopted: prioritize invoking smaller models to handle simple tasks (such as date format parsing, templated semantic extraction and analysis, etc.), and only trigger the general large language model in complex scenarios (such as multilingual mixed scenarios, large model intent recognition, etc.).

[0077] In this embodiment, when the feature analysis results include structured information and / or unstructured information with low linguistic complexity, a target business domain model corresponding to the business scenario of the original business information is selected from multiple candidate business domains. The target business domain model can understand the professional subject of the corresponding business scenario or domain, solving the problem of insufficient model generalization ability in traditional technologies. When the feature analysis results also include unstructured information with high linguistic complexity, a general large language model is selected. Dynamically selecting the target model based on the feature analysis results enables on-demand model invocation, is applicable to complex scenarios, and also reduces the invocation cost of large language models.

[0078] In an exemplary embodiment, generating business form field data based on original business information and a target model includes: inputting structured information and / or unstructured information with low language complexity from the original business information into a target business domain model for classification processing to obtain classification results; inputting unstructured information with high language complexity from the original business information into a general large language model for intent recognition to obtain intent recognition results; combining the classification results and intent recognition results to obtain a combined result, and generating business form field data based on the combined result.

[0079] In some implementations, such as Figure 4 The diagram illustrates the process of dynamic model combination. Structured information and / or low-complexity unstructured information from the original business information are input into the target business domain model. The target business domain model classifies the input information to obtain classification results. For example, the target business domain model could be a collateral classification model to parse collateral types ("real estate" → "real estate mortgage"). High-complexity unstructured information from the original business information is input into a general large language model for intent recognition to obtain intent recognition results. For example, an LLM model can be used to parse user intent ("resource borrowing application") and implicit needs (such as the purpose of resource borrowing, repayment ability, etc.), or a credit scoring model can be used to output a credit score. The classification results and intent recognition results are combined to obtain a combined result, and business form field data, such as resource borrowing amount, collateral type, purpose, etc., are generated based on the combined result. Then, automatic field filling is performed: the model predicts user intent, and business form field data is automatically generated.

[0080] In this embodiment, an asynchronous coordination mechanism is introduced to support parallel invocation and result integration of multiple models (such as LLM and domain-specific small models), avoiding the impact of large model inference delay on the overall process. This solves the problem that traditional technologies have long inference times for large models, which cannot meet the real-time interaction requirements of financial businesses (such as real-time risk assessment), resulting in poor real-time performance.

[0081] In an exemplary embodiment, the method further includes: performing compliance verification on the business form field data; and after passing the compliance verification, adding the business form field data to the corresponding business form field of the business form.

[0082] In some implementations, contextual semantic validation can also be performed on business form field data, such as verifying the compliance of the income-to-debt ratio. Compliance checks on business form field data are automatically performed using a financial regulatory rule base. If the compliance check passes, the business form field data is added to the corresponding business form field. If compliance fails, an error message is generated to alert the user to potential input errors, such as "The income certificate you entered does not match your bank statement," to facilitate error correction.

[0083] In this embodiment, by performing compliance checks before fields are written into the business form, the accuracy, completeness, and legality of the business form field data can be ensured, preventing erroneous fields from entering the business processing flow, thereby improving data quality and the reliability of business parsing. This mechanism not only prevents field errors generated by model parsing from propagating backward, but also ensures that the entire financial business process complies with regulatory and institutional standards, enhancing the system's security, stability, and scalability, and enabling flexible adaptation to the differences in business forms among different financial institutions.

[0084] In an exemplary embodiment, the method further includes: generating dialog prompt information for missing, ambiguous, or conflicting fields in the business form; receiving response information to the dialog prompt information; and optimizing the business form based on the response information.

[0085] like Figure 5 The diagram illustrates the compliance verification and multi-round dialogue guidance process. Multi-round dialogue guidance: If there are missing, ambiguous, or conflicting fields in the business form, proactively ask follow-up questions, such as, "Do you specifically mean real estate or vehicles when you mention 'collateral'?", i.e., inquire about the type of collateral. After receiving user feedback, optimize the business form based on the response information.

[0086] Furthermore, it also includes a feedback learning process: collecting user feedback data (such as user-corrected / annotated fields), retraining the target model based on the user feedback data, and continuously optimizing model performance. Caching and incremental learning can also be performed: caching the output results of high-frequency task models and continuously optimizing model performance through incremental learning (such as optimizing a credit scoring model based on historical resource borrowing and approval data).

[0087] In this embodiment, by generating dialog prompts when business form fields are detected to be missing, ambiguous, or conflicting, and dynamically optimizing field content based on user responses, the system can ensure the completeness, accuracy, and consistency of the final business form. This solution not only improves data parsing capabilities under complex semantic inputs but also significantly reduces the impact of single-time parsing errors in large models, enhancing the controllability of business processes and user experience, thereby increasing the success rate of form generation and strengthening the system's adaptability to different business scenarios.

[0088] Furthermore, a business information processing system runs on the terminal, and the aforementioned business information processing methods are applied to this system. Regarding deployment and optimization: The system is deployed on the bank's private cloud, supporting elastic scaling and data encryption. The optimality of the model combination strategy is verified through A (Control) / B (Treatment) testing methods, and model performance is continuously iterated. Regarding compliance and security design: Data privacy protection: Federated learning technology is used to ensure that user data is processed locally and not uploaded to the cloud. Enhanced interpretability: Field parsing basis is provided through model interpretation tools (e.g., "The source of the income field is the payroll OCR result").

[0089] The above methods also include: calling the dynamic model selection interface as a bridge between model input and output, receiving raw data from the business information processing system, coordinating the model selection process, and returning the final parsing result. Specifically, the asynchronous coordination mechanism supports parallel calls and result integration of multiple models (such as LLM and vertical small models), avoiding the impact of large model inference delays on the overall process. Dynamic data binding automatically adapts data formats (such as text, images, and multilingual mixed formats) according to the needs of financial business scenarios (such as resource borrowing approval and KYC compliance verification). Data required for dynamic model selection (such as user input fields, model parameters, and rule bases) can be stored in the process's private memory through the model attribute storage unit. Specifically, domain-customized storage designs data structures for financial scenarios (such as collateral classification and credit scoring) to optimize memory access efficiency; the incremental update mechanism, combined with the feedback learning module, dynamically updates the stored content (such as user-corrected fields and added compliance rules). It can also generate dynamic model selection process topologies according to business needs and obtain or create models from the model library; the process topology includes customized steps for the financial domain (such as collateral classification and compliance verification); the above methods support dynamic load balancing and incremental learning.

[0090] In one exemplary embodiment, a business information processing method is provided, comprising:

[0091] Keyword and contextual features are extracted from the original business information input by the user to obtain initial business features.

[0092] The initial business characteristics are analyzed for field types and language complexity to obtain the feature analysis results.

[0093] When the feature analysis results include structured information and / or unstructured information with low linguistic complexity, select the target business domain model corresponding to the business scenario of the original business information from multiple candidate business domain models; and when the feature analysis results also include unstructured information with high linguistic complexity, select a general large language model.

[0094] The structured information from the original business information, and / or the unstructured information with low language complexity, are input into the target business domain model for classification processing to obtain the classification results.

[0095] The unstructured information with high linguistic complexity in the original business information is input into a general large language model for intent recognition, and the intent recognition result is obtained.

[0096] The classification results and intent recognition results are combined to obtain the combined result, and business form field data is generated based on the combined result.

[0097] Perform compliance validation on the business form field data; after passing the compliance validation, add the business form field data to the corresponding business form field.

[0098] Generate dialog prompts for missing, ambiguous, or conflicting fields in business forms; receive response information to the dialog prompts and optimize the business forms based on the response information.

[0099] In the embodiments of this application, traditional technologies attempt to solve some problems through OCR, speech recognition or a single NLP model, but they have limitations such as insufficient model generalization ability, poor real-time performance and poor flexibility. The embodiments of this application propose a business information processing method based on multi-model dynamic intelligent recognition, which significantly improves the computing efficiency and recognition accuracy of application software. Moreover, the model has significant autonomous evolution capability as the amount of calls and time accumulate, and can be extended to a greater extent in terms of the breadth and depth of application scenarios.

[0100] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0101] Based on the same inventive concept, this application also provides a business information processing system for implementing the business information processing method described above. The solution provided by this system is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more business information processing system embodiments provided below can be found in the limitations of the business information processing method described above, and will not be repeated here.

[0102] In one exemplary embodiment, such as Figure 6 As shown, a business information processing system is provided, including: a preprocessing module 602, a model selection module 604, a dynamic combination module 606, a result optimization module 608, and a feedback learning module 610.

[0103] For specific limitations of this system, please refer to the limitations on business information processing methods mentioned above, which will not be repeated here.

[0104] Based on the same inventive concept, this application also provides a business information processing apparatus for implementing the business information processing method described above. The solution provided by this apparatus is similar to the implementation scheme described in the above method; therefore, the specific limitations in one or more business information processing apparatus embodiments provided below can be found in the limitations of the business information processing method described above, and will not be repeated here.

[0105] In one exemplary embodiment, such as Figure 7 As shown, a business information processing device is provided, including: a feature extraction module 702, a field generation module 704, and a form filling module 706, wherein:

[0106] The feature extraction module 702 is used to extract features from the original business information input by the user to obtain initial business features.

[0107] The field generation module 704 is used to select the target model corresponding to the business scenario of the original business information based on the initial business characteristics, and generate business form field data based on the original business information and the target model.

[0108] The form filling module 706 is used to add business form field data to the corresponding business form fields.

[0109] In an exemplary embodiment, the field generation module 704 is further configured to perform field type and language complexity analysis on the initial business features to obtain feature analysis results; and select a target model corresponding to the business scenario of the original business information based on the feature analysis results.

[0110] In an exemplary embodiment, the target model includes a target business domain model and a general large language model; the field generation module 704 is further configured to select a target business domain model corresponding to the business scenario of the original business information from multiple candidate business domain models when the feature analysis results include structured information and / or unstructured information with low language complexity; and to select a general large language model when the feature analysis results also include unstructured information with high language complexity.

[0111] In an exemplary embodiment, the field generation module 704 is further configured to input the structured information and / or the unstructured information with low language complexity from the original business information into the target business domain model for classification processing to obtain classification results; input the unstructured information with high language complexity from the original business information into the general large language model for intent recognition to obtain intent recognition results; combine the classification results and intent recognition results to obtain combined results, and generate business form field data based on the combined results.

[0112] In an exemplary embodiment, the feature extraction module 702 is further configured to extract keywords and contextual features from the original business information input by the user to obtain initial business features.

[0113] In one exemplary embodiment, the compliance verification module is used to perform compliance verification on the business form field data; after passing the compliance verification, the business form field data is added to the corresponding business form field of the business form.

[0114] In one exemplary embodiment, the form optimization module is used to generate dialog prompts for missing, ambiguous, or conflicting fields in a business form; receive response information to the dialog prompts; and optimize the business form based on the response information.

[0115] Each module in the aforementioned business information processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0116] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a business information processing method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0117] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0118] In one embodiment, a computer device is provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0119] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon, which, when executed by a processor, implements the steps in the above method embodiments.

[0120] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0121] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0122] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0123] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0124] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A business information processing method, characterized in that, The method includes: Feature extraction is performed on the raw business information input by the user to obtain initial business features; Based on the initial business characteristics, a target model corresponding to the business scenario of the original business information is selected, and business form field data is generated based on the original business information and the target model; Add the data from the business form fields to the corresponding business form fields of the business form.

2. The method according to claim 1, characterized in that, The step of selecting a target model corresponding to the business scenario based on the initial business characteristics and the original business information includes: The initial business characteristics are analyzed for field types and language complexity to obtain the feature analysis results; Based on the feature analysis results, a target model corresponding to the business scenario of the original business information is selected.

3. The method according to claim 2, characterized in that, The target model includes a target business domain model and a general language model; the step of selecting the target model corresponding to the business scenario of the original business information based on the feature analysis results includes: When the feature analysis results include structured information and / or unstructured information with low linguistic complexity, a target business domain model corresponding to the business scenario of the original business information is selected from multiple candidate business domain models; and When the feature analysis results also include unstructured information with high linguistic complexity, a general large language model is selected.

4. The method according to claim 3, characterized in that, The step of generating business form field data based on the original business information and the target model includes: The structured information and / or the unstructured information with low language complexity in the original business information are input into the target business domain model for classification processing to obtain the classification result; The unstructured information with high linguistic complexity in the original business information is input into the general large language model for intent recognition, and the intent recognition result is obtained. The classification result and the intent recognition result are combined to obtain a combined result, and the business form field data is generated based on the combined result.

5. The method according to claim 1, characterized in that, The initial business features obtained by extracting features from the user-inputted raw business information include: The initial business features are obtained by extracting keywords and contextual features from the original business information input by the user.

6. The method according to any one of claims 1 to 5, characterized in that, The method further includes: Perform compliance verification on the data in the business form fields; After passing the compliance verification, the data from the business form fields will be added to the corresponding business form fields.

7. The method according to any one of claims 1 to 5, characterized in that, The method further includes: For missing, ambiguous, or conflicting fields in the business form, generate dialog prompt messages; Receive response information to the dialog prompt information, and optimize the business form based on the response information.

8. A business information processing device, characterized in that, The device includes: The feature extraction module is used to extract features from the raw business information input by the user to obtain initial business features; The field generation module is used to select a target model corresponding to the business scenario of the original business information based on the initial business characteristics, and to generate business form field data based on the original business information and the target model; The form filling module is used to add the data from the business form fields to the corresponding business form fields of the business form.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to 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 a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images