Vault mode application reason compliance identification method, apparatus and device, and medium

Abstract

The invention discloses a vault mode application reason compliance identification method and device, equipment and a medium. The method comprises the following steps: in response to an initial vault authorization request of a target object, performing semantic recognition on an initial vault application reason in the initial vault authorization request based on a vault word segmentation model and a vault scene standard word bank to obtain an application reason keyword; performing scene matching on the application reason keyword according to the application reason standard segmented word under each scene in the vault scene standard word bank to obtain a target scene corresponding to the application reason keyword; determining a plurality of application reason candidate templates corresponding to the target scene according to a vault scene reason recommendation library, and feeding back the application reason candidate templates to the target object; and in response to an update vault authorization request of the target object based on the application reason target template, reason template parameters are extracted from the update vault application reason of the update vault authorization request to perform compliance verification. According to the scheme, the manual approval efficiency can be improved, and potential safety hazards caused by excessive authorization are reduced.

Description

Technical Field

[0001] This invention relates to the field of big data technology, and in particular to a method, apparatus, equipment and medium for verifying the compliance of application reasons for a vault model. Background Technology

[0002] The "Vault Mode" employs a dual-person control model for high-risk and sensitive operations: one person operates the operation, and another approves it. This mode enhances the security of such operations.

[0003] In existing technologies, the vault model application adopts a basic text-based open submission method, where the reason for the vault authorization application is solely written by the applicant. This method has the following problems in the actual vault authorization process: 1. It only judges whether the application reason is filled in, without basic identification of invalid application reasons such as "test" or "adc", increasing the amount of invalid business and reducing the work efficiency of approvers; 2. There are irregular approval situations in the vault approval scenario: for example, when recharging a user, if the vault application requires checking the user's call details, there is a mismatch between the application reason and the business scenario, which poses a risk of over-authorization. Summary of the Invention

[0004] This invention provides a method, apparatus, equipment, and medium for verifying the compliance of vault application reasons. It performs intelligent semantic recognition and scenario-related template recommendation on vault application reasons, which can ensure the validity of the submitted application reasons, thereby improving approval efficiency. At the same time, it reduces the potential over-authorization caused by unreasonable business acceptance and lowers the security risks of excessive authorization.

[0005] According to one aspect of the present invention, a method for verifying the compliance of a vault-style application is provided, the method comprising:

[0006] In response to the initial vault authorization request from the target object, semantic recognition is performed on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario;

[0007] Based on the standard word segmentation of the application reason in each scenario in the standard vocabulary library of the vault scenario, the application reason keywords are matched to obtain the target scenario corresponding to the application reason keywords;

[0008] Based on the vault scenario reason recommendation library, multiple application reason candidate templates corresponding to the target scenario are determined, and the application reason candidate templates are fed back to the target object; wherein, the vault scenario reason recommendation library is used to describe the association between application reason templates and scenarios;

[0009] In response to the target object's update vault authorization request based on the application reason target template, the reason template parameter is extracted from the update vault application reason of the update vault authorization request as a parameter to be verified, and the parameter to be verified is subjected to compliance verification to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object.

[0010] According to another aspect of the present invention, a device for verifying the compliance of a vault-style application is provided, the device comprising:

[0011] The keyword extraction module is used to respond to the initial vault authorization request of the target object, and to perform semantic recognition on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario;

[0012] The scenario matching module is used to perform scenario matching on the application reason keywords according to the standard word segmentation of the application reason under each scenario in the standard word library of the vault scenario to obtain the target scenario corresponding to the application reason keywords;

[0013] The template recommendation module is used to determine multiple candidate templates for application reasons corresponding to the target scenario based on the vault scenario reason recommendation library, and to feed back the candidate templates for application reasons to the target object; wherein, the vault scenario reason recommendation library is used to describe the relationship between application reason templates and scenarios;

[0014] The parameter verification module is used to respond to the target object's update vault authorization request based on the application reason target template, extract the reason template parameter from the update vault application reason of the update vault authorization request as the parameter to be verified, and perform compliance verification on the parameter to be verified to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object.

[0015] According to another aspect of the present invention, an electronic device is provided, the electronic device comprising:

[0016] At least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores a computer program executable by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the vault pattern application justification compliance identification method according to any embodiment of the present invention.

[0017] According to another aspect of the present invention, a computer-readable storage medium is provided, the computer-readable storage medium storing computer instructions for causing a processor to execute and implement the vault pattern application reason compliance identification method according to any embodiment of the present invention.

[0018] The technical solution of this invention, in response to an initial vault authorization request from a target object, performs semantic recognition on the initial vault application reason in the initial vault authorization request based on a vault word segmentation model and a vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario; according to the standard word segmentation of the application reason under each scenario in the vault scenario standard lexicon, the application reason keywords are matched with the scenario to obtain the target scenario corresponding to the application reason keywords; according to the vault scenario reason recommendation library, multiple application reason candidate templates corresponding to the target scenario are determined, and the application reason candidate templates are fed back to the target object; wherein, the vault scenario reason recommendation library is used to describe the association relationship between the application reason template and the scenario; in response to an updated vault authorization request from the target object based on the application reason target template, the reason template parameters are extracted from the updated vault application reason of the updated vault authorization request as parameters to be verified, and compliance verification is performed on the parameters to be verified to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object. This technical solution, through intelligent semantic recognition and scenario-related template recommendation of the reasons for treasury applications, can ensure the validity of the submitted reasons, reduce the tedious manual identification workload caused by invalid reasons, and thus improve approval efficiency. At the same time, it reduces the potential over-authorization caused by unreasonable business acceptance and lowers the security risks of excessive authorization.

[0019] It should be understood that the description in this section is not intended to identify key or essential features of the embodiments of the present invention, nor is it intended to limit the scope of the invention. Other features of the invention will become readily apparent from the following description. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 This is a flowchart of a method for verifying the compliance of a vault model application reason according to an embodiment of the present invention;

[0022] Figure 2This is a flowchart of another method for verifying the compliance of a vault model application reason provided by an embodiment of the present invention;

[0023] Figure 3 This is a schematic diagram of a device for verifying the compliance of a vault application reason according to an embodiment of the present invention;

[0024] Figure 4 This is a schematic diagram of the structure of an electronic device that implements a method for verifying the compliance of a vault mode application reason according to an embodiment of the present invention. Detailed Implementation

[0025] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0026] It should be noted that the terms "first," "second," "target," etc., used in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0027] Example 1

[0028] Figure 1 This is a flowchart illustrating a method for verifying the compliance of a vault-pattern application reason according to Embodiment 1 of the present invention. This embodiment is applicable to situations requiring efficient verification of the compliance of vault-pattern application reasons. This method can be executed by a vault-pattern application reason compliance verification device, which can be implemented in hardware and / or software and can be configured in an electronic device with data processing capabilities. Figure 1 As shown, the method includes:

[0029] S110, in response to the initial vault authorization request of the target object, semantic recognition is performed on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain the application reason keywords.

[0030] The standard lexicon for vault scenarios describes the relationship between standard word segments for application reasons and scenarios, specifically, one scenario is associated with multiple standard word segments for application reasons. A vault authorization request can refer to an operational instruction requesting authorization in vault mode. The target object can refer to the object that makes the vault authorization request. The vault word segmentation model can refer to an artificial intelligence model that can extract keywords for application reasons based on the standard lexicon for vault scenarios and through semantic recognition of the application reasons.

[0031] In this embodiment, a vault word segmentation model needs to be established in advance. Optionally, the training process of the vault word segmentation model includes: obtaining historical vault application reasons and their corresponding keyword annotations as a first training set, and obtaining standard vault application reasons and their corresponding keyword annotations as a second training set; wherein, the standard vault application reasons are generated based on synthetic data technology; using low-rank matrix technology and DeepSpeed ​​technology, the artificial intelligence model is supervised and trained according to the first training set and the second training set to obtain the vault word segmentation model.

[0032] Specifically, the process begins by using a secure big data platform as the data source. Multiple historical vault application reasons are retrieved from this platform, and keyword annotations for each reason are determined manually. These historical reasons and their corresponding keyword annotations form the first training set. Historical vault application reasons can refer to existing, actual vault application reasons, including both compliant and non-compliant ones. Keyword annotations can involve extracting keywords from the application reasons and labeling their components (e.g., subject, predicate, object, adverbial) and parts of speech (e.g., noun, verb, adjective). Simultaneously, compliant vault application reasons are generated using synthetic data technology as standard vault application reasons. Keyword annotations for each standard reason are determined manually, and these standard reasons and their corresponding keyword annotations form the second training set. The dataset for model training can then be constructed based on the first and second training sets. Synthetic data technology effectively addresses issues related to data supply and secure data utilization, while also improving model performance and domain adaptability.

[0033] Next, using Low-Rank Adaptation (LoRA) and DeepSpeed ​​techniques, the AI ​​model is trained under supervision using the first and second training sets to obtain the Jinku word segmentation model. LoRA, by freezing the original model parameters, adding a bypass to reduce the dimensionality of the input data, and then increasing the dimensionality of the reduced representation, effectively captures and represents the main information of the original high-dimensional parameter matrix, achieving efficient adaptation and fine-tuning of the AI ​​model. This not only reduces computational and storage overhead but also maintains high model performance. For example, the AI ​​model can be a Large Language Model (LLM). DeepSpeed ​​optimization techniques improve the efficiency and scalability of model training, including the following: 1. Memory optimization: ZeRO (Zero Redundancy Optimizer) technology is divided into multiple levels, from ZeRO-1 to ZeRO-3, gradually decomposing model parameters, optimization states, and gradients onto multiple GPUs, reducing the memory footprint of each GPU. It also supports efficient distributed data parallelism (DDP) strategies, optimizing communication efficiency and error handling between multiple GPUs through parameter optimization. 2. Gradient: Uses the gradient descent algorithm to find a set of model parameters that minimizes the target loss function. 3. Training Management and Resource Allocation: Supports learning rate scheduling, warm-up ratio settings, logging strategies, and automatic saving of model checkpoints for better monitoring and control of the training process.

[0034] It should be noted that this embodiment does not impose specific limitations on the model training method, and can be flexibly set according to actual needs. For example, one artificial intelligence model can be selected as the base model, and supervised training of the base model can be performed using both a first training set and a second training set. The trained base model can then be used as the Jinku word segmentation model. Alternatively, two artificial intelligence models can be selected as base models. The first base model can be supervised training using the first training set, and the second base model can be supervised training using the second training set. The training results of the second base model can then be integrated into the first base model for further training. Finally, the trained first base model can be used as the Jinku word segmentation model. It should also be noted that the Jinku application reason is a short text (generally within 200 characters) and has characteristics such as feature sparsity, singularity, dynamism, and interleaving. To overcome the limitations of traditional methods that rely solely on a single segmentation approach and achieve more accurate and faster segmentation, this invention constructs a first basic model based on a Trie tree structure. This model efficiently scans the entire sentence containing the reasons for the vault application, thereby improving processing speed. Simultaneously, a second basic model is constructed based on a Bidirectional Long Short-Term Memory Network (BiLSTM) to enhance segmentation accuracy. Specifically, the Trie tree structure is generated using a standard vault scenario lexicon and segmentation methods based on statistics and string matching. Specifically, it searches the standard vault scenario lexicon one by one using forward maximum matching, backward maximum matching, bidirectional maximum matching, and minimum segmentation (i.e., shortest path) scanning methods to generate a valid Trie tree.

[0035] In this embodiment, after establishing the vault word segmentation model, semantic recognition of the initial vault application reason in the initial vault authorization request can be performed based on the vault word segmentation model and the vault scenario standard lexicon to obtain application reason keywords. Optionally, performing semantic recognition of the initial vault application reason in the initial vault authorization request to obtain application reason keywords based on the vault word segmentation model and the vault scenario standard lexicon includes: generating a directed acyclic graph composed of all word formation cases of the initial vault application reason based on the vault word segmentation model as candidate word segmentation paths for the initial vault application reason; using a dynamic programming method to find the most probable path among the candidate word segmentation paths as the target word segmentation path; and determining the application reason keywords based on the target word segmentation path.

[0036] Specifically, firstly, a vault word segmentation model based on a Trie tree structure is used to efficiently scan the initial vault application reason, generating a directed acyclic graph (DAG) composed of all possible word combinations of Chinese characters in the sentence. For example, for "new salesperson", an acyclic graph is generated by referring to the existing Trie tree, and the corresponding candidate word segmentation paths are: the segmentation scheme of path 1 is "new|salesperson", and the segmentation scheme of path 2 is "new|sales|person". Then, a dynamic programming method (such as the Viterbi algorithm) is used to find the words that have been segmented in the candidate word segmentation paths and determine the frequency (number of times / total number) of the word. If the word does not exist, the frequency of the word with the lowest frequency in the standard vocabulary of the vault scenario is taken as the frequency of the word, that is, P(new) = FREQ.get('new', min_freq), and the Chinese words are arranged into a sequence according to B (begin, i.e., the starting position) - E (end, i.e., the ending position) - M (middle, i.e., the middle position) - S (single, i.e., the position of a single word, without before and after). For example, the word "new" can be tagged as BE, which is new / B-increase / E, meaning "new" is the starting position and "increase" is the ending position; "new salesperson" would be tagged as BMME, which is start-middle-middle-end. The method of finding the maximum probability path using dynamic programming involves calculating the maximum probability of the sentence from right to left, i.e., P(NodeN)=1.0, P(NodeN-1)=P(NodeN)×Max(P(last word))... and so on. The maximum probability path can be obtained as the target word segmentation path, and the segmentation combination corresponding to the target word segmentation path can be determined as the application reason keywords. It should be noted that in the vault word segmentation model, three probability tables can be obtained through training (including the initial state probability table, the state transition probability table, and the observation probability table). Based on these three probability tables, the Viterbi algorithm can obtain a BEMS sequence with the highest probability. By recombining the initial vault application reasons according to the order of starting with B and ending with E, the word segmentation result (i.e., the application reason keywords) is obtained.

[0037] It should be noted that in reality, there may be intentionally filled-in invalid vault application reasons such as "fever so high that it leads to sag 15". Such application reasons cannot be extracted using the standard vault scenario terminology. In this case, the target verification result is directly determined as verification failure, and the vault mode authorization is deemed unsuccessful.

[0038] S120: Based on the standard word segmentation of the application reason in each scenario in the standard word library of the vault scenario, the application reason keywords are matched with the target scenario corresponding to the application reason keywords.

[0039] In this embodiment, after obtaining the keywords for the initial vault application reason, the keywords can be matched against the standard word segmentation of the application reason in each scenario within the vault scenario standard thesaurus to obtain the target scenario corresponding to the keywords. Optionally, matching the keywords against the application reason in each scenario within the vault scenario standard thesaurus to obtain the target scenario includes: determining the similarity between the standard word segmentation of the application reason in each scenario within the vault scenario standard thesaurus and the keywords to obtain multiple candidate similarities; if the candidate similarity is greater than or equal to a preset similarity threshold, the candidate similarity is determined as a reference similarity; a maximum value is determined from the reference similarities as the target similarity, and the scenario corresponding to the target similarity is determined as the target scenario.

[0040] Specifically, firstly, the similarity between the keywords of the application reason and the standard word segments of the application reason in each scenario of the standard vocabulary library for the vault scenario is calculated to obtain the candidate similarity for each scenario. Then, each candidate similarity is compared with a preset similarity threshold. If the candidate similarity is greater than or equal to the preset similarity threshold, it indicates that the scenario relevance is strong (at this time, the scenario is considered to be successfully matched), and the candidate similarity is determined as the reference similarity. Then, the maximum value among the multiple reference similarities is determined as the target similarity. In special cases, if there is only one reference similarity, it is directly used as the target similarity, and the scenario corresponding to the target similarity is determined as the target scenario.

[0041] S130: Based on the vault scenario reason recommendation library, determine multiple application reason candidate templates corresponding to the target scenario, and feed back the application reason candidate templates to the target object.

[0042] The vault scenario reason recommendation library describes the relationship between application reason templates and scenarios. Specifically, a scenario is associated with multiple application reason templates, and these templates are ordered according to their weight (based on the frequency of most recent template use) (higher weights rank higher). The vault scenario reason recommendation library is constructed as follows: Based on the standard vault scenario thesaurus, existing vault application reasons are further categorized using an unsupervised approach. Expert annotation is then introduced, and a parameter model (i.e., application reason template) is designed to be validated at the granular level of the application reason. This parameter model conforms to the parameters used to call the reasonableness validation interface for the scenario. For example, in the scenario "4A front-end batch master-slave account binding permission change operation," applicants need to batch bind master-slave accounts for adding employees from A to Z. Accounts A, Z, and Z are newly added employee accounts. By adding employees using keywords, employee accounts from A to Z are separated, and the new employee business scenario is identified. The vault scenario reason recommendation library can be continuously updated according to business rules.

[0043] In this embodiment, after determining the target scenario corresponding to the keywords of the application reason, scenarios in the vault scenario reason recommendation library can be searched according to the target scenario. Based on the association between the application reason template and the scenario, a preset number of templates with higher weights among the multiple application reason templates corresponding to the target scenario are selected as candidate application reason templates. These candidate templates can be fed back to the target object in a list format to fill in the template for a standardized vault application reason matching the recommended scenario for the target object. For example, for a vault application reason such as "User applies for transfer of points at that time," five application reason templates with higher weights in the corresponding scenario can be provided as candidate application reason templates, such as "User XX applied for transfer at XX time, xx business is similar, therefore applying for transfer of remaining points!"

[0044] Furthermore, after receiving multiple candidate templates for application reasons, the target can click on the list to select the most suitable one as the target template for the application reason according to the actual needs. Based on the target template, the target can fill in key parameters to form an updated vault application reason. This achieves a standardized update of the initial vault application reason based on the target template, and then re-initiates the vault authorization request (i.e., update the vault authorization request) based on the updated vault application reason.

[0045] S140, in response to the target object's update vault authorization request based on the target template of the application reason, extract the reason template parameter from the update vault application reason of the update vault authorization request as the parameter to be verified, and perform compliance verification on the parameter to be verified to obtain the target verification result.

[0046] In this embodiment, after receiving the vault update authorization request initiated by the target object based on the application reason target template, it is necessary to extract the key parameters (i.e., reason template parameters) filled in from the vault update authorization request as parameters to be verified, and then perform compliance verification on the parameters to be verified to obtain the target verification result. Here, the application reason target template refers to a candidate application reason template currently used by the target object. Optionally, performing compliance verification on the parameters to be verified to obtain the target verification result includes: calling the parameter verification interface corresponding to the target scenario to perform compliance verification on the parameters to be verified to obtain the target verification result; wherein, the target verification result is either verification passed or verification failed; if the target verification result is verification passed, the weights of the multiple application reason candidate templates corresponding to the target scenario are adjusted.

[0047] Specifically, all parameters to be verified can be combined into an XML file. By calling the parameter verification interface corresponding to the target scenario, and combining it with the treasury audit recommendation library, the parameters to be verified are automatically checked for compliance. This yields the target verification result (verification passed or failed). The treasury audit recommendation library stores parameter verification rules for different scenarios. For example, for the scenarios of adding and modifying three accounts, the corresponding parameter verification interfaces may include account closure audit interfaces, points audit interfaces, balance audit interfaces, and transfer audit interfaces. If the target verification result is a pass, the treasury mode authorization is deemed successful, and the weights of multiple application reason candidate templates corresponding to the target scenario are adjusted to rank them based on the adjusted weights. If the target verification result is a fail, the treasury mode authorization is deemed unsuccessful. Furthermore, the analysis results from the previous steps can be output in conjunction with the 4A system and comprehensively displayed in a visual chart format, achieving the effects of daily learning, real-time analysis, and real-time processing.

[0048] The technical solution of this invention, in response to an initial vault authorization request from a target object, performs semantic recognition on the initial vault application reason in the initial vault authorization request based on a vault word segmentation model and a vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario; according to the standard word segmentation of the application reason under each scenario in the vault scenario standard lexicon, the application reason keywords are matched with the scenario to obtain the target scenario corresponding to the application reason keywords; according to the vault scenario reason recommendation library, multiple application reason candidate templates corresponding to the target scenario are determined, and the application reason candidate templates are fed back to the target object; wherein, the vault scenario reason recommendation library is used to describe the association relationship between the application reason template and the scenario; in response to an updated vault authorization request from the target object based on the application reason target template, the reason template parameters are extracted from the updated vault application reason of the updated vault authorization request as parameters to be verified, and compliance verification is performed on the parameters to be verified to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object. This technical solution, through intelligent semantic recognition and scenario-related template recommendation of the reasons for treasury applications, can ensure the validity of the submitted reasons, reduce the tedious manual identification workload caused by invalid reasons, and thus improve approval efficiency. At the same time, it reduces the potential over-authorization caused by unreasonable business acceptance and lowers the security risks of excessive authorization.

[0049] In this embodiment, optionally, the method further includes: if the similarity of multiple candidates is less than a preset similarity threshold, then the target verification result of the initial vault authorization request is determined to be verification failure; determine whether the keywords of the application reason match the candidate scenario, where the candidate scenario refers to other scenarios outside the vault scenario standard thesaurus; if so, generate an audit record based on the initial vault application reason and its corresponding candidate scenario, and determine whether to authorize the vault mode based on the manual review result of the audit record.

[0050] In this embodiment, if the similarity of multiple candidates is less than a preset similarity threshold, it indicates that the scene relevance is weak (at this time, the scene matching is considered unsuccessful). Therefore, the target verification result of the initial vault authorization request is directly determined to be verification failure, and the vault mode authorization is deemed unsuccessful, which helps improve the security of vault mode authorization. For example, a vault application submitted through a recharge scenario with the reason "User XX recharged at XX time and processed XX service, and needs to apply for call detail record inquiry!" is clearly described, but the recharge service and the call detail record inquiry service are obviously mismatched. In this case, vault mode authorization is directly rejected.

[0051] Furthermore, it can be determined whether the keywords of the application reason match the candidate scenarios (i.e., scenarios other than those in the standard vault scenario thesaurus). If so, it indicates that there may be risky trial-and-error behavior. In this case, it is necessary to generate an audit record based on the initial vault application reason and its corresponding candidate scenarios, so that the results of manual review of the audit record can be used to determine whether there is indeed risky trial-and-error behavior.

[0052] Example 2

[0053] Figure 2 This is a flowchart of a method for identifying the compliance of vault application reasons according to Embodiment 2 of the present invention. This embodiment is an optimization based on the above embodiment. Specifically, the optimization includes: before determining multiple candidate templates for application reasons corresponding to the target scenario according to the vault scenario reason recommendation library, it further includes: determining whether the target object has a need for application reason template recommendations; if so, determining multiple candidate templates for application reasons corresponding to the target scenario according to the vault scenario reason recommendation library; otherwise, submitting the initial vault application reason to manual review, and updating the vault scenario standard term library and the vault scenario reason recommendation library based on the initial vault application reason that has passed the manual review.

[0054] like Figure 2 As shown, the method in this embodiment specifically includes the following steps:

[0055] S210, in response to the initial vault authorization request of the target object, semantic recognition is performed on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain the application reason keywords.

[0056] Among them, the standard vocabulary for the vault scenario is used to describe the relationship between the standard word segmentation of the application reason and the scenario.

[0057] S220: Based on the standard word segmentation of the application reason in each scenario in the standard word library of the vault scenario, the application reason keywords are matched with the target scenario corresponding to the application reason keywords.

[0058] S230, determine whether the target applicant has a need for a recommended application reason template.

[0059] In this embodiment, before recommending the application reason template, it is necessary to ask the target object (e.g., provide a "yes" or "no" option in a pop-up window) to confirm whether the target object has a need for the application reason template recommendation. If the target object selects "yes", it indicates that the target object has a need for the application reason template recommendation, and steps S240-S250 are executed; if the target object selects "no", it indicates that the target object does not have a need for the application reason template recommendation, and step S260 is executed.

[0060] S240: Based on the vault scenario reason recommendation library, determine multiple application reason candidate templates corresponding to the target scenario, and feed back the application reason candidate templates to the target object.

[0061] Among them, the vault scenario reason recommendation library is used to describe the relationship between application reason templates and scenarios.

[0062] S250, in response to the target object's update vault authorization request based on the target template of the application reason, extract the reason template parameter from the update vault application reason of the update vault authorization request as the parameter to be verified, and perform compliance verification on the parameter to be verified to obtain the target verification result.

[0063] Among them, the application reason target template refers to a candidate application reason template currently used by the target object.

[0064] S260: The initial vault application reasons are submitted for manual review. Based on the manually approved initial vault application reasons, the standard terminology library for vault scenarios and the recommended library for vault scenario reasons are updated.

[0065] In this embodiment, the standard vocabulary library and the recommendation library for vault scenarios can be updated based on vault application reasons that have not used the application reason template and have passed manual review, so as to improve the richness and adaptability of the two libraries. At the same time, the vault word segmentation model is incrementally trained to improve the accuracy and adaptability of the model.

[0066] Specifically, Chinese is modeled based on expert-annotated parts-of-speech tags and statistical features, meaning the model parameters are trained using observed data (annotated corpus). A self-extraction method for training data is employed, automatically mining data from search logs. Implicit user annotations regarding term importance are extracted from massive log data. The resulting training data integrates annotations from hundreds of millions of users, offering broader coverage and deriving from real search data. The training results closely approximate the distribution of the labeled target set, resulting in more accurate training data. For example, self-explanatory features of terms include term proper noun type, term part-of-speech tag, term ID, positional features, and term length. Cross-feature features between terms and text strings include literal cross-features between the term and other terms in the text string, transition probability features of the term moving to other terms in the text string, text classification of the term, text classification of the topic and text string, and cross-features of the topic. Training data is extracted from search session data. Since a user's core search intent remains constant throughout a search session, terms corresponding to the core intent are of higher importance. Training data is extracted from a historical short string relation resource library. In the short string extended relations, the more frequently a term appears, the more important it is. New scenarios and corresponding parameter sets are separated to prepare data and interfaces for the next step of data rationality verification.

[0067] The technical solution of this invention, before determining multiple candidate templates for application reasons corresponding to a target scenario based on the vault scenario reason recommendation library, further determines whether the target object has a need for application reason template recommendations. If so, multiple candidate templates for application reasons corresponding to the target scenario are determined based on the vault scenario reason recommendation library; otherwise, the initial vault application reason is submitted for manual review, and the vault scenario standard terminology library and the vault scenario reason recommendation library are updated based on the manually approved initial vault application reason. This technical solution, through intelligent semantic recognition and scenario-related template recommendation of vault application reasons, can ensure the validity of submitted application reasons, reduce the tedious manual identification workload caused by invalid reasons, and thus improve approval efficiency. At the same time, it reduces the potential authorization over-limit caused by unreasonable business acceptance and lowers the security risks of excessive authorization. In addition, when there is no need for application reason template recommendations, updating the library based on manually approved vault application reasons can improve the richness and adaptability of the library, which helps to improve the accuracy of compliance identification.

[0068] Example 3

[0069] Figure 3This is a schematic diagram of a device for verifying the compliance of a vault-pattern application, provided in Embodiment 3 of the present invention. This device can execute the method for verifying the compliance of a vault-pattern application provided in any embodiment of the present invention, and possesses the corresponding functional modules and beneficial effects of the method. For example... Figure 3 As shown, the device includes:

[0070] The keyword extraction module 310 is used to respond to the initial vault authorization request of the target object, and to perform semantic recognition on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario;

[0071] The scenario matching module 320 is used to perform scenario matching on the application reason keywords according to the application reason standard word segmentation of each scenario in the vault scenario standard word library to obtain the target scenario corresponding to the application reason keywords;

[0072] The template recommendation module 330 is used to determine multiple candidate templates for application reasons corresponding to the target scenario based on the vault scenario reason recommendation library, and to feed back the candidate templates for application reasons to the target object; wherein, the vault scenario reason recommendation library is used to describe the relationship between the application reason templates and the scenario;

[0073] The parameter verification module 340 is used to respond to the target object's update vault authorization request based on the application reason target template, extract the reason template parameter from the update vault application reason of the update vault authorization request as the parameter to be verified, and perform compliance verification on the parameter to be verified to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object.

[0074] Optionally, the apparatus further includes: a model training module, used for:

[0075] Historical vault application reasons and their corresponding keyword annotations are obtained as the first training set, and standard vault application reasons and their corresponding keyword annotations are obtained as the second training set; wherein, the standard vault application reasons are generated based on synthetic data technology;

[0076] Using low-rank matrix technology and DeepSpeed ​​technology, the AI ​​model is trained under supervision based on the first training set and the second training set to obtain the Jinku word segmentation model.

[0077] Optionally, the vault word segmentation model is constructed based on a Trie tree structure, which is generated using a word segmentation method based on statistics and string matching using the standard vocabulary of the vault scenario.

[0078] Accordingly, the keyword extraction module 310 is used for:

[0079] The directed acyclic graph formed by all word combinations of the initial vault application reason generated by the vault word segmentation model is used as the candidate word segmentation path for the initial vault application reason;

[0080] The dynamic programming method is used to find the path with the highest probability among the candidate word segmentation paths as the target word segmentation path, and the application reason keywords are determined based on the target word segmentation path.

[0081] Optionally, the scene matching module 320 is used for:

[0082] Multiple candidate similarities are obtained by determining the similarity between the standard word segmentation of the application reason in each scenario in the standard thesaurus of the vault scenario and the keywords of the application reason;

[0083] If the candidate similarity is greater than or equal to a preset similarity threshold, then the candidate similarity is determined as the reference similarity.

[0084] The maximum value is determined from the reference similarity as the target similarity, and the scene corresponding to the target similarity is determined as the target scene.

[0085] Optionally, the device further includes: an audit log module, used for:

[0086] If the similarity of all the candidate similarities is less than the preset similarity threshold, then the target verification result of the initial vault authorization request is determined to be verification failure.

[0087] Determine whether the keywords of the application reason match the candidate scenario, where the candidate scenario refers to other scenarios outside the standard thesaurus of the vault scenario;

[0088] If so, an audit record will be generated based on the initial vault application reason and its corresponding candidate scenario.

[0089] Optionally, the apparatus further includes: a template requirement determination module, used for:

[0090] Before determining multiple application reason candidate templates corresponding to the target scenario based on the vault scenario reason recommendation library, it is determined whether the target object has a need for application reason template recommendations;

[0091] If so, then multiple application reason candidate templates corresponding to the target scenario are determined based on the vault scenario reason recommendation library;

[0092] Otherwise, the initial vault application reasons will be submitted for manual review, and the vault scenario standard terminology library and the vault scenario reason recommendation library will be updated based on the initial vault application reasons that have passed the manual review.

[0093] Optionally, the parameter verification module 340 is used for:

[0094] The parameter verification interface corresponding to the target scenario is called to perform compliance verification on the parameter to be verified to obtain the target verification result; wherein, the target verification result is either verification passed or verification failed;

[0095] If the target verification result is successful, the weights of the multiple application reason candidate templates corresponding to the target scenario will be adjusted.

[0096] The vault model application reason compliance identification device provided in this embodiment of the invention can execute the vault model application reason compliance identification method provided in any embodiment of the invention, and has the corresponding functional modules and beneficial effects of the execution method.

[0097] Example 4

[0098] Figure 4 A schematic diagram of an electronic device 10, which can be used to implement embodiments of the present invention, is shown. The electronic device is intended to represent various forms of digital computers, such as laptop computers, desktop computers, workstations, personal digital assistants, servers, blade servers, mainframe computers, and other suitable computers. The electronic device can also represent various forms of mobile devices, such as personal digital processors, cellular phones, smartphones, wearable devices (e.g., helmets, glasses, watches, etc.), and other similar computing devices. The components shown herein, their connections and relationships, and their functions are merely illustrative and are not intended to limit the implementation of the invention described and / or claimed herein.

[0099] like Figure 4 As shown, the electronic device 10 includes at least one processor 11 and a memory, such as a read-only memory (ROM) 12 or a random access memory (RAM) 13, communicatively connected to the at least one processor 11. The memory stores computer programs executable by the at least one processor. The processor 11 can perform various appropriate actions and processes based on the computer program stored in the ROM 12 or loaded from storage unit 18 into the RAM 13. The RAM 13 can also store various programs and data required for the operation of the electronic device 10. The processor 11, ROM 12, and RAM 13 are interconnected via a bus 14. An input / output (I / O) interface 15 is also connected to the bus 14.

[0100] Multiple components in electronic device 10 are connected to I / O interface 15, including: input unit 16, such as keyboard, mouse, etc.; output unit 17, such as various types of displays, speakers, etc.; storage unit 18, such as disk, optical disk, etc.; and communication unit 19, such as network card, modem, wireless transceiver, etc. Communication unit 19 allows electronic device 10 to exchange information / data with other devices through computer networks such as the Internet and / or various telecommunications networks.

[0101] Processor 11 can be a variety of general-purpose and / or special-purpose processing components with processing and computing capabilities. Some examples of processor 11 include, but are not limited to, central processing unit (CPU), graphics processing unit (GPU), various special-purpose artificial intelligence (AI) computing chips, various processors running machine learning model algorithms, digital signal processors (DSPs), and any suitable processor, controller, microcontroller, etc. Processor 11 performs the various methods and processes described above, such as the vault pattern application justification compliance verification method.

[0102] In some embodiments, the vault pattern application justification compliance verification method may be implemented as a computer program tangibly contained in a computer-readable storage medium, such as storage unit 18. In some embodiments, part or all of the computer program may be loaded and / or installed on electronic device 10 via ROM 12 and / or communication unit 19. When the computer program is loaded into RAM 13 and executed by processor 11, one or more steps of the vault pattern application justification compliance verification method described above may be performed. Alternatively, in other embodiments, processor 11 may be configured to perform the vault pattern application justification compliance verification method by any other suitable means (e.g., by means of firmware).

[0103] Various embodiments of the systems and techniques described above herein can be implemented in digital electronic circuit systems, integrated circuit systems, field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload-programmable logic devices (CPLDs), computer hardware, firmware, software, and / or combinations thereof. These various embodiments may include implementations in one or more computer programs that can be executed and / or interpreted on a programmable system including at least one programmable processor, which may be a dedicated or general-purpose programmable processor, capable of receiving data and instructions from a storage system, at least one input device, and at least one output device, and transmitting data and instructions to the storage system, the at least one input device, and the at least one output device.

[0104] Computer programs used to implement the methods of the present invention may be written in any combination of one or more programming languages. These computer programs may be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device, such that when executed by the processor, the computer programs cause the functions / operations specified in the flowcharts and / or block diagrams to be performed. The computer programs may be executed entirely on a machine, partially on a machine, or as a standalone software package, partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0105] In the context of this invention, a computer-readable storage medium can be a tangible medium that may contain or store a computer program for use by or in conjunction with an instruction execution system, apparatus, or device. A computer-readable storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination thereof. Alternatively, a computer-readable storage medium may be a machine-readable signal medium. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0106] To provide interaction with a user, the systems and techniques described herein can be implemented on an electronic device having: a display device (e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor) for displaying information to the user; and a keyboard and pointing device (e.g., a mouse or trackball) through which the user provides input to the electronic device. Other types of devices can also be used to provide interaction with the user; for example, feedback provided to the user can be any form of sensory feedback (e.g., visual feedback, auditory feedback, or tactile feedback); and input from the user can be received in any form (including sound input, voice input, or tactile input).

[0107] The systems and technologies described herein can be implemented in computing systems that include backend components (e.g., as data servers), or middleware components (e.g., application servers), or frontend components (e.g., user computers with graphical user interfaces or web browsers through which users can interact with implementations of the systems and technologies described herein), or any combination of such backend, middleware, or frontend components. The components of the system can be interconnected via digital data communication of any form or medium (e.g., communication networks). Examples of communication networks include local area networks (LANs), wide area networks (WANs), blockchain networks, and the Internet.

[0108] A computing system can include clients and servers. Clients and servers are generally located far apart and typically interact through communication networks. The client-server relationship is created by computer programs running on the respective computers and having a client-server relationship with each other. The server can be a cloud server, also known as a cloud computing server or cloud host, which is a hosting product within the cloud computing service system to address the shortcomings of traditional physical hosts and VPS services, such as high management difficulty and weak business scalability.

[0109] It should be understood that the various forms of processes shown above can be used, with steps reordered, added, or deleted. For example, the steps described in this invention can be executed in parallel, sequentially, or in different orders, as long as the desired result of the technical solution of this invention can be achieved, and this is not limited herein.

[0110] The specific embodiments described above do not constitute a limitation on the scope of protection of this invention. Those skilled in the art should understand that various modifications, combinations, sub-combinations, and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of this invention should be included within the scope of protection of this invention.

Claims

1. A method for verifying the compliance of application reasons for a vault model, characterized in that, The method includes: In response to the initial vault authorization request from the target object, semantic recognition is performed on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario; Based on the standard word segmentation of the application reason in each scenario in the standard vocabulary library of the vault scenario, the application reason keywords are matched to obtain the target scenario corresponding to the application reason keywords; Based on the vault scenario reason recommendation library, multiple application reason candidate templates corresponding to the target scenario are determined, and the application reason candidate templates are fed back to the target object; wherein, the vault scenario reason recommendation library is used to describe the association between application reason templates and scenarios; In response to the target object's update vault authorization request based on the application reason target template, the reason template parameter is extracted from the update vault application reason of the update vault authorization request as a parameter to be verified, and the parameter to be verified is subjected to compliance verification to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object.

2. The method according to claim 1, characterized in that, The training process of the vault word segmentation model includes: Historical vault application reasons and their corresponding keyword annotations are obtained as the first training set, and standard vault application reasons and their corresponding keyword annotations are obtained as the second training set; wherein, the standard vault application reasons are generated based on synthetic data technology; Using low-rank matrix technology and DeepSpeed ​​technology, the AI ​​model is trained under supervision based on the first training set and the second training set to obtain the Jinku word segmentation model.

3. The method according to claim 1 or 2, characterized in that, The vault word segmentation model is constructed based on a Trie tree structure, which is generated using a word segmentation method based on statistics and string matching using a standard vocabulary for the vault scenario. Accordingly, based on the vault word segmentation model and the standard vault scenario lexicon, semantic recognition is performed on the initial vault application reason in the initial vault authorization request to obtain application reason keywords, including: The directed acyclic graph formed by all word combinations of the initial vault application reason generated by the vault word segmentation model is used as the candidate word segmentation path for the initial vault application reason; The dynamic programming method is used to find the path with the highest probability among the candidate word segmentation paths as the target word segmentation path, and the application reason keywords are determined based on the target word segmentation path.

4. The method according to claim 1, characterized in that, Based on the standard word segmentation of application reasons in various scenarios in the vault scenario standard thesaurus, the application reason keywords are matched to obtain the target scenarios corresponding to the application reason keywords, including: Multiple candidate similarities are obtained by determining the similarity between the standard word segmentation of the application reason in each scenario in the standard thesaurus of the vault scenario and the keywords of the application reason; If the candidate similarity is greater than or equal to a preset similarity threshold, then the candidate similarity is determined as the reference similarity. The maximum value is determined from the reference similarity as the target similarity, and the scene corresponding to the target similarity is determined as the target scene.

5. The method according to claim 4, characterized in that, The method further includes: If the similarity of all the candidate similarities is less than the preset similarity threshold, then the target verification result of the initial vault authorization request is determined to be verification failure. Determine whether the keywords of the application reason match the candidate scenario, where the candidate scenario refers to other scenarios outside the standard thesaurus of the vault scenario; If so, an audit record will be generated based on the initial vault application reason and its corresponding candidate scenario.

6. The method according to claim 1, characterized in that, Before determining multiple candidate templates for application reasons corresponding to the target scenario based on the vault scenario reason recommendation library, the process also includes: Determine whether the target object has a need for application reason template recommendations; If so, then multiple application reason candidate templates corresponding to the target scenario are determined based on the vault scenario reason recommendation library; Otherwise, the initial vault application reasons will be submitted for manual review, and the vault scenario standard terminology library and the vault scenario reason recommendation library will be updated based on the initial vault application reasons that have passed the manual review.

7. The method according to claim 1, characterized in that, The compliance verification of the parameters to be verified is performed to obtain the target verification result, including: The parameter verification interface corresponding to the target scenario is called to perform compliance verification on the parameter to be verified to obtain the target verification result; wherein, the target verification result is either verification passed or verification failed; If the target verification result is successful, the weights of the multiple application reason candidate templates corresponding to the target scenario will be adjusted.

8. A device for verifying the compliance of a vault-style application, characterized in that, The device includes: The keyword extraction module is used to respond to the initial vault authorization request of the target object, and to perform semantic recognition on the initial vault application reason in the initial vault authorization request based on the vault word segmentation model and the vault scenario standard lexicon to obtain application reason keywords; wherein, the vault scenario standard lexicon is used to describe the association relationship between the standard word segmentation of the application reason and the scenario; The scenario matching module is used to perform scenario matching on the application reason keywords according to the standard word segmentation of the application reason under each scenario in the standard word library of the vault scenario to obtain the target scenario corresponding to the application reason keywords; The template recommendation module is used to determine multiple candidate templates for application reasons corresponding to the target scenario based on the vault scenario reason recommendation library, and to feed back the candidate templates for application reasons to the target object; wherein, the vault scenario reason recommendation library is used to describe the association between application reason templates and scenarios; The parameter verification module is used to respond to the target object's update vault authorization request based on the application reason target template, extract the reason template parameter from the update vault application reason of the update vault authorization request as the parameter to be verified, and perform compliance verification on the parameter to be verified to obtain the target verification result; wherein, the application reason target template refers to an application reason candidate template currently used by the target object.

9. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the vault pattern application justification compliance identification method according to any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer instructions that, when executed by a processor, implement the method for verifying the compliance of a vault pattern application justification as described in any one of claims 1-7.

Images