Process page display method and device, electronic equipment and computer readable medium

Abstract

Embodiments of the present disclosure disclose a process page display method and device, electronic equipment and a computer readable medium. A specific embodiment of the method comprises: in response to receiving a process page rendering request, obtaining code information and a set of business processing criteria information; generating initial full business process information; verifying the execution relationship between the initial full business process information, obtaining full business process information; rendering the full business process information and rendering and displaying it on the process page; in response to selecting selection information, generating a matching business batch as a first business batch; and displaying the corresponding business process processing tree on the process page. This embodiment can accurately render the corresponding business process information on the process page after receiving the business data related to the business process generation uploaded to the process page, and can set various page filtering operations for the business process information, greatly improving the rendering efficiency of the business process.

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of computer technology, and more specifically to a method, apparatus, electronic device, and computer-readable medium for displaying process pages. Background Technology

[0002] Currently, with the continuous development of artificial intelligence, various existing businesses are gradually integrating intelligent technologies to streamline business processes and improve work efficiency. The typical approach to streamlining and generating business processes is as follows: First, professionals highly familiar with the business processes streamline the process and generate the necessary information. Then, relevant web developers generate a rendering result based on this business process information.

[0003] However, when using the above method, the following technical problems often arise: The requirement for highly skilled professionals and developers with extensive knowledge of the business processes leads to high demands on business process generation and rendering. Furthermore, the business process generation and rendering phases are handled by different objects, resulting in information gaps between the two phases. When the target business system has a large amount of complex business content, the workload for generating the business process is substantial, and the page rendering cycle is lengthy, leading to low rendering efficiency for the corresponding business processes in the target business system.

[0004] The information disclosed in this background section is only intended to enhance the understanding of the background of the inventive concept, and therefore may contain information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The summary portion of this disclosure is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description portion. This summary portion is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0006] Some embodiments of this disclosure provide methods, apparatuses, electronic devices, and computer-readable media for displaying process pages to address one or more of the technical problems mentioned in the background section above.

[0007] In a first aspect, some embodiments of this disclosure provide a process page display method, comprising: responding to receiving a process page rendering request for a target business system, obtaining code information and a set of business processing rules information for the target business system; generating initial full business process information for each business batch based on the code information and using a first large language model deployed on a first server, wherein the thought chain length corresponding to the first large language model satisfies a target length condition; verifying the execution relationship between business batches of the initial full business process information based on the set of business processing rules to obtain full business process information; generating page rendering information for the full business process information using the first large language model, and rendering and displaying the full business process information on the process page based on the page rendering information; responding to selecting selection information for at least one scene field on the process page, generating a business batch matching the at least one scene field as a first business batch based on the at least one scene field and the full business process information and using the first large language model; and displaying a stored business process processing tree corresponding to the first business batch on the process page.

[0008] Secondly, some embodiments of this disclosure provide a process page display device, including: an acquisition unit configured to, in response to receiving a process page rendering request for a target business system, acquire code information and a set of business processing guidelines information for the target business system; a first generation unit configured to, based on the code information and using a first large language model deployed on a first server, generate initial full business process information for each business batch, wherein the thought chain length corresponding to the first large language model satisfies a target length condition; and an adjustment unit configured to, based on the set of business processing guidelines, verify the execution relationship between business batches in the initial full business process information, thereby obtaining... The system includes: a first display unit configured to generate page rendering information for the full business process information using the first large language model, and to render and display the full business process information on the process page based on the page rendering information; a second generation unit configured to, in response to selection information for at least one scene field on the process page, generate a business batch matching the at least one scene field as a first business batch based on the at least one scene field and the full business process information using the first large language model; and a second display unit configured to display the stored business process processing tree corresponding to the first business batch on the process page.

[0009] Thirdly, some embodiments of this disclosure provide an electronic device, including: one or more processors; and a storage device having one or more programs stored thereon, such that when the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any implementation of the first aspect.

[0010] Fourthly, some embodiments of this disclosure provide a computer-readable medium having a computer program stored thereon, wherein the program, when executed by a processor, implements the method as described in any implementation of the first aspect.

[0011] The above embodiments of this disclosure have the following beneficial effects: Through the process page display method of some embodiments of this disclosure, after receiving business data related to business process generation uploaded to the process page, the corresponding business process information can be accurately rendered directly on the process page, and various page filtering operations supporting business process information can be set, greatly improving the rendering efficiency of the business process. Specifically, the reason for the low rendering efficiency of the relevant business processes is that it requires professionals and developers who are very familiar with the business processes, resulting in high requirements for business process generation and rendering. In addition, the business process generation stage and the business process rendering stage are handled by different objects, and there is an information gap between the two stages. Furthermore, when the target business system has a large amount of complex business content, the workload of business process generation is large and the rendering cycle of the page is long, resulting in low rendering efficiency of the corresponding business process of the target business system. Based on this, the process page display method of some embodiments of this disclosure first, in response to receiving a process page rendering request for the target business system, obtains the code information and business processing rule information set for the target business system. After selecting a request for one-click business process generation and rendering for the target business system, code information is obtained to facilitate the analysis of various business logics within the target business system. This allows the primary language model to learn these logics and accurately generate the corresponding business processes. By acquiring a set of business processing guidelines, the initial business process information generated can avoid inter-batch business interaction issues, improving the accuracy of subsequent full business process information generation. Then, based on the code information, the primary language model deployed on the primary server accurately and efficiently generates initial full business process information for each business batch, where the thought chain length corresponding to the primary language model meets the target length condition. Here, by setting the thought chain length of the primary language model to meet the target length condition, the primary language model can perform more thought steps during the process of learning code information to generate initial full business process information, resulting in more accurate business process information. Next, based on the set of business processing guidelines, the execution relationships between business batches of the initial full business process information are verified to obtain full business process information, resulting in more accurate process information with precise business logic. Furthermore, utilizing the first major language model, page rendering information for the entire business process can be accurately and automatically generated, and based on this page rendering information, the entire business process information can be effectively rendered and displayed on the process page. Here, using the first major language model to generate page rendering information improves the accuracy of the generated information. Specifically, because the first major language model can learn more deeply the business content related to the target business system, the generated page rendering information is also more accurate.Furthermore, in response to the selection information of at least one scenario field on the process page, based on the at least one scenario field and the full business process information, and utilizing the first large language model, the first large language model's proficiency in business content can be leveraged to accurately and efficiently infer and generate a business batch matching the at least one scenario field, which serves as the first business batch. Finally, the stored business process processing tree corresponding to the first business batch is displayed on the process page. In summary, through the first large language model deployed on the first server and the setting of the thought chain length, a series of processes can be implemented for full business process information, including process information mastery and generation, page rendering, and process search. Based on this, accurate and intelligent generation of corresponding business process information can be achieved. In addition, by performing one-click process generation for the target business system on the process page, accurate display of business process information on the process page can be achieved without wasting a lot of human resources. Furthermore, the ability to set up various page filtering operations for business process information greatly improves the rendering efficiency of the business process. Attached Figure Description

[0012] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and elements are not necessarily drawn to scale.

[0013] Figure 1 These are flowcharts of some embodiments of the process page display method according to this disclosure; Figure 2 This diagram illustrates the content display of business batches under various scenario fields. Figure 3 This diagram illustrates the content displayed after selecting a function. Figure 4 This diagram illustrates the content display after selecting the target business content for calculation. Figure 5 These are schematic diagrams illustrating the structure of some embodiments of the process page display device according to this disclosure; Figure 6 This is a schematic diagram of the structure of an electronic device suitable for implementing some embodiments of the present disclosure. Detailed Implementation

[0014] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0015] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other.

[0016] It should be noted that the concepts of "first" and "second" mentioned in this disclosure are used only to distinguish different devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0017] It should be noted that the terms "a" and "a plurality of" used in this disclosure are illustrative rather than restrictive, and those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0018] The names of messages or information exchanged between multiple devices in the embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.

[0019] This disclosure will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] refer to Figure 1 The diagram illustrates a process 100 according to some embodiments of a process page display method based on the present disclosure. This process page display method includes the following steps: Step 101: In response to receiving a process page rendering request for the target business system, obtain the code information and business processing guidelines information set for the target business system.

[0021] In some embodiments, in response to receiving a process page rendering request for a target business system, the executing entity of the aforementioned process page display method (e.g., an electronic device) can obtain code information and a set of business processing guidelines for the target business system via wired or wireless means. The target business system can be a system that processes business matters related to the target business. Here, the target business system can be a business system for which a corresponding business process is to be built. For example, the target business system can be a financial system or a logistics system. The process page rendering request can be a request to render the business process corresponding to the target business system. That is, the business process can be the business execution flow between various businesses within the target business system. In practice, after uploading the code information and the set of business processing guidelines corresponding to the target business system to the process page, clicking the one-click generation control displayed on the process page can directly generate the business process corresponding to the target business system and render the business process on the process page. The code information can be the system source file corresponding to the target business system. The code information can represent the overall business execution logic corresponding to the target business system. The business processing guidelines information is a systematic set of rules used to guide the processing flow and methods of various businesses. Business processing guidelines information can also characterize the execution logic between various business processes and the precautions that need to be taken when executing each business process.

[0022] For example, for securities business, the business processing standard information could be the position profit and loss standard. Specifically, the position profit and loss standard can be an important module of the position main function, and other modules corresponding to the main function include: position flow calculation, contract position profit and loss accounting, exchange-traded option position profit and loss accounting, bond interest accrual, and end-of-day profit and loss accounting. The main function corresponding to the position profit and loss standard consists of nearly 40 sequentially executed batch statistics. Each batch calculates various statistical indicators for proprietary accounts for a specific set of business scenarios, including: quantity measured, position cost, capital occupation, price difference income, interest income, and expense expenditure. The position profit and loss standard corresponding to the accounting process can include two steps: "generating the position profit and loss flow table" and "calculating position profit and loss statistical indicators".

[0023] The step "Generate Position Profit and Loss Statement" can be achieved by supplementing the basic position statement with relevant information. Specifically, before running each accounting batch, the preprocessing stored procedure filters data from the basic position statement and outputs three input statements (a statement recording normal business processing flow, a statement recording short selling business processing flow, and a statement with the same representation as the basic position statement, recording all original flow). In practice, each business batch uses a specific input statement as the input information for calculation, depending on the specific circumstances of the accounting business scenario. The step "Calculate Position Profit and Loss Statistical Indicators" corresponds to the input of the position profit and loss statement, the original position profit and loss statement (mainly contributing fields containing the keyword "OLD" in the formula), and the fair price table, and outputs the position profit and loss statement and the position profit and loss change statement. The position profit and loss criteria include: Criterion 1, Criterion 2, Criterion 3, and Criterion 4. Rule 1 can be a buy-first-sell-later rule (mainly for securities trading). Rule 2 can be a transfer-out-later-transfer-in rule (mainly for cost transfer transactions). Rule 3 can be a transfer-out-later-transfer-in rule (mainly for non-aggregated liability transactions). Rule 4 can be a business type conversion rule (mainly for aggregated liability transactions). Rule 1, for securities trading, involves multiple transactions with the same account, market, and security code each day. The buy transactions need to be centrally calculated and the profit and loss of the holdings updated first, followed by the sell transactions. The "buy-first-sell-later" rule uses a moving weighted average method to calculate the change in capital occupied by the sell transactions, is reasonably influenced by the change in capital occupied by the buy transactions on the same day, and conforms to the business accounting logic. The transfer-out-later-transfer-in rule, for cost transfer transactions, requires calculating the change in capital occupied by the transfer-out transactions first, and then calculating the change in capital occupied by the transfer-in transactions. The purpose is to ensure that the sum of the change in capital occupied by the transfer-in transactions and the change in capital occupied by the transfer-out transactions is zero. The "transfer out first, transfer in later" principle can be based on the different statistical needs for various liability transactions, dividing them into non-aggregated liability transactions and aggregated liability transactions, with corresponding differences in accounting methods. For non-aggregated liability transactions, the original transaction categories in the input log remain unchanged during the accounting process, and the changes in funds for both normally held assets and lent assets are simultaneously accounted for for each transaction type. The transaction type conversion principle can be based on the aggregated liability transaction accounting process, where the transaction logs of multiple transaction types are first aggregated to generate a new aggregated transaction type (this is usually a requirement of the business department), and then statistical indicators are calculated for the aggregated transaction type.

[0024] It should be noted that the various business processing guidelines in the business processing guidelines information set can be summarized based on the business knowledge corresponding to the target business system.

[0025] Step 102: Based on the code information, use the first large language model deployed on the first server to generate initial full business process information for each business batch.

[0026] In some embodiments, the aforementioned execution entity can generate initial full business process information for each business batch based on the code information and using the first large language model deployed on the first server. The first server can be a server deployed with a large language model. For example, the first server can have the following basic configuration: "GPU: ≥80GB VRAM or multi-card parallelism (e.g., 2-6 A100 cards)," "CPU: 32 cores or more," "Memory: 256GB DDR5," and "Storage: 2TB NVMe SSD." The first large language model can be a large language model capable of code parsing and generating business process diagrams. In practice, the first large language model can be a large language model based on the MoE architecture (e.g., the DeepSeek series). In practice, the large language model based on the MoE architecture dynamically activates some expert sub-networks at the feedforward network (FFN) layer by introducing a routing mechanism (e.g., Top-k routing). The advantage of the large language model based on the MoE architecture is that it can significantly increase the model capacity (the number of parameters can reach hundreds of billions or even trillions) while strictly controlling the computational overhead of a single forward / backward propagation. The MoE architecture's large language model can first map the input to a low-dimensional latent space, and then perform lightweight attention operations on this compressed representation. In terms of computational efficiency, the MoE architecture's large language model can reduce quadratic complexity, making it particularly suitable for long text modeling and code generation. The training phase of the first large language model includes: a pre-training phase, an instruction fine-tuning phase, and a reinforcement learning optimization phase. The thought chain length of the first large language model satisfies the target length condition. Specifically, for the first large language model with the MoE architecture, one or two most relevant expert subnetworks can be selected based on the input content. Then, for the selected expert subnetworks, the problem is decomposed into multiple reasoning steps according to the thought chain pattern for step-by-step derivation. Finally, the reasoning results of each expert are weighted to obtain the final output. A business batch can be the smallest business unit. That is, each business batch has a unique corresponding business, corresponding to different business matters. For example, a business batch can be a business unit, representing the corresponding business step (i.e., the business execution logic in a certain scenario). For example, a business batch can be holding profit and loss, bond interest accrual, etc.

[0027] The initial full business process information can be the initially generated process information representing the execution flow corresponding to each business batch in the target business system. In practice, the initial full business process information can be graph-based architecture information. The thought chain length corresponding to the first large language model satisfies the target length condition. The thought chain length can be the maximum depth at which the model can maintain coherent reasoning and logical chains when completing a task, usually measured by the number of reasoning steps. In practice, the target length condition can be that the number of steps corresponding to reasoning steps exceeds a target threshold. The target threshold can be a threshold used to measure whether the large model is engaging in deep thinking. That is, if the number of steps corresponding to reasoning steps exceeds the target threshold, it indicates that the large model is engaging in deep thinking.

[0028] As an example, firstly, the aforementioned executing entity can fill in the code information into the process generation prompt template to obtain process generation prompts. Then, the process generation prompts are input into the first major language model to obtain the initial full business process information.

[0029] In some optional implementations of certain embodiments, the business processing rule information in the business processing rule information set is generated through the following steps: The first step is to obtain the code block corresponding to the business processing rule information, which will be used as the rule code block. This code block can be the code content representing the business rule logic corresponding to the business processing rule information.

[0030] The second step is to extract the business description information and key content of the business processing guidelines corresponding to the business processing guidelines information. The business description information can be the guideline description content of the business processing guidelines information. The key content of the guidelines can be the key field content in the business processing guidelines information.

[0031] The third step involves using the second language model deployed on the second server to perform the following generation steps: Sub-step 1 involves searching for the business processing content corresponding to the business description information and the rule-related content corresponding to the rule key content. The rule-related content includes: external rule code blocks; the second major language model is obtained through transfer learning based on the first major language model, and the thought chain length corresponding to the first major language model is longer than that corresponding to the second major language model. Rule-related content can be related content corresponding to the rule key content. For example, for rule key content consisting of various keywords, the rule-related content can be a summary of related content retrieved based on each keyword. External rule code blocks can be code blocks retrieved from an external knowledge base that are related to the business processing rules. Business processing content can be the business processing situation corresponding to the business description information.

[0032] Sub-step 2 involves parsing the code logic of the rule code block and the external rule code block to obtain the initial business processing rule information.

[0033] Sub-step 3 involves sending the business processing content and the rule-related content to the first server to generate a rule-related question set and a response result set based on the deployed first large language model. There is a one-to-one correspondence between the rule-related questions in the rule-related question set and the response results in the response result set. Rule-related questions can be questions posed regarding business processing rule information. Response results can be answers to the rule-related questions.

[0034] The fourth step involves verifying and converting the initial business processing rule information based on the business processing content, the rule-related content, the rule-related question set, and the response result set, to obtain the business processing rule information.

[0035] In some optional implementations of certain embodiments, the executing entity may, based on the code information and utilizing a first large language model deployed on a first server, generate initial full business process information for each business batch, including the following steps: The first step is to extract the business batch set and the corresponding field group for each business batch based on the code information. The individual business batches in the business batch set can be extracted from the code information. The field group can be the key fields from the business content corresponding to each business batch.

[0036] As an example, the aforementioned execution entity can utilize the first major language model to extract content related to business batches from the code comments corresponding to the code information, thereby obtaining a business batch content set. It then determines the business batch set corresponding to the business batch content set and extracts the corresponding field groups.

[0037] The second step involves, in response to determining that the service batch set includes the various service batches, extracting the code blocks corresponding to each service batch from the code information as service code blocks. Here, the fact that the service batch set includes the various service batches indicates that each service batch is a subset of the service batch set.

[0038] As an example, based on the code comment information, the first major language model can be used to extract the code blocks corresponding to each business batch from the code information as business code blocks.

[0039] The third step involves, in response to determining that the service batch set does not include the various service batches, obtaining at least one service batch from the various service batches that is not present in the service batch set. The fact that the service batch set does not include the various service batches indicates that the various service batches in the service batch set are incomplete. That is, there is missing code content in the code information, resulting in the exclusion of code content corresponding to a portion of the service batches in the service batch set. In other words, code content corresponding to at least one service batch is missing.

[0040] Fourth, for each of the at least one business batch, based on the field group corresponding to the business batch, filter out the associated business batch group that has a business relationship with the business batch from the business batch set. Here, an associated business batch can be a business batch whose business content is related to the corresponding business content of the business batch. That is, an associated business batch can represent businesses with related content.

[0041] Fifth, based on the local code blocks corresponding to the obtained associated business batch set and at least one field group corresponding to the at least one business batch, the first large language model is used to generate local code blocks corresponding to the at least one business batch as supplementary code blocks. Each business batch has a corresponding field group, and the supplementary code blocks can be code describing the code logic corresponding to at least one business batch.

[0042] Step 6: Generate business code blocks corresponding to each business batch based on the supplementary code blocks and the code information.

[0043] As an example, the aforementioned execution entity can combine supplementary code blocks and code information at corresponding locations to obtain business code blocks.

[0044] Step 7: Based on the business code block, use the first large language model to generate initial full business process information.

[0045] In some optional implementations of certain embodiments, before step 102, the steps further include: The first step is to obtain the code volume and process generation task information corresponding to the code information. The code volume can be the number of lines of code corresponding to the code information. The code volume can represent the complexity of the code. The process generation task information can be the task content for generating the process.

[0046] The second step involves generating task information based on the code size and the process flow. A pre-trained classification algorithm for difficulty classification is then used to determine the processing difficulty of the task corresponding to the generated task information. The difficulty classification algorithm can determine the task complexity of the generated process given the code size. In practice, processing difficulty can be presented as a numerical value; a higher processing difficulty value indicates a higher level of difficulty in the generated process. In practice, the classification algorithm can be a linear regression algorithm or a regression algorithm based on fully connected layers.

[0047] Thirdly, in response to the processing difficulty being higher than a first difficulty value and the current model's resource usage being monitored by the model resource monitor meeting the resource conditions, it is determined that the first major language model deployed on the first server will be used to complete the process generation task. Here, the first difficulty value can be a numerical value used to determine whether the difficulty of generating the process is high. The first difficulty value can be a pre-set value. A value higher than the first difficulty value indicates a higher difficulty in generating the process. The model resource monitor can be a monitoring device that monitors the model's resource usage in real time (e.g., it can monitor the server). The current model's resource usage can be the computing resources that the model can access at the current time. The resource conditions can be that the current model's resource usage supports calling the first major language model to execute the process generation task.

[0048] Step 103: Based on the business processing rule information set, verify the execution relationship between business batches of the initial full business process information to obtain the full business process information.

[0049] In some embodiments, the aforementioned executing entity can verify the execution relationships between business batches of the initial full business process information based on the business processing rule information set, thereby obtaining full business process information. Verifying the execution relationships between business batches can involve verifying the business associations between business batches during execution to determine if there are any issues with the execution logic between business batches. The full business process information can be based on the business processing rule information set to verify the initial business process information and modify the processed accurate business process.

[0050] As an example, firstly, processing prompts are generated to verify and correct the execution relationships between business batches based on the business processing criteria information set. Then, the processing prompts are input into the first large language model to obtain the full business process information.

[0051] In some optional implementations of certain embodiments, the executing entity may verify the execution relationship between business batches of the initial full business process information based on the business processing rule information set to obtain the full business process information, including the following steps: The first step is to filter out a subset of business processing rule information corresponding to each business batch from the set of business processing rule information. Each business processing rule information in the subset is a set of rules that are related to each business batch.

[0052] As an example, a subset of business processing criteria information is obtained by filtering out accurate business processing information from the corresponding business batch information that is the same as any business batch in each business batch.

[0053] The second step is to obtain the importance information of the criteria for the subset of business processing criteria information set on the process page rendering request page. The process page rendering request page can be a page that pops up after clicking the rendering request control on the process page. This page can be a page for filling in some requirements during the process rendering. These requirements may include: rendering requirements and the importance of criteria. In practice, the importance of criteria can be determined by the various labels for each criterion on the process page rendering request page. Different labels represent different levels of criterion importance. For example, the labels corresponding to a criterion can be one of the following: first label, second label, and third label. A business criterion under the first label is more important than a business criterion under the second label. A business criterion under the second label is more important than a business criterion under the third label. The criterion importance information can be the labels corresponding to each business processing criterion in the subset of business processing criteria information. The criterion importance information can represent the importance of each business processing criterion in the subset of business processing criteria information.

[0054] The third step involves determining whether to invoke the second major language model in the second server based on the importance information of the criteria and the criterion complexity corresponding to the subset of business processing criterion information. Criterion complexity characterizes the criterion complexity of each business processing criterion within the subset of business processing criterion information. In practice, the more criterion content, formulas, parameters, and associated business batches a business processing criterion corresponds to, the more complex the corresponding business processing criterion becomes. Criterion complexity can be generated by comprehensively considering the criterion complexity corresponding to each piece of business processing criterion information. For example, the criterion complexity can be a value between 1 and 100. The higher the value, the more complex the criterion content. Criterion complexity can also be obtained by weighting the complexity of each criterion. Furthermore, criterion complexity can be determined based on a preset complexity formula. For example, the complexity formula can be a weighted formula that assigns values ​​based on the frequency of formula occurrence, the amount of criterion content, and the number of associated business batches.

[0055] As an example, in response to determining that the importance of the criteria indicates that there are more than a target number of business processing criteria whose corresponding labels are the first label and / or whose criterion complexity is higher than the target complexity, the second large language model is invoked. In response to determining that the importance of the criteria indicates that there are more than a target number of business processing criteria whose corresponding labels are not the first label and whose criterion complexity is lower than the target complexity, the relevant criterion verification module can be invoked (a module that performs automated criterion verification based on criterion matching rather than the large language model).

[0056] The fourth step, in response to the confirmed call, utilizes the second large language model to perform rule information summary output and rule information sorting on the subset of business processing rule information, resulting in a rule information group sequence and a rule summary group sequence. The relationships between each rule information in the rule information group are adjusted synchronously. The rule information summary output can generate summary content corresponding to the business processing rule information. The rule information sorting can determine the sequential verification order of each business processing rule information. The sequential verification order differs depending on the business batch corresponding to the business processing rule information. There is a one-to-one correspondence between the rule information in the rule information group sequence and the rule summaries in the rule summary group sequence. Each rule information in the rule information group is processed for verification simultaneously. The rule information groups in the rule information group sequence are verified sequentially. For example, the rule information group sequence includes: a first rule information group, a second rule information group, and a third rule information group. Therefore, verification is performed first for each rule information in the first rule information group, then for each rule information in the second rule information group, and finally for each rule information in the third rule information group.

[0057] Here, by outputting a summary of the rule information, it is determined whether the second language model fully understands the rule content during the learning process of each business process rule information. Only by ensuring that the second language model fully understands the rules can the accuracy of the rule information group sequence be guaranteed. In addition, the second language model can group rule information processing the same business batch together to facilitate subsequent rule verification. Based on the execution order of the business batches, the rule groups are sorted to obtain the rule information group sequence.

[0058] In the specific implementation, corresponding prompt words can be set to generate a sequence of criterion information groups and a sequence of criterion summary groups.

[0059] Fifth, based on the criterion summary group sequence, generate the sequence order accuracy value corresponding to the criterion information group sequence. The sequence order accuracy value characterizes the accuracy of each accurate information group in the accurate information group sequence. The sequence order accuracy value can be in numerical form. The higher the sequence order accuracy value, the more accurate the sequence order corresponding to the criterion information group sequence.

[0060] As an example, firstly, the criterion summary group sequence, the subset of business processing criterion information, and the corresponding prompts regarding the group partitioning accuracy and sequence accuracy of the determined accurate summary group sequence are sent to the first server. The first server then uses the first large language model to generate the group partitioning accuracy and sequence accuracy for the criterion summary group sequence. Next, the obtained group partitioning accuracy and sequence accuracy are weighted to obtain the sequence order accuracy value.

[0061] Step 6: In response to the sequence order accuracy value being higher than the target accuracy value, the execution relationship between business batches of the initial full business process information is adjusted according to the criterion information group sequence to obtain the full business process information. The target accuracy value can be a pre-set accuracy rate. The target accuracy value can be a numerical value used to measure the accuracy of the criterion information group sequence. For example, the target accuracy value could be 97%.

[0062] As an example, the aforementioned executing entity can, in accordance with the sequential order of each criterion information group in the criterion information group sequence, use the second major language model to adjust the execution relationship between the business batches corresponding to the criterion information groups in the initial full business process information to obtain the full business process information.

[0063] In some optional implementations of certain embodiments, the executing entity may adjust the execution relationship between business batches of the initial full business process information based on the criterion information group sequence to obtain the full business process information, including the following steps: The first step, for the target criterion information group in the criterion information group sequence, is to perform the following adjustment steps: Sub-step 1 involves filtering out code blocks related to the target criterion information group from the code information, and designating them as target code blocks. The target code block can be code logic content reflecting at least one business batch corresponding to the target criterion information group. The target criterion information group can be a combination of various criterion information currently being verified. The code information consists of the full source file code.

[0064] Sub-step 2: Obtain the full business process processing information corresponding to the previous criterion information group. The previous criterion information group can be the last criterion information group that was adjusted, corresponding to the target criterion information group.

[0065] Sub-step 3: For each target criterion information, based on the target code block and the target criterion information, using the second large language model, perform business execution content verification on multiple business batches related to the target criterion information in the full business process processing information, and obtain a verification result. The verification result is one of the following: a verification result that passes the verification, a verification result indicating that the code lacks logical support, or a verification result indicating that there is redundant constraint logic. Specifically, a verification result that passes the verification can be a verification result indicating that the code logic corresponding to the code block matches the criterion content corresponding to the target criterion information. A verification result indicating that the code lacks logical support can be a verification result indicating that the code block lacks corresponding criterion logical support. A verification result indicating that there is redundant constraint logic can be a verification result indicating that there is redundant criterion information.

[0066] In practice, prompt word technology can be used to verify the business execution content of multiple business batches related to the target criteria information in the full business process processing information based on the target code block and the target criteria information, using the second major language model, and obtain the verification results.

[0067] Sub-step 4: Based on the obtained verification result group, generate the target full business process processing information corresponding to the target criterion information group. The target full business process processing information can be the full business process processing information corresponding to the target criterion information.

[0068] As an example, based on the verification result group, the second largest language model is used to process the verification results in each verification result group that have problems, so as to adaptively adjust the corresponding full business process processing information and obtain the target full business process processing information.

[0069] Sub-step 5: In response to the target criterion information group being the last criterion information group in the sequence, the target full business process processing information is determined as full business process information.

[0070] The second step is to determine the next criterion information group corresponding to the target criterion information group as the target criterion information group in response to the target criterion information group not being the last criterion information group in the sequence, and continue to execute the adjustment steps.

[0071] In some optional implementations of certain embodiments, before step 103, the steps further include: The first step is to determine the semantic complexity and the number of rules corresponding to the business processing rule information set. The semantic complexity (which can be numerical information) refers to the rule complexity of each business processing rule in the business processing rule information set. The number of rules refers to the number of individual rules included in the business processing rule information set.

[0072] The second step is to determine the adjustment difficulty corresponding to the execution relationship adjustment task based on the semantic complexity and the number of criteria. Details will not be elaborated further.

[0073] Third, in response to the adjustment difficulty being lower than the first difficulty value but higher than the second difficulty value, it is determined that the second large language model deployed on the second server is used to complete the relation adjustment task.

[0074] Step 104: Using the first large language model, generate page rendering information for the full business process information, and render and display the full business process information on the process page according to the page rendering information.

[0075] In some embodiments, the aforementioned executing entity can utilize the first large language model to generate page rendering information for the entire business process information, and render and display the entire business process information on the process page based on the page rendering information. The page rendering information can be the rendering content for rendering the entire business process information. In practice, page rendering information may include: node data, edge data, layout configuration, style definitions, and interaction logic. Node data may include: node identifier, node content, node position, node status, and node attributes. Edge data may include: connection relationships, edge types, path information, and conditional expressions. Layout configuration may include: layout algorithm, layout parameters, layout direction, whether automatic layout is enabled, and layout stability. Style definitions may include: node styles, edge styles, theme configuration, state styles, and CSS variables. Interaction logic may include: event listeners, interactive feedback, dynamic updates, user operations, and state synchronization.

[0076] As an example, firstly, rendering prompts are generated to display the entire business process information on the page. Then, the rendering prompts are input into the first language model to obtain the page rendering information.

[0077] In adopting a technical solution to address the technical problems mentioned in the background, and considering the application scenario—rendering pages of full business process information—the following technical issues arise: when the full business process information corresponds to a large amount of process content and involves multi-stage process rendering, synchronous rendering at multiple stages leads to numerous rendering errors at each page level, resulting in chaotic process rendering. We have decided to adopt the following solution: In some optional implementations of certain embodiments, the executing entity may utilize the first large language model to generate page rendering information for the full business process information, including the following steps: The first step is to obtain the rendering requirements information for each stage of the multi-stage process information for the entire business process. Each stage has at least one rendering requirement, and there are process content relationships between the stages. Furthermore, there are also process content relationships between the rendering requirements. That is, the process content relationships can connect not only the relationships between stages within the multi-stage process but also the relationships between multiple rendering requirements. The multi-stage process can refer to the various stages that render the entire business process information. Here, by performing multi-stage process rendering, the hierarchical rendering of the process content at different page levels can be ensured, avoiding the situation where multiple stages of synchronous rendering lead to numerous rendering errors at various page levels. Each stage has a corresponding set of rendering requirement information. The rendering requirement information can be the rendering requirement for a specific process information within the entire business process information. Specifically, each rendering requirement can be set on the relevant process rendering page.

[0078] The second step is to generate a process rendering requirement information graph based on the various process rendering requirement information. In this graph, nodes represent process rendering requirement information, and edges represent the process content relationships between these requirement information. It's important to note that the process rendering requirement information graph has a multi-layered structure, with each layer corresponding to different process rendering requirement information within a specific stage. These multi-layered structures are connected through process content relationships. For example, the process rendering requirement information graph can be a three-layered structure, including: a first process rendering requirement subgraph at the top, a second process rendering requirement subgraph in the middle, and a third process rendering requirement subgraph at the bottom. Each of these subgraphs contains multiple corresponding process rendering requirement information. Nodes in each subgraph represent process rendering requirement information, and edges represent the process content relationships between connected requirement information. Specifically, the flowchart rendering of the first process rendering requirement subgraph can be performed first, followed by the second, and finally the third. The first, second, and third process rendering requirement subgraphs are connected by edges based on the process content relationships.

[0079] The third step involves constructing multi-stage process rendering prompts based on the aforementioned process rendering requirement information graph. Each stage has a corresponding set of process rendering prompts. These prompts can be used to simultaneously prompt the first language model to perform process rendering (i.e., to perform process rendering processing synchronously). Alternatively, they can be used to render specific parts of the process content.

[0080] Fourth, for the target stage in the multi-stage process, perform the following information generation steps: Sub-step 1: Determine the process rendering prompt phrase corresponding to the target stage, and use it as the target process rendering prompt phrase.

[0081] Sub-step 2 involves inputting the target flow rendering prompt phrases and the previously generated rendering results corresponding to the target stage into the first large language model to obtain the page rendering sub-information corresponding to the target stage. Each rendering result can be the completed flow rendering result corresponding to each stage rendered before the target stage. The page rendering sub-information can be the rendering content to be rendered on the page. The rendering content corresponding to the page rendering sub-information is the rendering execution result of the rendering requirements information for each flow corresponding to the target stage. The page rendering sub-information can include: page rendering component information and component layout information.

[0082] Here, by inputting the previously generated rendering results corresponding to the target stage, the first language model can be prompted to render the various process contents under the target stage based on the previously generated rendering results, ensuring the continuity of the rendering content between the target stage and the previously rendered stages.

[0083] Sub-step 3 involves rendering the page on the process page based on the page rendering sub-information to obtain a rendering result. The rendering result can be the page display effect of the page rendering sub-information on the process page.

[0084] Sub-step 4: In response to the rendering result passing verification and the target stage being the last of the multiple stages, an association graph is constructed for each obtained rendering result to obtain a rendering effect diagram, which serves as the page rendering information. The verification of the rendering result can be a representation of the relevant objects after confirmation based on visual effects. Each rendering result can be a result generated in the target stage or at least one previous stage.

[0085] Fifth step: In response to the rendering result passing the verification and the target stage not being the last stage in the multi-stage process, the next stage corresponding to the target stage is taken as the target stage, and the information generation step is continued.

[0086] Here, "steps one through five" serves as another inventive point of this disclosure. Specifically, this disclosure first obtains rendering requirement information for each stage of the multi-stage process of the entire business process information to ensure the accuracy of subsequent page rendering information and the generation of the process rendering requirement information diagram. Then, by generating the process rendering requirement information diagram, the key content used in rendering the entire business process information can be efficiently displayed. Finally, through prompt word technology, the rendering results of each stage in the multi-stage process are generated sequentially, ensuring the hierarchical rendering of each process content at different page levels and avoiding the occurrence of numerous rendering errors at each page level due to simultaneous multi-stage rendering.

[0087] Step 105: In response to selecting at least one scenario field on the process page, based on the at least one scenario field and the full business process information, using the first large language model, generate a business batch matching the at least one scenario field as the first business batch.

[0088] In some embodiments, in response to the selection information of at least one scenario field selected on the process page, the execution entity can generate a business batch matching the at least one scenario field, as the first business batch, based on the at least one scenario field and the full business process information, using the first large language model. The scenario field can be a field related to a scenario. Here, the scenario is a scenario related to the target business system. For example, the scenario field can be... Here, the selection information can be inputting at least one scenario field, or selecting the corresponding at least one scenario field from at least one field selection box. The matched business batch can be one where the business content corresponding to the business batch matches the semantic content of the field corresponding to at least one scenario field. It should be noted that the business batch matching at least one scenario field can be at least one. That is, the first business batch can include multiple business batches.

[0089] As an example, the aforementioned execution entity can utilize the first major language model to extract the comprehensive semantic information of at least one scenario field and the semantic information of each business batch. Then, it can select the business batch with the highest semantic similarity between the comprehensive semantic information of the scenario field and the semantic information of the business batch from among the various business batches, and designate it as the first business batch.

[0090] like Figure 2 This shows a schematic diagram illustrating the content display of business batches under various scenario fields.

[0091] See Figure 2After selecting "First Scenario Field", "Second Scenario Field", "Third Scenario Field" and "Fourth Scenario Field" on the process page, the batch content corresponding to the business batch will be displayed in the page area below.

[0092] The rectangles within each displayed business batch represent the calculation logic and function of each business data point or content, while the edges represent the processing steps between two data points. For example, business data 1, after processing steps 1 and 2, generates business data 2 and business data 3. Business data 3, after processing steps 3 and 4, generates business data 4 and business data 5 respectively. Following the "target business content calculation logic," business data 6 is generated. Business data 5 and the business data are processed by function 1 to obtain business data 7 and business data 8. Business data 8, after processing step 5, outputs business data 0. Clicking the edge corresponding to processing step 2 on the workflow page will navigate to the storage content illustration page of the corresponding stored procedure for processing step 2.

[0093] like Figure 3 As shown, this is a schematic diagram illustrating the content displayed after selecting a function.

[0094] See Figure 3 Clicking the rectangle corresponding to function 1 on the process page will take you to the display page for the various rules associated with function 1. This page shows rule 1, rule 2, rule 3, and rule 4. Selecting rule 1 will navigate to the rule content for the corresponding scenario: "First Scenario Field," "Second Scenario Field," "Third Scenario Field," or "Fourth Scenario Field." Specifically, the rule content can include: the formula corresponding to rule 1, the formula parameters corresponding to rule 1, and the definition of the formula parameters corresponding to rule 1.

[0095] like Figure 4 As shown, this is a schematic diagram illustrating the content display after selecting the target business content calculation logic.

[0096] See Figure 4 Clicking "Target Business Content Calculation Logic" on the process page will take you to the display page for the corresponding calculation logic. The corresponding calculation logic can be... Figure 4 The calculation and judgment logic of each node is shown in the figure.

[0097] In addressing the technical problems mentioned in the background section, and considering the application scenario—where users select scene fields by clicking various scene selection controls on a workflow page, with each control corresponding to a large number of scene fields—the following technical issues arise: the large number of selectable scene fields for each control necessitates users scrolling up and down to determine the desired field, resulting in a time-consuming and laborious process and a poor user experience. Given the following requirements for this application scenario, particularly the large number of scene fields, we have decided to adopt the following solution: In some optional implementations of certain embodiments, before generating a business batch matching the at least one scenario field as the first business batch based on the at least one scenario field and the full business process information using the first large language model in response to selecting at least one scenario field on the process page, the method further includes: The first step involves responding to the selection of a first scene selection control from at least one scene selection control displayed on the process page, which then pops up a list of candidate scene fields under the first scene selection control. There is a one-to-one correspondence between the scene selection controls in the at least one scene selection control and the scene fields in the at least one scene field. The scene selection control can be a control for selecting a scene field. Each candidate scene field can be a field selectable under the first scene selection control.

[0098] The second step involves, in response to receiving a scene field selected from the candidate scene fields, replacing the first scene selection control displayed on the process page with the selected scene field. That is, the scene field is displayed in the display bar corresponding to the first scene selection control.

[0099] The third step is to remove the first scene selection control from the at least one scene selection control to obtain a scene selection control set. The scene selection controls in the scene selection control set can be controls that have not yet selected a scene field.

[0100] Fourth, for the selected control set in the scene, perform the following processing steps: Sub-step 1 involves updating the field space of the candidate scene field set corresponding to each scene selection control in the scene selection control set, based on the mapping relationship tree between scene fields. The mapping relationship tree represents the mapping relationships between scene fields. Each mapping relationship in the mapping relationship tree can be determined based on business logic. For example, the mapping relationship could be "After scene field 1 is displayed, scene fields 2 and 3 are not supported for display." The field space can be the various fields that can be selected under the scene selection control. Each scene selection control has a corresponding field space, which is limited by the scene fields selected by other scene selection controls. The corresponding field space will be shrunk to ensure that the scene fields selected under this scene selection control match the already selected scene fields.

[0101] As an example, firstly, the aforementioned execution entity can obtain the various scene fields that have already been displayed. Then, it filters out the mapping relationships corresponding to the various scene fields that have already been displayed from the mapping relationship tree. Finally, based on each mapping relationship, it updates the field space of the candidate scene field set corresponding to each scene selection control to avoid the appearance of scene fields with opposing mapping relationships in the field space.

[0102] Sub-step 2: In response to selecting the second scene selection control from the scene selection control set displayed on the process page, pop up each candidate scene field in the updated field space corresponding to the second scene selection control.

[0103] Sub-step 3: In response to receiving a scene field selected from each candidate scene field, the second scene selection control displayed on the process page is replaced with the scene field.

[0104] Sub-step 4 involves removing the second scene selection control from the at least one scene selection control to obtain a set of scene selection controls after removal. Further details are omitted.

[0105] Sub-step 5: In response to the removed scene selection control set being empty and / or no field selection information being received within the target time period, each scene field displayed on the process page is determined to be at least one scene field. The target time period can be a period of time after the current time. The field selection information can be request information for selecting a scene field.

[0106] Fifth step: In response to the fact that the removed scene selection control set is not empty and field selection information is received within the target time period, the removed scene selection control set is used as the scene selection control set, and the scene selection control set is continued to be executed.

[0107] Here, "step one to step five" is another inventive point of this disclosure. This disclosure automatically updates the field space corresponding to the remaining scene selection controls that have not yet selected scene fields by using a mapping relationship tree after the user selects scene fields for the scene selection control. This reduces the display of subsequent mutually exclusive scene fields, making the selection efficiency of subsequent scene fields increasingly higher and improving the user experience.

[0108] Step 106: Display the stored business process processing tree corresponding to the first business batch on the process page.

[0109] In some embodiments, the aforementioned execution entity may display the stored business process processing tree corresponding to the first business batch on the process page. The business process processing tree can represent the business process logic corresponding to the first business batch.

[0110] In some optional implementations of certain embodiments, after step 106, the steps further include: The first step involves responding to a question / inquiry about a specific part of the overall business process information selected on the process page. Using the first large language model, a response is generated corresponding to the question / inquiry. The specific part of the business process information can be a portion of the overall business process information. For example, it could be a branch process within the overall business process information. The question / inquiry can be a question raised regarding the specific part of the business process information. For example, the question / inquiry could be: "Provide the business logic formula for the selected specific part of the business process." The response is a reply to the question / inquiry.

[0111] It should be noted that for selecting specific business process content, you can take a screenshot or double-click on the process page to select the content.

[0112] As an example, the aforementioned implementing entity can utilize the business Q&A function corresponding to the first major language model to generate response content for the inquiry information.

[0113] The second step involves responding to the input of business search information in the corresponding search bar on the process page, and generating a response based on the first large language model. The search bar can be a component for searching business content. The business search information can be content related to searching for business-related knowledge.

[0114] As an example, the aforementioned implementing entity can utilize the business question-and-answer function corresponding to the first major language model to generate response content corresponding to business search information.

[0115] The third step involves responding to a query on the process page that selects at least one business processing rule information corresponding to a specific business process content. Using the first language model, descriptive information for the at least one business processing rule information is generated. The query information can be a query request for various related business processing rule information corresponding to the specific business process content. The descriptive information may include: rule identifiers corresponding to at least one business processing rule information, and key descriptive content of the rule corresponding to at least one business processing rule information. The specific implementation method is not detailed here, but the corresponding descriptive content can be generated based on the process understanding of the entire business process information using the first language model.

[0116] The fourth step involves responding to the selection information for the second business batch on the process page, and determining the business process processing content corresponding to the second business batch from the full business process information. The second business batch can be a business batch to be highlighted. In practice, a description of the second business batch or an identifier corresponding to the second business batch can be entered on the process page for subsequent highlighting of the corresponding business processing content. The business process processing content here can be a business process chain including process nodes corresponding to the second business batch.

[0117] Fifth, highlight the process information corresponding to the business process content displayed on the process page. This highlighting can be achieved by using a yellow background to highlight the corresponding business process link.

[0118] In adopting technical solutions to address the technical problems mentioned in the background, and considering the application scenario—maintaining business process content—the following technical issues arise: the business process content is large in volume, and updates cannot be detected promptly, resulting in content lag and slow update speed. Given the following requirements for this application scenario: large update volume and failure to detect updates promptly, we have decided to adopt the following solution: In some optional implementations of certain embodiments, after step 106, the steps further include: The first step involves setting up a first business processing agent to support the maintenance of business process content, in response to determining that the business level corresponding to the target business system is the target level. The business level characterizes the importance of the target business system. The higher the corresponding business level, the higher the importance of the target business system. The target level can be used to measure whether the target business system is a high-importance business system. The first business processing agent can be an agent that supports the automated processing of various businesses corresponding to the target business system. In practice, the first business processing agent can be a reinforced agent based on a first large language model after transfer learning. In practice, the first business processing agent can be an agent trained on a large amount of business data. The first business processing agent can encapsulate a large language model to process various businesses. The first business processing agent can be a self-developed agent based on business processes.

[0119] The second step involves, in response to the detection that the current time is a business process update time, acquiring the pre-selected stored actual business dataset and business enhancement dataset. The business enhancement dataset is generated based on the first business processing agent and validated by the second business processing agent. The business process update time can be the time when business process information is updated. Actual business data can be data generated in actual business operations. Business enhancement data can be data derived from actual business data using an agent and validated. The second business processing agent can be an agent specifically designed to validate and process business data information. The second business processing agent can also be an agent used to ensure the accuracy and usability of business data logic. Unlike the first business processing agent, the large language model corresponding to the second business processing agent can learn based on more business data logic knowledge and business validation knowledge. Alternatively, it can learn using the system's internal basic system knowledge to ensure that the second business processing agent can effectively and accurately determine the usability of business data. The model architecture of the large language model deployed for the second business processing agent can be the same as that for the first business processing agent, but its primary responsibilities differ. The second business processing agent can be a business processing agent that is developed internally and validated with a large amount of data before being launched internally.

[0120] The third step is to instruct the first business processing agent to perform the following steps: Sub-step 1: Based on the real-time business dataset and the business enhancement dataset, generate process change information and new rule information for the entire business process information. Process change information can be changes to the content of the business process. New rule information can be newly added business rule information.

[0121] As an example, leveraging the process generation and rule generation capabilities of the large language model deployed by the first business processing agent, and based on prompt word generation technology, process change information and new rule information for the full business process information are generated according to the real-time business dataset and business enhancement dataset.

[0122] Sub-step 2: Send the process change information and the new criteria information to the second business processing intelligent agent to verify the process change content and the new criteria content.

[0123] Sub-step 3: In response to confirming that the process change information and the new criterion information are verified to be correct, the process information of the entire business process is adjusted according to the process change information to obtain the adjusted process information of the entire business.

[0124] Sub-step 4 involves verifying the execution relationship between business batches based on the newly added criteria information to obtain the full business adjustment process information. The specific implementation method will not be elaborated further.

[0125] In some optional implementations of certain embodiments, the actual business dataset and the business enhancement dataset are obtained through the following steps: The first step involves acquiring the first set of process questions and the first set of candidate criteria information input by the target object, as well as the second set of process questions and the second set of candidate criteria information generated by the first business processing agent based on a knowledge base. The target object can be an entity that processes the business process. For example, the target object can be a professional who processes the business process. The first set of process questions can consist of various questions posed regarding the business process.

[0126] The second step involves labeling and fusing the first process problem set and the second process problem set to obtain a process problem set, and labeling and fusing the first candidate criterion message set and the second candidate criterion information set to obtain a candidate criterion information set.

[0127] The third step is to instruct the first business processing agent to perform the following steps: Sub-step 1 involves identifying the third business processing intelligent agents corresponding to the first business processing intelligent agent in each of the external business systems with which they have communication interactions. Each external business system has a corresponding third business processing intelligent agent. The external business system corresponds to the target business system in terms of system responsibilities, but is deployed for different objects. The third business processing intelligent agent is an intelligent agent with the same functions and generation method as the first business processing intelligent agent. The difference between the two lies in the objects they serve.

[0128] Sub-step 2 involves sending the process question set and the candidate criterion information set to each of the third business processing agents to generate question-response sets for the process question set and criterion-response sets for the candidate criterion information set. Each question response and each criterion response has a corresponding illusion tag. There is a one-to-one correspondence between each third business processing agent and each question-response set in the question-response set. That is, each third business processing agent generates a corresponding question-response set for the process question set. Question responses can be the results of answering process questions. Specific details of each criterion response set are not elaborated here. Criterion responses can be responses regarding the validity and usability of candidate criterion information. Illusion tags can be tags representing the possibility of output illusion content generated by the third business processing agent. The purpose of illusion tags is to subsequently remind the second business processing agent to consider the illusion questions corresponding to the third business processing agent when verifying each question-response set and each criterion-response set.

[0129] Sub-step 3 involves sending the obtained question-response sets and criterion-response sets to the second business processing agent for verification and aggregation of the question and criterion responses, resulting in an aggregated response set, which serves as the business enhancement dataset. Here, the aggregated response set can be the data-enhanced response set based on the various agents of the external business system.

[0130] Sub-step 4 involves acquiring the data permissions between the respective third service processing agents. These data permissions can be the permissions granted to the third service processing agents for data transmission. Each third service processing agent has corresponding set data permissions.

[0131] Here, data permissions are defined to determine which data is supported for transmission by the external business system. The data transmitted here is not the agent's predicted output, but rather actual business data occurring within the external business system.

[0132] Sub-step 5: Based on the respective data permissions, obtain the respective business datasets stored by each third business processing agent. Each third business processing agent has a corresponding business dataset. The business datasets can be various business data related to the process problem set and the candidate criterion information set. The business data are real data from external business systems.

[0133] Sub-step 6: Summarize the data from each business dataset to obtain the initial business dataset.

[0134] Sub-step 7: The initial business dataset is sent to the second business processing agent for data verification and data filtering based on the corresponding business logic of the target business system to obtain the actual business dataset.

[0135] Here, as another inventive point of this disclosure, by setting up a first business processing intelligent agent and a second business processing intelligent agent, based on the diverse datasets of real-time business datasets and business enhancement datasets, it can ensure the timely discovery and accuracy of updated process content and newly added rule content. Based on this, it can achieve the first-time discovery and processing of updated content. Furthermore, when the user or the first business processing intelligent agent generates a process problem set and / or rule message set, by interacting with various third business processing intelligent agents deployed in various external business systems, a larger amount of business enhancement datasets generated by each intelligent agent can be obtained. That is, when facing process problems and / or rule problems, by comprehensively utilizing the predictive capabilities of various intelligent agents corresponding to various external business systems, a more comprehensive and diverse business enhancement dataset can be generated. In addition, data verification by the second business processing intelligent agent ensures that inaccurate data caused by model illusion problems is removed during the generation of business enhancement datasets, making the business enhancement datasets more accurate. Secondly, based on various data permissions, it can obtain real data related to the problems corresponding to various external business systems, so as to improve the accuracy of subsequent verification of newly added rule information. That is, by using augmented datasets and real datasets generated based on diverse intelligent agents, a more comprehensive and high-quality dataset can be obtained.

[0136] The above embodiments of this disclosure have the following beneficial effects: Through the process page display method of some embodiments of this disclosure, after receiving business data related to business process generation uploaded to the process page, the corresponding business process information can be accurately rendered directly on the process page, and various page filtering operations supporting business process information can be set, greatly improving the rendering efficiency of the business process. Specifically, the reason for the low rendering efficiency of the relevant business processes is that it requires professionals and developers who are very familiar with the business processes, resulting in high requirements for business process generation and rendering. In addition, the business process generation stage and the business process rendering stage are handled by different objects, and there is an information gap between the two stages. Furthermore, when the target business system has a large amount of complex business content, the workload of business process generation is large and the rendering cycle of the page is long, resulting in low rendering efficiency of the corresponding business process of the target business system. Based on this, the process page display method of some embodiments of this disclosure first, in response to receiving a process page rendering request for the target business system, obtains the code information and business processing rule information set for the target business system. After selecting a request for one-click business process generation and rendering for the target business system, code information is obtained to facilitate the analysis of various business logics within the target business system. This allows the primary language model to learn these logics and accurately generate the corresponding business processes. By acquiring a set of business processing guidelines, the initial business process information generated can avoid inter-batch business interaction issues, improving the accuracy of subsequent full business process information generation. Then, based on the code information, the primary language model deployed on the primary server accurately and efficiently generates initial full business process information for each business batch, where the thought chain length corresponding to the primary language model meets the target length condition. Here, by setting the thought chain length of the primary language model to meet the target length condition, the primary language model can perform more thought steps during the process of learning code information to generate initial full business process information, resulting in more accurate business process information. Next, based on the set of business processing guidelines, the execution relationships between business batches of the initial full business process information are verified to obtain full business process information, resulting in more accurate process information with precise business logic. Furthermore, utilizing the first major language model, page rendering information for the entire business process can be accurately and automatically generated, and based on this page rendering information, the entire business process information can be effectively rendered and displayed on the process page. Here, using the first major language model to generate page rendering information improves the accuracy of the generated information. Specifically, because the first major language model can learn more deeply the business content related to the target business system, the generated page rendering information is also more accurate.Furthermore, in response to the selection information of at least one scenario field on the process page, based on the at least one scenario field and the full business process information, and utilizing the first large language model, the first large language model's proficiency in business content can be leveraged to accurately and efficiently infer and generate a business batch matching the at least one scenario field, which serves as the first business batch. Finally, the stored business process processing tree corresponding to the first business batch is displayed on the process page. In summary, through the first large language model deployed on the first server and the setting of the thought chain length, a series of processes can be implemented for full business process information, including process information mastery and generation, page rendering, and process search. Based on this, accurate and intelligent generation of corresponding business process information can be achieved. In addition, by performing one-click process generation for the target business system on the process page, accurate display of business process information on the process page can be achieved without wasting a lot of human resources. Furthermore, the ability to set up various page filtering operations for business process information greatly improves the rendering efficiency of the business process.

[0137] Further reference Figure 5 As an implementation of the methods shown in the above figures, this disclosure provides some embodiments of a process page display device, which are similar to... Figure 1 Corresponding to the method embodiments shown, this process page display device can be specifically applied to various electronic devices.

[0138] like Figure 5As shown, a process page display device 500 includes: an acquisition unit 501, a first generation unit 502, an adjustment unit 503, a first display unit 504, a second generation unit 505, and a second display unit 506. The acquisition unit 501 is configured to, in response to receiving a process page rendering request for a target business system, acquire code information and a set of business processing guidelines for the target business system; the first generation unit 502 is configured to, based on the code information and using a first large language model deployed on a first server, generate initial full business process information for each business batch, wherein the thought chain length corresponding to the first large language model satisfies a target length condition; the adjustment unit 503 is configured to, based on the set of business processing guidelines, verify the execution relationship between business batches in the initial full business process information to obtain full business process information; the first display unit 504, the first generation unit 505, and the second display unit 506. Unit 504 is configured to use the first large language model to generate page rendering information for the full business process information, and to render and display the full business process information on the process page based on the page rendering information; Unit 505 is configured to respond to the selection information of at least one scene field selected on the process page, and to generate a business batch matching the at least one scene field as a first business batch based on the at least one scene field and the full business process information using the first large language model; Unit 506 is configured to display the stored business process processing tree corresponding to the first business batch on the process page.

[0139] It is understandable that the various units described in the process page display device 500 and the references Figure 1 The steps described in the method correspond to each other. Therefore, the operations, features, and beneficial effects described above for the method also apply to the process page display device 500 and the units contained therein, and will not be repeated here.

[0140] The following is for reference. Figure 6 It shows a schematic diagram of the structure of an electronic device (e.g., an electronic device) 600 suitable for implementing some embodiments of the present disclosure. Figure 6 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of the embodiments of this disclosure.

[0141] like Figure 6As shown, electronic device 600 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 601, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 602 or a program loaded from storage device 608 into random access memory (RAM) 603. RAM 603 also stores various programs and data required for the operation of electronic device 600. Processing device 601, ROM 602, and RAM 603 are interconnected via bus 604. Input / output (I / O) interface 605 is also connected to bus 604.

[0142] Typically, the following devices can be connected to I / O interface 605: input devices 606 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 607 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 608 including, for example, magnetic tapes, hard disks, etc.; and communication devices 609. Communication device 609 allows electronic device 600 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 6 An electronic device 600 with various devices is shown; however, it should be understood that it is not required to implement or possess all of the devices shown. More or fewer devices may be implemented or possessed alternatively. Figure 6 Each box shown can represent a device or multiple devices as needed.

[0143] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device 609, or installed from a storage device 608, or installed from a ROM 602. When the computer program is executed by the processing device 601, it performs the functions defined above in the methods of some embodiments of this disclosure.

[0144] It should be noted that, in some embodiments of this disclosure, the computer-readable medium described above may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In some embodiments of this disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer-readable medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0145] In some implementations, clients and servers can communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol) and can interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future-developed networks.

[0146] The aforementioned computer-readable medium may be included in the aforementioned electronic device; or it may exist independently and not assembled into the electronic device. The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to: In response to receiving a process page rendering request for a target business system, acquire code information and a set of business processing guidelines for the target business system; based on the code information, use a first large language model deployed on a first server to generate initial full business process information for each business batch, wherein the thought chain length corresponding to the first large language model satisfies a target length condition; verify the execution relationship between business batches of the initial full business process information based on the set of business processing guidelines to obtain full business process information; use the first large language model to generate page rendering information for the full business process information, and render and display the full business process information on the process page based on the page rendering information; in response to selecting selection information for at least one scene field on the process page, generate a business batch matching the at least one scene field as a first business batch based on the at least one scene field and the full business process information using the first large language model; and display a stored business process processing tree corresponding to the first business batch on the process page.

[0147] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0148] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0149] The units described in some embodiments of this disclosure can be implemented in software or hardware. The described units can also be housed in a processor; for example, a processor may be described as including an acquisition unit, a first generation unit, an adjustment unit, a first display unit, a second generation unit, and a second display unit. The names of these units do not necessarily limit the specific unit itself; for example, the acquisition unit may also be described as "a unit that, in response to receiving a process page rendering request for a target business system, acquires code information and a set of business processing guidelines information for the target business system."

[0150] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), system-on-a-chip (SoCs), complex programmable logic devices (CPLDs), and so on.

[0151] The above description is merely a selection of preferred embodiments of this disclosure and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A method for displaying a process page, characterized in that, include: In response to receiving a process page rendering request for the target business system, obtain the code information and business processing guidelines information set for the target business system; Based on the code information, the first large language model deployed on the first server is used to generate initial full business process information for each business batch, wherein the thought chain length corresponding to the first large language model satisfies the target length condition. Based on the business processing rule information set, the execution relationship between business batches of the initial full business process information is verified to obtain the full business process information; Using the first large language model, generate page rendering information for the full business process information, and render and display the full business process information on the process page according to the page rendering information; In response to the selection information of at least one scenario field selected on the process page, based on the at least one scenario field and the full business process information, a business batch matching the at least one scenario field is generated using the first large language model, which is then used as the first business batch. The process page displays the stored business process processing tree corresponding to the first business batch.

2. The method according to claim 1, characterized in that, The method further includes: In response to selecting a question about a specific business process within the overall business process information on the process page, the system uses the first large language model to generate a response to the question. In response to the input of business search information in the corresponding search bar on the process page, the system generates the corresponding response content based on the first large language model. In response to selecting query information for at least one business processing rule information corresponding to a local business process content on the process page, the first large language model is used to generate descriptive information for the at least one business processing rule information. In response to the selection information for the second business batch being executed on the process page, the business process processing content corresponding to the second business batch is determined from the full business process information; The process information corresponding to the business process content is highlighted in the full business process information displayed on the process page.

3. The method according to claim 1, characterized in that, The business processing criteria information in the business processing criteria information set is generated through the following steps: In response to the existence of a rule code block corresponding to the business processing rule information, extract the business description information and key rule content corresponding to the business processing rule information; Using the second largest language model deployed on the second server, perform the following generation steps: Search for the business processing content corresponding to the business description information and the rule-related content corresponding to the key rule content. The rule-related content includes: external rule code blocks, the second language model is obtained by transfer learning based on the first language model, and the thought chain length corresponding to the first language model is higher than the thought chain length corresponding to the second language model. The code logic of the rule code block and the external rule code block is parsed to obtain the initial business processing rule information; The business processing content and the rule-related content are sent to the first server to generate a rule-related question set and a response result set based on the deployed first large language model. Based on the business processing content, the related content of the criteria, the set of related questions of the criteria, and the set of response results, the initial business processing criterion information is validated and its format is converted to obtain business processing criterion information.

4. The method according to claim 1, characterized in that, The step of verifying the execution relationship between business batches of the initial full business process information based on the business processing rule information set to obtain the full business process information includes: From the business processing rule information set, a subset of business processing rule information corresponding to each business batch is selected; Obtain key information about the criteria for the subset of business processing criteria information set on the process page rendering request page; Based on the important information of the criteria and the criterion complexity corresponding to the subset of business processing criteria information, determine whether to call the second largest language model in the second server; In response to the confirmed call, the second large language model is used to perform rule information summary output and rule information sorting on the subset of business processing rule information to obtain a rule information group sequence and a rule summary group sequence, wherein the relationship of each rule information in the rule information group is adjusted synchronously; Based on the criterion summary group sequence, generate the sequence order accuracy value corresponding to the criterion information group sequence; In response to the sequence order accuracy value being higher than the target accuracy value, the execution relationship between business batches of the initial full business process information is adjusted according to the criterion information group sequence to obtain the full business process information.

5. The method according to claim 4, characterized in that, The step of adjusting the execution relationship between business batches of the initial full business process information according to the criterion information group sequence to obtain the full business process information includes: For the target criterion information group in the criterion information group sequence, perform the following adjustment steps: The code blocks that are related to the target criterion information group are selected from the code information and used as target code blocks; Obtain the full business process processing information corresponding to the previous criterion information group; For each target criterion information, based on the target code block and the target criterion information, the second major language model is used to perform business execution content verification on multiple business batches related to the target criterion information in the full business process processing information, and a verification result is obtained. The verification result is one of the following: a verification result that passes the verification, a verification result that the code lacks logical support, or a verification result that has redundant constraint logic. Based on the obtained verification result group, generate the target full business process processing information corresponding to the target criterion information group; In response to the target criterion information group being the last criterion information group in the sequence, the target full business process processing information is determined as full business process information; In response to the fact that the target criterion information group is not the last criterion information group in the sequence, the next criterion information group corresponding to the target criterion information group is determined as the target criterion information group, and the adjustment steps are continued.

6. The method according to claim 1, characterized in that, Based on the code information, the first language model deployed on the first server is used to generate initial full business process information for each business batch, including: Based on the code information, extract the business batch set and the field group corresponding to each business batch; In response to determining that the service batch set includes each service batch, the code block corresponding to each service batch is extracted from the code information and used as the service code block; In response to determining that the service batch set does not include the respective service batches, at least one service batch that does not exist in the service batch set is obtained from the respective service batches; For each of the at least one business batch, based on the field group corresponding to the business batch, filter out the associated business batch group that has a business relationship with the business batch from the business batch set; Based on the local code block corresponding to the obtained associated business batch set and at least one field group corresponding to the at least one business batch, the first large language model is used to generate a local code block corresponding to the at least one business batch as a supplementary code block. Based on the supplementary code block and the code information, generate business code blocks corresponding to each business batch; Based on the business code block, the initial full business process information is generated using the first large language model.

7. The method according to claim 1, characterized in that, in Before generating initial full business process information for each business batch based on the code information and using the first large language model deployed on the first server, the method further includes: Obtain the code volume and process generation task information corresponding to the code information; Based on the amount of code and the task information generated by the process, the processing difficulty of the task corresponding to the task information generated by the process is determined using a pre-trained classification algorithm for difficulty classification. In response to the processing difficulty being higher than a first difficulty value and the current model call resources monitored by the model resource monitor meeting the resource conditions, it is determined that the first large language model deployed on the first server will be used to complete the process generation task; and Before verifying the execution relationship between business batches of the initial full business process information based on the business processing rule information set to obtain the full business process information, the method further includes: Determine the semantic complexity and number of criteria corresponding to the business processing criterion information set; The adjustment difficulty corresponding to the execution relationship adjustment task is determined based on the semantic complexity and the number of criteria. In response to the fact that the adjustment difficulty is lower than the first difficulty value but higher than the second difficulty value, it is determined that the second largest language model deployed on the second server is used to complete the relation adjustment task.

8. A process page display device, characterized in that, include: The acquisition unit is configured to, in response to receiving a process page rendering request for the target business system, acquire code information and a set of business processing guidelines information for the target business system. The first generation unit is configured to generate initial full business process information for each business batch based on the code information and using the first large language model deployed on the first server, wherein the thought chain length corresponding to the first large language model satisfies the target length condition. The adjustment unit is configured to verify the execution relationship between business batches of the initial full business process information based on the business processing criterion information set, so as to obtain the full business process information. The first display unit is configured to use the first large language model to generate page rendering information for the full business process information, and to render and display the full business process information on the process page according to the page rendering information. The second generation unit is configured to, in response to the selection information of at least one scenario field selected on the process page, generate a business batch matching the at least one scenario field as the first business batch based on the at least one scenario field and the full business process information using the first large language model. The second display unit is configured to display the stored business process processing tree corresponding to the first business batch on the process page.

9. An electronic device, characterized in that, include: One or more processors; Storage device, on which one or more programs are stored, When the one or more programs are executed by the one or more processors, the one or more processors implement the method as described in any one of claims 1-7.

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

Images