An abnormal information processing method, device, equipment, medium and product

Abstract

The application discloses an abnormal information processing method, device and equipment, and a medium and product in the technical field of computers. The method comprises the following steps: receiving a configuration operation, the configuration operation being capable of indicating first configuration information corresponding to a first object; and displaying abnormal information, the abnormal information comprising at least one of first information and second information. The first information indicates that the first configuration information has a semantic exception, so that the abnormal information comprising the first information can at least indicate that the first configuration information has a semantic exception. In addition, the second information indicates that the first configuration information does not satisfy a configuration constraint, and the configuration constraint is determined based on second configuration information corresponding to a second object similar to the first object, so that the abnormal information comprising the second information can at least indicate that the first configuration information has an exception in a general practice rule. It can be seen that the abnormal information can comprehensively indicate the exception of the first configuration information as much as possible, so as to avoid the influence caused by the fact that the user does not know the exception.

Description

Technical Field

[0001] This relates to the field of computer technology, and in particular to a method, apparatus, device, medium, or product for processing abnormal information. Background Technology

[0002] In some fields, specific individuals configure objects within that field, such as promotional activities or special offers. However, the information configured for these objects may contain anomalies, leading to various consequences. Summary of the Invention

[0003] An abnormal information processing method, apparatus, equipment, medium, or product is provided to overcome the aforementioned impacts.

[0004] In one scenario, an anomaly information processing method includes: receiving a configuration operation, the configuration operation indicating first configuration information; displaying anomaly information, the anomaly information including at least one of first information and second information, the first information indicating that the first configuration information has a semantic anomaly, the second information indicating that the first configuration information does not meet configuration constraints, the configuration constraints being determined based on second configuration information, and a first object corresponding to the first configuration information being similar to a second object corresponding to the second configuration information.

[0005] In one scenario, an anomaly information processing device includes: a receiving unit for receiving a configuration operation, the configuration operation indicating first configuration information; and a display unit for displaying anomaly information, the anomaly information including at least one of first information and second information, the first information indicating that the first configuration information has a semantic anomaly, the second information indicating that the first configuration information does not meet configuration constraints, the configuration constraints being determined based on second configuration information, and a first object corresponding to the first configuration information being similar to a second object corresponding to the second configuration information.

[0006] In one embodiment, an electronic device includes: a processor and a memory; the memory for storing instructions or computer programs; and the processor for executing the instructions or computer programs in the memory to cause the electronic device to perform the exception information handling method provided in this specification.

[0007] In one scenario, a computer-readable medium storing instructions or a computer program that, when executed on a device, causes the device to perform the exception information handling method provided in this specification.

[0008] In one scenario, a computer program product includes a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the exception information handling methods provided in this specification.

[0009] Based on the above, the technical solution provided in this specification includes: receiving a configuration operation, which indicates first configuration information corresponding to a first object (such as a new activity); and displaying exception information, which includes at least one of first information and second information. Specifically, because the first information indicates a semantic error in the first configuration information (e.g., a new activity is applicable to a new user, but the restriction "the activity is only applicable to new users" is not configured, resulting in a semantic error), the first information can indicate a semantically present configuration error (such as a logical error, semantic conflict, etc.) in the first configuration information, thereby ensuring that the exception information including the first information at least indicates a semantically present configuration error in the first configuration information. Furthermore, because the second information indicates that the first configuration information does not meet configuration constraints (such as the general practice of "configuring frequency limits when the discount exceeds a threshold"), and these constraints are determined based on the second configuration information corresponding to a second object similar to the first object (such as historical activities), the second information can indicate that the first configuration information deviates significantly from the configuration constraints determined from the configuration information of similar objects. Therefore, the anomaly information including the second information can at least indicate configuration errors in the first configuration information based on some general practices. Thus, the anomaly information can comprehensively indicate various anomalies in the first configuration information (such as semantic contradictions, non-compliance with historical patterns, etc.), enabling users to recognize these anomalies and avoid consequences caused by users' lack of awareness of these anomalies. Attached Figure Description

[0010] To more clearly illustrate the technical solutions, the accompanying drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the accompanying drawings described below only illustrate some embodiments. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0011] Figure 1 A schematic diagram of an implementation environment for the exception information handling method provided in this specification; Figure 2 A flowchart of an exception information processing method provided in this specification; Figure 3 This is an example diagram of a configuration process provided in this manual; Figure 4This is a schematic diagram of the structure of an abnormal information processing device provided in this specification; Figure 5 This is a schematic diagram of the structure of an electronic device provided in this specification. Detailed Implementation

[0012] The following description, with reference to the accompanying drawings, will detail one or more embodiments of the technical solution under various conditions. While some embodiments of the technical solution are shown in the drawings, it should be understood that the technical solution can be implemented in various forms under various conditions 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 the technical solution. It should be understood that the accompanying drawings and embodiments are for illustrative purposes only and are not intended to limit the scope of protection of the technical solution.

[0013] It should be understood that the steps described in the method implementation of the technical solution can be performed in different orders and / or in parallel. Furthermore, the method implementation may include additional steps and / or omit the steps shown. The scope of the technical solution is not limited in this respect.

[0014] The term "comprising" and its variations, used in describing technical solutions, are open-ended inclusions, meaning "including but not limited to," so that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The term "based on" means "at least partially based on." The term "according to" means "at least partially based on." The term "one embodiment" means "at least one embodiment"; the term "another embodiment" means "at least one additional embodiment"; the term "some embodiments" means "at least some embodiments." Definitions of other terms will be given in the following description.

[0015] It should be noted that the concepts of "first" and "second" mentioned in the technical solution are only used to distinguish different entities, operations, devices, modules or units, and are not used to limit the order of functions performed by these devices, modules or units or their interdependencies.

[0016] It should be noted that the terms "one" and "more" mentioned in the technical solution are illustrative rather than restrictive. Those skilled in the art should understand that, unless otherwise expressly indicated in the context, they should be understood as "one or more".

[0017] The names of messages or information exchanged between multiple devices in the various embodiments of the technical solution are for illustrative purposes only and are not intended to limit the scope of these messages or information.

[0018] It is understood that the data involved in this technical solution (including but not limited to the data itself, the acquisition, use, storage or deletion of the data) shall comply with the requirements of relevant laws, regulations and related provisions.

[0019] It is understood that before using the technical solutions disclosed in the various embodiments of the technical solution, the relevant users should be informed of the type, scope of use, and usage scenarios of the information involved in the technical solution and their authorization should be obtained in accordance with relevant laws and regulations through appropriate means. The relevant users may include any type of rights holder, such as individuals, enterprises, or groups.

[0020] For example, in response to receiving an active request from a user, a prompt message is sent to the relevant user to clearly inform the user that the requested operation will require obtaining and using the user's information, thereby enabling the relevant user to choose whether to provide information to the software or hardware such as the electronic device, application, server, or storage medium that performs the operation of any embodiment of the technical solution involved in the technical solution based on the prompt message.

[0021] As an optional but non-restrictive implementation, in response to a user's active request, a prompt message can be sent to the user, such as a pop-up window, where the prompt message can be presented in text format. Furthermore, the pop-up window can also include a selection control allowing the user to choose "agree" or "disagree" to provide information to the electronic device.

[0022] It is understood that the above notification and user authorization process are merely illustrative and do not constitute a limitation on the implementation methods involved in the technical solution. Other methods that comply with relevant laws and regulations may also be applied to the implementation methods involved in the technical solution.

[0023] Before providing a detailed introduction to the abnormal information handling methods provided in this manual, we will combine... Figure 1 This document describes one possible implementation environment for the anomaly information processing method. However, this specification does not limit the implementation environment of the anomaly information processing method; it can be any of the following: Figure 1 The implementation environment shown can also be other implementation environments. Among them, the Figure 1 This diagram illustrates an implementation environment for the exception information handling method provided in this specification. The implementation environment includes a configuration platform front-end and an exception analysis service.

[0024] For the above configuration platform front-end (such as...) Figure 1In the case of the configuration platform frontend shown, the configuration platform frontend is configured to provide object configuration services to users, enabling them to trigger configuration operations on an object (such as an activity) using the configuration platform frontend, obtain the object's configuration information, and subsequently perform other processing flows based on this configuration information, such as saving the configuration information to the backend service corresponding to the configuration platform frontend, publishing the configuration information online, or sending the configuration information to other services (such as any third-party service or...). Figure 1 The process includes handling procedures such as anomaly analysis services. The backend service is configured to provide data services to the configuration platform frontend, such as data storage services, data query services, and data parsing services. This specification does not limit the backend service; for example, in one scenario, the backend service can be implemented using any device capable of providing data services to the frontend, such as any type of server (e.g., a standalone server, a cluster server, or a cloud server). Furthermore, this specification does not limit the configuration platform frontend; for example, in one scenario, the configuration platform frontend can be implemented using any platform frontend capable of providing configuration services to users, such as a platform frontend presented via a terminal device. This terminal device can be a smartphone, computer, personal digital assistant (PDA), tablet computer, etc.

[0025] For the above-mentioned anomaly analysis services (such as...) Figure 1In the case of the anomaly analysis service shown, this anomaly analysis service is configured to perform anomaly analysis on the configuration information sent by the configuration platform frontend to obtain anomaly information and corresponding modification suggestions. This allows the anomaly information to indicate the type of anomaly in the configuration information (such as semantic configuration errors, configuration constraint errors, etc.), and the modification suggestions to indicate how to modify the configuration information to overcome the anomaly. Subsequently, the anomaly analysis service will feed back the anomaly information and corresponding modification suggestions to the user in a certain way, such as displaying the anomaly information and corresponding modification suggestions through the configuration platform frontend, or sending the anomaly information and corresponding modification suggestions back to the user through other tools (such as instant messaging services). This ensures that the user is aware of the anomalies in the configuration information and their solutions as soon as possible after configuration is completed, overcoming the impact of the user's lack of awareness of the anomaly (such as resource loss, over-issuance of discounts, etc.). This allows the user to subsequently use the configuration platform frontend to adjust the configuration information based on the anomaly information, thereby improving the configuration effect. Furthermore, this specification does not limit the deployment method between the anomaly analysis service and the configuration platform front-end. For example, in one scenario, the anomaly analysis service can operate independently of the configuration platform front-end, while still being able to communicate with it. Alternatively, in another scenario, the anomaly analysis service can be deployed internally within the configuration platform front-end, enabling it to provide not only object configuration services but also configuration anomaly analysis services. Moreover, this specification does not limit the anomaly analysis service; for instance, in one scenario, the anomaly analysis service can be implemented using any device capable of providing anomaly analysis services, such as a standalone server, a cluster server, or a cloud server.

[0026] To enable those skilled in the art to better understand the technical solutions provided in this specification, the technical solutions will be clearly and completely described below with reference to the accompanying drawings of the embodiments involved. Obviously, the described embodiments are only some embodiments, not all embodiments. Based on these embodiments, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the technical solutions.

[0027] To better understand the technical solution, the abnormal information handling method will be explained below with reference to some accompanying diagrams. For example... Figure 2 As shown, the abnormal information processing method includes steps 201-202 as described below. Among them, the... Figure 2 This is a flowchart of an exception information processing method provided in this specification.

[0028] Step 201: Receive configuration operation, which indicates the first configuration information.

[0029] The configuration operation refers to the operation triggered when a user configures a first object (such as a promotional activity) using the configuration platform front end, so that the configuration operation can indicate the configuration information set for the first object, such as the activity name, the activity time, and the scope of use of the activity.

[0030] It should be noted that the first object mentioned above refers to an object with configuration requirements, and this specification does not limit the first object. For example, in one case, the first object can be any object with configuration requirements, such as a certain activity (such as a promotional activity), a certain application, or a certain device.

[0031] In another scenario, the configuration operation described above may be received via a configuration platform front-end, enabling the configuration operation to indicate the configuration information set by the user for the first object using the front-end. This configuration platform front-end is configured to provide object configuration services to the user; however, this specification does not limit the scope of the configuration platform front-end.

[0032] Furthermore, in one scenario, when the configuration platform frontend provides object configuration services to the user through a configuration page, the aforementioned configuration operation can be received via this configuration page, enabling the configuration operation to indicate the configuration information set by the user for the first object using the configuration page. Here, the configuration page is configured to provide object configuration services to the user; and this application does not limit the configuration page, for example, it can present the configuration information in the form of form data, so that the configuration information can be obtained subsequently by retrieving the form data.

[0033] Furthermore, this specification does not limit the above configuration operations. For example, in one scenario, the configuration operation may include at least one of a save operation and a submit operation. The save operation is configured to instruct the saving of the configuration information presented on the configuration page. The submit operation is configured to instruct the provision of the configuration information presented on the configuration page to subsequent processes so that these subsequent processes can use the configuration information to perform other tasks, such as configuration verification or configuration publishing.

[0034] The first configuration information refers to the configuration information obtained based on the above configuration operations, so that the first configuration information can indicate the configuration information set by the user for the first object, so that the first configuration information can describe some characteristics of the first object, such as name, text description, actual rules, scope of use, etc.

[0035] Furthermore, this specification does not limit the aforementioned first configuration information. For example, it can be implemented using any information capable of describing object configuration, such as key-value pairs. Therefore, in one scenario, the first configuration information includes at least one key-value pair. Specifically, a key-value pair is configured to record field information (e.g., <field name, field configuration value>), where the "key" is configured to record the field name, and the "value" is configured to record the field's configuration value. Based on this, the first configuration information can include at least one field, and different fields indicate different characteristics of the first object. For any given field, the field information includes a field name and a field configuration value, enabling the field information to represent the configuration result for a specific field in a structured manner (e.g., key-value pairs).

[0036] Furthermore, this specification does not limit the method of obtaining the aforementioned first configuration information. For example, in one scenario, when the configuration operation is received via a configuration platform front-end, the first configuration information may be obtained by the configuration platform front-end from its corresponding back-end service, so that the first configuration information includes the configuration information stored by the back-end service for the first object. The back-end service is configured to provide services (such as configuration information query services) to the configuration platform front-end, enabling the configuration platform front-end to better provide services to the user with the assistance of the back-end service. Additionally, this specification does not limit the scope of the back-end service.

[0037] As can be seen, in one scenario, after receiving the configuration operation, the configuration platform front-end can determine that the user has saved or submitted the configuration information set for the first object. Therefore, the configuration platform front-end queries the configuration information from the back-end service corresponding to the configuration platform front-end by calling the Application Programming Interface (API) to obtain the first configuration information.

[0038] Furthermore, in one scenario, when the aforementioned configuration platform frontend provides object configuration services to users via a configuration page, the process of obtaining the aforementioned first configuration information may include: after the configuration platform frontend detects that the configuration operation triggered by the user has been received via the configuration page, the configuration platform frontend first initiates an exception analysis request to Functions as a Service (FaaS) asynchronously, so that the request carries the associated information of the first object, such as the object identifier of the first object; then, the FaaS searches for the associated information of the configuration page based on the information carried in the request, such as the access address of the configuration page, the login information of the configuration page, etc.; then, the FaaS calls the headless browser module based on the associated information of the configuration page, so that the headless browser module can simulate the real user environment of the user logging into the configuration page based on the associated information of the configuration page. In this way, the headless browser technology is used to completely execute the page script of the configuration page, render dynamic content, etc., so as to collect the document object model (Document Object Model) of the complete configuration page finally presented to the user. The Model (DOM) is used to fully describe the complete front-end form data presented by the configuration page, including at least all nested configurations. This allows the DOM to fully represent the configuration information presented by the configuration page, enabling subsequent determination of the first configuration information based on the DOM. This ensures that the first configuration information includes the DOM, allowing for the acquisition of complete configuration information through page DOM retrieval. This overcomes the shortcomings caused by some page content crawling methods being unable to handle dynamically generated content (such as content accessed through the address presented on the current page). Thus, the accuracy of configuration information retrieval can be improved without affecting front-end performance, overcoming the defects caused by incomplete configuration information retrieval and improving the anomaly detection effect.

[0039] It should be noted that the aforementioned "asynchronous mode" is configured to decouple the collection process of the first configuration information from the main flow of the configuration platform frontend, so that the collection process can be executed through an independent, stateless function computing service (such as FaaS). This ensures that even if the collection process is time-consuming, the configuration platform frontend can still respond quickly to user operations. Furthermore, this specification does not limit the access address of the aforementioned configuration page. For example, the access address can be implemented using any data that can uniquely identify the page access path, such as a Uniform Resource Locator (URL). In addition, this specification does not limit the login information of the configuration page. For example, when a user triggers the aforementioned configuration operation through the configuration page, the login information can include the identity credentials (such as temporary identity credentials) assigned to the user when using the configuration page, so that the identity credentials and headless browser technology can be used to simulate the user's login and access to the configuration page, thereby collecting the complete DOM of the configuration page. Also, this specification does not limit the aforementioned headless browser module. For example, the headless browser module can be implemented using any headless browser technology. Furthermore, this specification does not limit the aforementioned FaaS and headless browser modules. For example, both can be deployed within the configuration platform frontend, enabling the frontend to asynchronously retrieve page information (such as the DOM) of the configuration page. Alternatively, the FaaS and headless browser modules can be independent of the configuration platform frontend, allowing subsequent asynchronous retrieval of the configuration page's information through data communication between the three.

[0040] Based on the above two paragraphs, it can be understood that, in one scenario, when the configuration operation is received via a configuration page, the first configuration information can be determined based on the page information (such as the DOM) of the configuration page, so that the first configuration information includes the page information, thereby enabling the first configuration information to represent the configuration information set for the first object in a DOM-based manner. Here, the page information is used to describe the configuration page; moreover, this specification does not limit the page information, for example, the page information may at least include the DOM of the configuration page.

[0041] Based on the aforementioned content of the first configuration information, it can be understood that, in one scenario, the first configuration information may include at least one of the configuration information obtained from the aforementioned backend service and the page information (such as DOM) of the aforementioned configuration page, so that the first configuration information can represent the configuration information set for the first object.

[0042] Based on the relevant content of step 201 above, in one scenario, when the configuration platform front-end displays a configuration page, in response to the configuration platform front-end detecting the above-mentioned configuration operation received through the configuration page, the configuration platform front-end obtains the first configuration information indicated by the configuration operation through a certain method (such as searching for configuration information from the back-end service, or collecting the front-end page by using asynchronous calls to FaaS and headless browser technology, etc.), so that configuration anomaly identification can be realized based on the first configuration information in the future.

[0043] Step 202: Display abnormal information, which includes at least one of first information and second information. The first information indicates that the first configuration information has a semantic abnormality, and the second information indicates that the first configuration information does not meet the configuration constraints. The configuration constraints are determined based on the second configuration information, and the first object corresponding to the first configuration information is similar to the second object corresponding to the second configuration information.

[0044] The exception information can indicate the exceptions that exist in the first configuration information and the related information of the exception, such as the name of the exception, the description of the problem, and the potential impact of the exception. This allows users to understand not only what configuration errors exist in the first configuration information, but also some characteristics of these configuration errors.

[0045] In another scenario, step 202 may include displaying anomaly information and corresponding modification suggestions, so that users can not only understand which configuration errors exist in the first configuration information through the anomaly information, but also understand the solutions to the configuration errors through the modification suggestions. This allows users to subsequently decide whether to modify the first configuration information to overcome the problems caused by the anomaly based on the anomaly information and its corresponding modification suggestions, thereby ensuring that users do not need to manually attribute the causes to obscure data (such as logs, indicator curves, etc.) to improve configuration effectiveness.

[0046] Based on the above content, in one scenario, the above-mentioned abnormal information handling method may also include: displaying modification suggestions corresponding to the above-mentioned abnormal information, so that users can quickly understand how to modify the above-mentioned first configuration information to overcome its abnormality through the modification suggestions, so as to ensure that users do not have to manually attribute obscure data (such as logs, indicator curves, etc.) to improve the configuration effect.

[0047] Furthermore, this specification does not limit the method of obtaining the above-mentioned abnormal information. For example, in one case, when a third method (such as anomaly detection according to hard rules) is configured to detect syntax errors from configuration information, the process of obtaining the abnormal information may include: processing the first configuration information according to the third method to obtain third information, so that the third information can indicate which syntax errors exist in the first configuration information and the associated information of the syntax errors, such as the name of the syntax error, the problem description of the syntax error, the potential impact of the syntax error, etc.; determining the abnormal information based on the third information, so that the abnormal information at least includes the third information, thereby enabling the abnormal information to at least indicate the syntax errors existing in the first configuration information and the associated information of the syntax errors.

[0048] It should be noted that the third method mentioned above is configured to identify whether any configuration information has a syntactic error. Moreover, this specification does not limit the third method. For example, in one case, the third method can be implemented using any method that can achieve syntax error detection, such as pre-built rules (such as regular expressions), scripts, or machine learning models with such detection function.

[0049] Therefore, in one scenario, the aforementioned third information can be obtained by processing the first configuration information using a syntax detection model, so that the third information can indicate the syntax errors present in the first configuration information and the associated information of those syntax errors. The syntax detection model has syntax error detection functionality; moreover, this specification does not limit the syntax detection model. For example, the syntax detection model can be any machine learning model with this detection functionality, such as a Large Language Model (LLM). The LLM can implement the syntax error detection function under the guidance of its corresponding prompt word, which is configured to guide the LLM on how to detect whether there are syntax errors in the first configuration information according to certain rules (such as grammar rules).

[0050] For example, in one scenario, the aforementioned third method can be implemented using any executable strategy, such as a preset strategy. Thus, in one scenario, the aforementioned third information can be obtained by using a preset strategy to identify anomalies in the first configuration information, enabling the third information to indicate syntax errors present in the first configuration information and their associated information. This preset strategy includes some hard rules (such as syntax rules) to indicate the syntax constraints that any configuration information must satisfy, so that subsequent detection of syntax errors in the configuration information can be based on this preset strategy. Different rules recorded in the preset strategy indicate different syntax constraints, allowing the preset strategy to comprehensively represent the numerous syntax constraints that the configuration information must satisfy. Furthermore, this preset strategy can be developed by relevant personnel or automatically generated based on a large amount of literature; this specification does not specifically limit its application.

[0051] Research has revealed that in some cases, configuration information may be syntactically correct, but it may contain other issues such as semantic logic errors (e.g., illogical business logic exceptions), semantic conflicts, non-compliance with general practices, or other anomalies that could lead to resource loss, making anomaly identification difficult. For example, when the first object pertains to a promotional activity applicable to new users, if the first configuration information corresponding to that first object does not record configuration information regarding the scope of use, it can be determined that the first configuration information contains a semantic logic error, such as the absence of the restriction "only available to new users."

[0052] Based on the above research, to overcome the difficulties described above, the process of obtaining the aforementioned abnormal information may include: processing the first configuration information using at least one method to obtain abnormal information, such that the abnormal information includes at least one of first information and second information, thereby enabling the abnormal information to at least indicate problems in the first configuration information other than syntax errors. Specifically, the at least one method includes at least one of a first method and a second method. The first method is configured to identify semantic anomalies from the first configuration information to obtain the first information, and the second method is configured to detect whether the first configuration information satisfies configuration constraints extracted from the configuration information of similar objects (such as a second object) to obtain the second information.

[0053] Regarding the first method described above, this method is configured to detect whether the first configuration information has semantic anomalies, such as semantic logic errors (e.g., the absence of the restriction "only available to new users" in the first configuration information, or some illogical business logic anomalies), semantic conflicts, or other anomalies that may cause resource loss. The aim is to identify syntactically correct but semantically problematic anomalies through this first method, ultimately obtaining first information that indicates a semantic configuration error in the first configuration information. Furthermore, this specification does not limit this first method; for example, this first method can be implemented using any machine learning model (such as LLM) capable of providing this detection function.

[0054] Therefore, in one scenario, the process of determining the aforementioned first information may include: processing the aforementioned first configuration information using a semantic detection model (such as LLM) to obtain the first information, so that the first information can indicate the semantic anomalies present in the first configuration information and the associated information of the semantic anomalies, such as the name of the semantic anomaly, the problem description of the semantic anomaly, and the potential impact of the semantic anomaly. The semantic detection model has a semantic anomaly detection function; moreover, this specification does not limit the semantic detection model. For example, the semantic detection model can be implemented using any machine learning model (such as LLM) with this detection function. It should be noted that the LLM can implement the semantic anomaly detection function under the guidance of its corresponding prompt words. The prompt words are configured to guide the LLM on how to identify various semantically existing problems, and the prompt words can be pre-set based on the actual application scenario.

[0055] It should be noted that this specification does not limit the input data of the semantic detection model. For example, when the first configuration information includes the page information of the configuration page, if the language used by the page information (such as Chinese) is the same as the language that the semantic detection model can understand, then the input data of the semantic detection model may at least include the page information. Similarly, when the first configuration information includes configuration information obtained from a backend service, if the language used by the "configuration information obtained from the backend service" (such as English) is different from the language that the semantic detection model can understand, then the input data of the semantic detection model may at least include the result obtained by language conversion of the "configuration information obtained from the backend service" to ensure that the input data can represent the configuration information set for the first object in the language that the semantic detection model can understand. Furthermore, when the first configuration information includes the page information of the configuration page and the configuration information obtained from the backend service, if the language used by the page information is the same as the language that the semantic detection model can understand, but the language used by the "configuration information obtained from the backend service" is different from the language that the semantic detection model can understand, then the input data of the semantic detection model may at least include the page information.

[0056] Regarding the second method described above, this second method is configured to detect whether the first configuration information satisfies the configuration constraints extracted from the configuration information of similar objects (such as constraints on the number of times each user can participate) to obtain the second information. Furthermore, this specification does not limit this second method; for example, this second method can be implemented using any machine learning model (such as LLM) capable of providing this detection function.

[0057] Therefore, in one scenario, the process of determining the second information may include: using a constraint detection model to compare the first configuration information with the configuration constraints to obtain the second information, so that the second information can indicate whether the first configuration information satisfies the configuration constraints extracted from the configuration information of similar objects (such as the second configuration information corresponding to the second object), thereby enabling the second information to indicate the anomalies presented by the first configuration information on the configuration constraints and the associated information of the anomalies, such as the name of the anomaly, the problem description of the anomaly, the potential impact of the anomaly, etc. The constraint detection model has the function of detecting whether the configuration constraints are satisfied; moreover, this specification does not limit the constraint detection model. For example, the constraint detection model can be implemented using any machine learning model (such as LLM) with this detection function. Furthermore, the LLM can detect whether a configuration information satisfies the configuration constraints under the guidance of its corresponding prompt word. The prompt word is configured to guide the LLM on how to detect whether the configuration information has anomalies based on the configuration constraints, and the prompt word can be pre-set based on the actual application scenario.

[0058] For the aforementioned second object, it refers to an object similar to the first object (such as historical activities of the same type), so that the second object can represent objects of the same kind as the first object. This allows the configuration information set for the second object to indicate, to some extent, certain rules (such as general practice rules) that need to be satisfied when configuring the first object. Furthermore, this specification does not limit the second object; for example, in one case, the second object can at least satisfy the following constraint: the second object is similar to the first object. Moreover, this specification does not limit the method of obtaining the second object; for example, it can be implemented using any method that can obtain similar objects based on an object.

[0059] It should be noted that the above "the second object is similar to the first object" means that the similarity between the two objects in terms of attributes reaches a certain condition (such as ranking higher, exceeding a threshold, etc.), so that "the second object is similar to the first object" can indicate that the configuration information set for the two objects has a high degree of similarity in some field information (such as the field information that records the attribute), such as exceeding a threshold.

[0060] In another scenario, the process of determining the second object may include: first, determining the attributes of the first object, such as scope of use and usage scenario, based on the first configuration information, so that the attributes accurately represent the characteristics of the first object; then, searching the database based on the attributes to obtain the second object, so that the similarity between the second object and the first object on the attribute reaches a preset condition (such as ranking higher, reaching a threshold, etc.). The database records the configuration information of historically published objects (such as the second configuration information corresponding to the second object), and some fields recorded in the second configuration information are configured to indicate the attributes of the second object, so that the attributes of the second object are determined based on the second configuration information. Furthermore, this specification does not limit the process of determining the second object. For example, the process of determining the second object may be implemented using any machine learning model (such as LLM) that implements the determination process. The LLM can perform similar object searches under the guidance of its prompt words, which are configured to guide the LLM on how to search for similar objects.

[0061] It is evident that, under certain circumstances, the aforementioned second object can at least satisfy the following constraints: the second object is similar to the first object, and the effective time (e.g., the time of deployment to the online system) of the configuration information set for the second object is earlier than the effective time of the configuration information set for the first object. This allows the second object to represent historical objects of the same type that were deployed online before the first object was deployed online. This also allows the configuration information set for the second object to indicate, to some extent, some general practice rules that need to be followed when configuring the first object. This enables subsequent detection of whether there are any anomalies in the configuration information set for the first object using these general practice rules, thereby improving the anomaly identification effect and overcoming the defects caused when such anomalies exist.

[0062] Based on the above, in one scenario, the second configuration information can at least satisfy the following constraints: the second object corresponding to the second configuration information is similar to the first object corresponding to the first configuration information, and the effective time of the second configuration information is earlier than the effective time of the first configuration information, so that the second configuration information can indicate some configuration constraints (such as general practice rules) that need to be followed when configuring the first object, so that these configuration constraints can be used to detect whether there is an anomaly in the first configuration information, thereby improving the anomaly identification effect and overcoming the defects caused when the anomaly exists.

[0063] Regarding the aforementioned configuration constraints, these constraints are determined based on the second configuration information corresponding to some second objects. This ensures that the configuration constraints accurately represent the constraints that most of these second objects must follow in their configuration information (e.g., 95% of the second objects have a constraint limiting the number of times each user can participate). This allows the configuration constraints to indicate the general practice rules corresponding to the type of these second objects, and thus indicate some general practice rules that must be followed when configuring objects of this type (such as the first object). Therefore, the configuration constraints can indicate the practical knowledge (such as general practice rules) involved in this type, thereby supplementing the knowledge gaps in the aforementioned semantic anomaly identification to a certain extent, thus improving the anomaly identification effect. It should be noted that this specification does not limit the configuration constraints. For example, the configuration constraints may include constraints of at least one dimension (such as semantic, syntactic, or logical dimensions) to comprehensively describe the content that must be followed when configuring objects of this type (such as the first object), ensuring that the configuration information satisfying the configuration constraints has as few anomalies as possible.

[0064] Furthermore, this specification does not limit the method of determining the above-mentioned configuration constraints. For example, it may specifically include: after obtaining the second configuration information of these second objects, firstly, performing statistical analysis on these second configuration information under the first field (such as discount level, usage threshold, participation frequency, etc.) to obtain the statistical analysis results corresponding to the first field, so that the statistical analysis results can indicate the distribution of these second configuration information under the first field; then, based on the statistical analysis results, determining the configuration constraints corresponding to the first field, so that the configuration constraints can indicate the configuration characteristics of most of these second objects under the first field, thereby enabling the configuration constraints to indicate the general practice rules that these second object types need to follow under the first field, and further enabling the configuration constraints to indicate some general practice rules that need to be followed when configuring objects of this type (such as the first object), so that it is possible to better identify whether the first configuration information corresponding to the first object is abnormal due to deviation from these general practice rules, thereby improving the anomaly identification effect.

[0065] Based on the above, in one scenario, the configuration constraints mentioned above can be obtained by statistical analysis of multiple second configuration information. Different second configuration information corresponds to different second objects, so that the configuration constraints can more accurately represent the general practice rules corresponding to the types of these second objects. Thus, the configuration constraints can indicate some general practice rules that need to be followed when configuring objects of this type (such as the first object), so that the first configuration information corresponding to the first object can be better identified based on the configuration constraints to determine whether the first configuration information is abnormal due to deviation from these practice constraints, thereby improving the anomaly identification effect.

[0066] It should be noted that this specification does not limit the usage of the above configuration constraints. For example, it can be specifically as follows: after determining the configuration constraints corresponding to the first field from the second configuration information of some second objects, firstly, based on the first configuration information corresponding to the first object, determine the status information of the first object under the first field, such as whether a configuration value exists and what the configuration value is; then compare the status information with the configuration constraints to obtain the anomaly identification result of the first object under the first field, so that the anomaly identification result can indicate whether the status information deviates significantly from the general practice rules indicated by the configuration constraints, thereby enabling the anomaly identification result to indicate whether there is a configuration problem of the first object under the first field, so that the second information can be determined based on the anomaly identification result.

[0067] Based on the aforementioned second information, it can be understood that, in one scenario, the process of determining the second information may include: after obtaining the first configuration information corresponding to the first object, firstly, using an LLM to extract the attributes of the first object from the first configuration information, such as usage scenarios and usage scope; then, using the LLM to retrieve some second objects from a database containing configuration information of a large number of historical objects based on these attributes, so that the similarity between the attributes of the second objects and the attributes of the first object reaches a certain condition (such as exceeding a threshold); then, using the LLM to perform statistical analysis on the configuration information of these second objects (such as the aforementioned second configuration information) under the first field, in order to extract the configuration constraints satisfied by most of these second objects under the first field, such as 95% of similar objects setting parameters for each user. The first configuration information is constrained by a limit on the number of times it can be used. Then, the configuration value corresponding to the first configuration information in the first field is compared with the configuration constraint using LLM to obtain a comparison result. The comparison result indicates whether the first configuration information deviates significantly from the general practice rule indicated by the configuration constraint in the first field. When the comparison result indicates that the first configuration information deviates significantly from the general practice rule in the first field, second information is generated based on the first field. The second information indicates the anomaly of the first configuration information compared to the general practice rule, so that the anomaly information can be determined based on the second information. In this way, LLM can be used to realize the anomaly of the current configuration information compared to the historical configuration information of similar objects in terms of general practice rule, so as to better improve the anomaly identification effect.

[0068] It should be noted that this specification does not limit the method for extracting the "attributes of the first object" mentioned above. For example, when the aforementioned first configuration information includes page information of the aforementioned configuration page, if the language used by the page information is the same as the language that LLM can understand, then the method for extracting the "attributes of the first object" may include: using LLM to extract the attributes of the first object from the page information. As another example, when the aforementioned first configuration information includes configuration information obtained from a backend service, if the language used by the "configuration information obtained from the backend service" is different from the language that LLM can understand, then the method for extracting the "attributes of the first object" may include: first performing language conversion on the "configuration information obtained from the backend service" to obtain converted information; then using LLM to extract the attributes of the first object from the converted information. For example, when the first configuration information includes the page information of the configuration page and the configuration information obtained from the backend service, if the language used by the page information is the same as the language that the semantic detection model can understand, but the language used by the "configuration information obtained from the backend service" is different from the language that the semantic detection model can understand, then the method for extracting the "attributes of the first object" may include: using LLM to extract the attributes of the first object from the page information.

[0069] Based on the aforementioned abnormal information, it is understood that, in one scenario, when the abnormal information includes at least one of the first and second pieces of information, at least a portion of the abnormal information can be determined using a large language model. This allows the large language model to leverage its inherent capabilities (such as common sense and reasoning abilities) to discover deeper-level anomalies that cannot be covered by rule-based methods, such as logical errors, semantic conflicts, semantic contradictions, and non-compliance with general practical rules, thereby improving the accuracy of anomaly identification. Specifically, the large language model is configured to perform anomaly identification on the first configuration information based on the corresponding prompt words (e.g., identifying whether semantic anomalies exist, identifying whether configuration constraints are met). The prompt words are configured to guide the large language model on how to perform the anomaly identification. Furthermore, the prompt words are pre-defined based on the current application scenario to ensure that the prompt words meet the anomaly identification requirements of that application scenario.

[0070] In another scenario, the process of determining the aforementioned anomaly information may include: after obtaining the first and second information, firstly, using an LLM to merge the first and second information and remove duplicates to obtain information 1; then, using the LLM to call a knowledge base (such as a business specification knowledge base) to verify each anomaly recorded in information 1, in order to delete anomalies that do not require notification in the current application scenario (such as anomalies that are normal configurations in the current application scenario, or anomalies that are too superficial), to obtain information 2; then, inputting information 2 into the LLM so that the LLM can generate the associated information of each anomaly in information 2 according to a pre-defined format, such as the name of the anomaly, the problem description of the anomaly, and the potential impact of the anomaly, so that the associated information of these anomalies can be presented to the user in clear and easy-to-understand natural language, so that the user can make decisions based on this associated information (e.g., Figure 3 (The decision shown). Figure 3 This is an example diagram illustrating a configuration process provided in this manual.

[0071] It should be noted that the knowledge base shown above is configured to record exceptions that do not require notification to the user, such as configuration errors arising from adapting to certain configuration requirements of the first object, exceptions that need to be exempted based on historical experience, and exceptions that have no value for attention. Furthermore, this specification does not limit the method of acquiring this knowledge base; for example, it can be built by relevant personnel. Alternatively, the knowledge base can be updated based on exceptions generated during historical configurations that were ignored. Also, the knowledge base can be updated based on the configuration requirements of the first object.

[0072] Based on the aforementioned anomaly information, it can be understood that, in one scenario, this anomaly information is determined using at least the first and second methods. The first method aims to identify anomalies (such as illogical logical errors) from the semantics carried by the first configuration information, while the second method identifies anomalies deviating from the configuration constraints indicated by historical configuration information (such as the aforementioned second configuration information) from the first configuration information. Furthermore, because the first method can still demonstrate good anomaly identification performance even without historical configuration information, it can compensate for the limitations of the second method in terms of applicability. Also, because this historical configuration information records knowledge that cannot be obtained through the knowledge base corresponding to the first method (such as group patterns reflected by the configuration information of a large number of objects), the second method can discover anomalies that cannot be obtained through the first method based on this historical configuration information, thus compensating for the knowledge blind spots of the first method. Therefore, the first and second methods complement each other, jointly improving the anomaly identification effect (such as accuracy).

[0073] Furthermore, this specification does not limit the timing of the execution of step 202. For example, in one scenario, step 202 may specifically include: displaying an error message in response to detecting the above-mentioned save operation or the above-mentioned submit operation, so as to significantly advance the discovery of the error to the time of configuration saving or configuration submission, thereby enabling the error to be discovered as early as possible before the first configuration information takes effect, so as to ensure that the user can modify the error as early as possible, and avoid defects caused by the late discovery of the error.

[0074] As can be seen from the above paragraph, in one scenario, the display time of the above-mentioned abnormal information is earlier than the effective time of the above-mentioned first configuration information (such as the time of publication online), so as to discover the abnormality as early as possible before the first configuration information takes effect, so as to ensure that the user can modify the abnormality as early as possible, and avoid defects caused by the abnormality being discovered too late.

[0075] Based on the relevant content of steps 201 to 202 above, the exception information processing method provided in this specification includes: receiving a configuration operation, which indicates first configuration information corresponding to a first object (such as a new activity); and displaying exception information, which includes at least one of first information and second information. Specifically, since the first information indicates that the first configuration information has a semantic exception (for example, a new activity is applicable to a new user, but the restriction "the activity is only applicable to new users" is not configured, thus resulting in a semantic exception), the first information can indicate a semantic configuration error (such as a logical error, semantic conflict, etc.) in the first configuration information, thereby ensuring that the exception information including the first information at least indicates a semantic configuration error in the first configuration information. Furthermore, because the second information indicates that the first configuration information does not meet configuration constraints (such as the general practice of "configuring frequency limits when the discount exceeds a threshold"), and these constraints are determined based on the second configuration information corresponding to a second object similar to the first object (such as historical activities), the second information can indicate that the first configuration information deviates significantly from the configuration constraints determined from the configuration information of similar objects. Therefore, the anomaly information including the second information can at least indicate configuration errors in the first configuration information based on some general practices. Thus, the anomaly information can comprehensively indicate various anomalies in the first configuration information (such as semantic contradictions, non-compliance with historical patterns, etc.), enabling users to recognize these anomalies and avoid consequences caused by users' lack of awareness of these anomalies.

[0076] Furthermore, this specification does not limit the executing entity of the above-described exception information handling method. For example, this method can be applied to a terminal device (such as a configuration platform front-end). Alternatively, this method can also be implemented using a terminal device and a server (such as...). Figure 1 The data interaction process between the anomaly analysis service (as shown) is implemented. This server can be a standalone server, a cluster server, or a cloud server.

[0077] Research has revealed that, in one scenario, the semantic understanding of the first configuration information determined based on the aforementioned DOM is relatively difficult because the semantics of this first configuration information are presented according to the DOM structure, thus affecting the effectiveness of semantic understanding.

[0078] Based on the above research, in order to overcome the difficulties shown in the previous paragraph, when the first configuration information includes the page information (such as DOM) of the configuration page, the process of determining the above-mentioned abnormal information may include: firstly, performing semantic parsing on the first configuration information to obtain first semantic information (such as a semantic tree), so that the first semantic information can represent the semantics carried by the first configuration information in a way that the model can understand; then using LLM to perform anomaly recognition processing (such as semantic anomaly detection, configuration constraint-based detection, etc.) on the first semantic information to obtain abnormal information. This can overcome the defects caused by the semantics carried by the first configuration information being difficult to understand, thereby improving the anomaly recognition effect.

[0079] Based on the foregoing, in one scenario, at least some of the aforementioned anomaly information (such as the first information, the second information, etc.) can be determined based on the first semantic information. This first semantic information indicates the semantics carried by the first configuration information, which is generated based on the page information of the configuration page, and the configuration operation is received via the configuration page. Because the first semantic information can represent the semantics carried by the first configuration information in a way that the model can understand, the anomaly recognition based on the first semantic information will not lead to recognition errors due to a lack of understanding of the semantics. This overcomes the shortcomings caused by the difficulty in understanding the semantics carried by the first configuration information, thereby improving the anomaly recognition effect.

[0080] It should be noted that this specification does not limit the way the first semantic information is used. For example, in one case, both the first information and the second information are determined based on the first semantic information. As another example, in one case, the first information is determined based on the first semantic information, but the second information is determined based on the first configuration information.

[0081] Regarding the aforementioned first semantic information, this first semantic information can describe the semantics carried by the aforementioned first configuration information in a more easily understandable way; moreover, this specification does not limit the representation method of the first semantic information. For example, the first semantic information can be represented using any type of structured data, such as a semantic tree. Furthermore, this specification does not limit the determination method of the first semantic information. For example, it can be implemented using any method capable of converting the DOM into a semantic tree, such as by using a pre-built semantic parsing module with this conversion function. The semantic parsing module is configured to convert the DOM into a semantic tree to obtain a complete, structured semantic tree, so that the semantic tree can clearly represent the Chinese tags (such as field names), corresponding values ​​(such as field configuration values), and corresponding hierarchical relationships (such as the level of the field) of each configuration recorded in the aforementioned DOM in a tree structure.

[0082] As can be seen, in one scenario, the process of determining the semantic tree can include: first, based on the stability of the front-end rendering structure, using DOM selectors to locate the containers of Chinese tags (keys) and their corresponding configuration values ​​(values) from the page information of the configuration page, and pairing them based on adjacent node relationships to obtain a semantic tree. This semantic tree can represent each configuration item (such as field information) recorded in the page information in a tree structure. Then, the text recorded in the semantic tree is cleaned (e.g., decorative characters, redundant whitespace, etc.) to obtain clean Chinese phrases that can be used for semantic analysis. Second, when it is detected that the configuration item recorded in the semantic tree includes a reference to other configuration entities (such as a URL), the semantic tree is updated by executing a recursive collection process so that the updated semantic tree also records the information corresponding to the configuration entity obtained through this process (such as subpage information). In addition, this determination process adopts a breadth-and-depth hybrid traversal strategy, and uses built-in deduplication tables and maximum depth limits to prevent circular references and infinite recursion, ensuring the stability and efficiency of this determination process. Furthermore, browser sessions and processes are reused as much as possible throughout the determination process to reduce performance overhead. Additionally, resource consumption strategies (such as concurrency limits and timeout settings) and degradation strategies are implemented for this determination process. The degradation strategy refers to: when recoverable problems such as network fluctuations occur, automatically retrying to obtain information corresponding to the current layer; when irrecoverable structural changes occur, reverting to a successfully extracted layer, ensuring that the final semantic tree can at least represent some of the semantics carried by the page information, thus avoiding complete blocking of the entire process.

[0083] Research has revealed that, in one scenario, the configuration information obtained from a backend service may be in a language (such as English) that the model can understand (such as Chinese), making semantic understanding of this first configuration information quite difficult.

[0084] Based on the above research, in order to overcome the difficulties shown in the previous paragraph, when the first configuration information includes configuration information obtained from the backend service (such as English data), the process of determining the above-mentioned abnormal information may include: firstly, according to the pre-built mapping relationship, the first configuration information is converted into language to obtain converted information, so that the converted information can represent the semantics carried by the first configuration information in a language that the model can understand (such as Chinese); then, the converted information is semantically parsed to obtain first semantic information (such as a semantic tree), so that the first semantic information can represent the semantics carried by the first configuration information in a way that the model can understand; then, the first semantic information is anomaly identified using LLM (such as semantic anomaly detection, configuration constraint-based detection, etc.) to obtain abnormal information. This can overcome the defects caused by the difficulty in understanding the semantics carried by the first configuration information, thereby improving the anomaly identification effect.

[0085] It should be noted that the above mapping relationship is configured to record the correspondence between words in different languages, so that the configuration information retrieved from the backend service can be converted into other languages ​​in the future to meet the language requirements of LLM, overcome the semantic understanding difficulties caused by language differences, and better improve the anomaly recognition effect.

[0086] Based on the above eight paragraphs, it can be seen that in one scenario, when the aforementioned abnormal information includes first information, and the aforementioned first semantic information indicates the semantics carried by the aforementioned first configuration information, the first information can be obtained by using a semantic detection model (such as LLM) to perform semantic anomaly detection on the first semantic information. This allows the first information to indicate the semantic anomalies present in the first configuration information and the associated information of those semantic anomalies. Since the first semantic information can represent the semantics carried by the first configuration information in a way that the model can understand, the semantic detection model will not make identification errors due to its inability to understand the semantics when processing the first semantic information. This overcomes the defects caused by the difficulty in understanding the semantics carried by the first configuration information, thereby improving the anomaly recognition effect.

[0087] It should be noted that this specification does not limit the semantic anomaly detection described in the preceding paragraph. For example, in one scenario, this semantic anomaly detection can be implemented using knowledge bases within the current application domain (such as business validation knowledge bases, business specification knowledge bases, etc.). These knowledge bases are configured to record semantic constraints that need to be satisfied when configuring a specific object (such as the first object), such as the semantic constraint "when the first object is applicable to new users, the restriction 'the activity only applies to new users' needs to be configured." Furthermore, the current application domain refers to the domain to which the first object belongs.

[0088] It can be seen that, in one scenario, the aforementioned first information can be obtained by using a knowledge base to perform semantic anomaly detection on the first semantic information, so that the first information can indicate whether the first configuration information satisfies the semantic constraints recorded in the knowledge base, thereby enabling the first information to indicate whether there are semantic anomalies in the first configuration information.

[0089] In another scenario, if the aforementioned knowledge base is configured to record the semantic constraints that need to be satisfied when configuring objects of various types, the process of determining the aforementioned first information may include: firstly, searching the knowledge base for knowledge information (such as business specifications, business instructions, etc.) that matches the first object, so that the knowledge information can indicate the semantic constraints that need to be satisfied when configuring the first object; then comparing the aforementioned first semantic information with the knowledge information to obtain the first information, so that the first information can indicate whether the first configuration information satisfies the semantic constraints recorded in the knowledge information, thereby enabling the first information to indicate whether there are any semantic anomalies in the first configuration information. This specification does not limit the implementation method of this comparison; for example, it can be implemented using any machine learning model (such as LLM) capable of implementing this comparison function.

[0090] Based on the foregoing, in one scenario, when the first information in the aforementioned abnormal information indicates a semantic anomaly in the first configuration information, this semantic anomaly may include the first configuration information not satisfying semantic constraints. This first information may be obtained by analyzing the first semantic information and knowledge information using at least a first model (such as LLM), where the knowledge information includes the semantic constraints. The first semantic information indicates the semantics carried by the first configuration information, thereby enabling the first model to combine domain knowledge to detect whether the first configuration information has semantic anomalies, thus improving the anomaly identification effect. This specification does not limit this analysis; for example, it can be implemented using any method capable of identifying whether the first semantic information matches the knowledge information, such as comparison methods, matching methods, or alignment methods. Therefore, in one scenario, the first information may be obtained by comparing the first semantic information and knowledge information using at least a first model (such as LLM).

[0091] It should be noted that this specification does not limit the method of obtaining the above knowledge information. For example, it can be implemented using any method that can find content that matches an object of a certain type in a knowledge base, such as a machine learning model (such as LLM) with such search function.

[0092] In another scenario, the process of determining the aforementioned knowledge information may include: after obtaining the first semantic information based on the first configuration information corresponding to the first object, firstly, using LLM to parse the first semantic information to obtain first intent information, so that the first intent information can indicate the intent of the first object, such as the expected discount, the expected scope of use, the expected usage scenario, etc., thereby enabling the first intent information to indicate some characteristics (such as attributes) of the first object; then, searching the knowledge base for content that matches the first intent information as the knowledge information corresponding to the first object, so that the knowledge information can represent the semantic constraints that need to be met when configuring the first object as comprehensively and accurately as possible, so that subsequent detection of semantic anomalies in the first configuration information can be based on the knowledge information, thereby improving the anomaly identification effect.

[0093] Based on the above, in one scenario, the aforementioned knowledge information can be determined based on first intent information. This first intent information is obtained by performing intent recognition on the aforementioned first semantic information, so that the first intent information can indicate the intent (such as attributes) of the first object. This allows the knowledge information determined based on the first intent information to represent the semantic constraints that need to be met when configuring the first object as comprehensively and accurately as possible, so that subsequent detection of semantic anomalies in the first configuration information can be performed based on this knowledge information, thereby improving the anomaly recognition effect.

[0094] Research has revealed that, in one scenario, the aforementioned semantic anomaly detection can be configured to at least detect semantic conflicts between different fields in the first configuration information. For example, when the text field corresponds to the configuration value "exclusive to new users," if the configuration value corresponding to the actual rule field does not restrict the usage scope to new users, then it can be determined that there is a semantic conflict between the field information corresponding to the actual rule field and the field information corresponding to the text field.

[0095] Based on the above research, in one scenario, when the first configuration information corresponding to the first object includes multiple field information (such as <object name, activity 1>, <text description, exclusive for new users>, <actual rules, rule sequence>, etc.), and the first semantic information determined based on the first configuration information includes the semantic content corresponding to each field information, the process of determining the first information can include: firstly, using a second model (such as LLM) to compare different semantic content in the first semantic information to obtain a comparison result, so that the comparison result can indicate which semantic content among these semantic contents has semantic conflicts and the related information of the conflicts; in response to the comparison result indicating that at least two semantic contents have semantic conflicts, first information is generated based on the field information corresponding to each semantic content among the at least two semantic contents, so that the first information can at least indicate which field information in the first configuration information has semantic conflicts. In this way, the second model can be used to perform semantic consistency verification on different field information in the first configuration information, in order to identify some field information with semantic conflicts from the first configuration information, so as to overcome the defects caused by the existence of these semantic conflicts and improve the configuration effect.

[0096] Based on the above, in one scenario, when the first information in the aforementioned abnormal information indicates a semantic anomaly in the first configuration information, and this first configuration information includes multiple fields, and the aforementioned first semantic information includes the semantic content corresponding to each field, the semantic anomaly may include semantic conflicts between at least two fields. This first information can be obtained by analyzing the different semantic content within the first semantic information using at least a second model (such as LLM), so that the first information can at least indicate whether semantic conflicts exist between different semantic content within the first semantic information. This allows the first information to at least indicate whether semantic conflicts exist between the fields corresponding to each semantic content, and thus, at least indicate whether semantic conflicts exist between different fields within the first configuration information. In this way, the second model can be used to perform semantic consistency verification on the different fields within the first configuration information, aiming to identify some fields with semantic conflicts in the first configuration information, thereby overcoming the defects caused by these semantic conflicts and improving the configuration effect. This specification does not limit this analysis; for example, it can be implemented using any method capable of identifying whether semantic conflicts exist between different semantic contents, such as semantic comparison methods, semantic conflict identification methods, and semantic consistency evaluation methods. It can be seen that, in one case, the first information can be obtained by comparing different semantic contents in the first semantic information using at least a second model (such as LLM).

[0097] It should be noted that the second model described above is configured to perform semantic consistency verification on different semantic contents in the first semantic information. The aim is to identify semantically conflicting content from the first semantic information using this second model, so that subsequent semantic conflict detection results can be generated based on the field information corresponding to these semantic contents. This result should at least indicate the semantic conflict. Furthermore, first information should be generated based on the semantic conflict detection results, so that the first information includes the result. Additionally, the input data of the second model can include not only the first semantic information but also the aforementioned first intent information. This allows the second model to detect not only fields that semantically conflict with other field information but also fields that semantically conflict with the first intent information itself, improving the accuracy of field configuration error identification. It also enables the second model to more accurately determine the cause of these conflicts (such as due to a field configuration error), allowing for better overcoming of these conflicts. Moreover, this specification does not limit the second model; for example, the second model can be implemented using any machine learning model (such as LLM) capable of performing this semantic consistency verification function. The LLM is configured to perform the semantic consistency check under the guidance of its prompt words. The prompt words are configured to guide the LLM on how to perform the semantic consistency check, and the prompt words can be pre-set based on the application scenario.

[0098] Based on the aforementioned first information, it can be understood that, in one scenario, the process of obtaining the aforementioned abnormal information may include: firstly, obtaining the first configuration information corresponding to the first object, such as configuration information obtained from the backend service or DOM extracted from the frontend; then, generating first semantic information (such as a semantic tree) based on the first configuration information, so that the first semantic information can describe the semantics carried by the first configuration information in a way that the model can understand; then, parsing the intent of the first object from the first semantic information, such as the expected discount, expected scope of use, expected usage scenario, etc.; then, inputting the intent and the first semantic information into the first model (LLM), so that the first model calls the knowledge base to obtain knowledge information matching the intent, and the first model compares the first semantic information with the knowledge information to detect whether the first semantic information satisfies the semantic constraints (such as business logic, business specifications, business rules, etc.) indicated by the knowledge information, so as to obtain and output data 1, so that data 1 can indicate that the first configuration information in the knowledge information The system first identifies the anomalies and their associated information in the configuration information. It then inputs the intent and the first semantic information into a second model (LLM). The second model performs semantic consistency checks on different semantic contents within the first semantic information based on the intent, resulting in and outputting data 2. This data 2 indicates the semantic consistency anomalies in the first configuration information and their associated information. Finally, data 1 and data 2 are integrated to obtain the first information, which indicates semantic anomalies in the first configuration information (such as non-compliance with a business specification, semantic conflicts between field information 1 and field information 2, etc.) and their associated information. This allows the first information to identify syntactically correct but semantically problematic configuration errors in the first configuration information and their associated information. This enables subsequent identification of these anomalies based on the first information for user viewing and use. By leveraging the model, the system automatically identifies semantic configuration anomalies, improving anomaly identification and overall configuration effectiveness.

[0099] Based on the aforementioned abnormal information handling method, in one scenario, the method may include the following steps: When a user is setting first configuration information for a first object using a configuration page displayed by the configuration platform frontend, if the frontend detects a save or submit operation, the frontend asynchronously initiates an abnormal analysis request to FaaS, so that the request carries the object identifier of the first object; then, FaaS determines the associated information of the configuration page (such as URL, user's identity credentials, etc.) based on the object identifier carried in the request, and FaaS transmits a page collection request carrying the associated information to the headless browser module, so that the headless browser module can simulate the user's login and access the configuration page based on the associated information carried in the page collection request, in order to collect the page information (such as DOM) of the configuration page, so that FaaS can determine the first configuration information based on the page information fed back by the headless browser module, so that the first configuration... The information includes the page information; then the FaaS sends the first configuration information to the semantic parsing module, so that the semantic parsing module can convert the first configuration information into a semantic tree according to the recursive parsing method, so that the FaaS can determine the first semantic information based on the semantic tree fed back by the semantic parsing module, so that the first semantic information includes the semantic tree, thereby enabling the first semantic information to indicate the semantics carried by the first configuration information in a structured way and in a way that is easier for the model to understand; then, the FaaS sends the first semantic information to the anomaly analysis service, so that the anomaly analysis service can perform anomaly analysis (such as syntax error detection, semantic anomaly detection, configuration constraint comparison, etc.) based on the first semantic information to obtain anomaly information, so that the FaaS can subsequently send the anomaly information fed back by the anomaly analysis service back to the front end for display, so that the user can view and use it, prompting the user to decide whether to modify or adjust the first configuration information.

[0100] It should be noted that this specification does not limit the deployment relationship between the aforementioned FaaS, headless browser module, semantic parsing module, and anomaly analysis service and the configuration platform frontend. For example, in one scenario, the FaaS, headless browser module, semantic parsing module, and anomaly analysis service are all deployed within the configuration platform frontend. In another scenario, a portion of the FaaS, headless browser module, semantic parsing module, and anomaly analysis service are deployed within the configuration platform frontend, while the remaining portions are independent of the configuration platform frontend. Yet another scenario, the FaaS, headless browser module, semantic parsing module, and anomaly analysis service are all independent of the configuration platform frontend.

[0101] Based on the above two paragraphs, it can be seen that in one situation, such as Figure 3As shown, after the user completes the configuration process for object 1 (as mentioned above) through the configuration page, the configuration information corresponding to object 1 (as mentioned above) is first obtained through a certain method; then, the configuration information is structured and parsed to obtain a semantic tree (as mentioned above); then, anomaly analysis is performed based on the semantic tree. If no anomaly exists, a confirmation message is directly generated, and the configuration information is labeled as "configured reasonably" so that the configuration information can be published online later; however, if an anomaly exists, an anomaly prompt is generated for the configuration information, and an anomaly report is generated using a large language model, so that the anomaly report includes the anomaly information and the corresponding modification suggestions, thus making the anomaly report include the name of the anomaly, the problem description of the anomaly, and the potential impact of the anomaly. The system receives the response and corresponding modification suggestions for the anomaly, enabling the anomaly report to be notified to the user in at least one way (such as displaying it on the front end of the configuration platform or delivering it via instant messaging service). This allows the user to view the notification and subsequently decide whether to modify the configuration based on the anomaly report. If the decision indicates no configuration modification, the anomalies can be ignored, and the process can end directly. However, if the decision indicates configuration modification, the anomalies cannot be ignored. Therefore, after obtaining the configuration information modified by the user based on the anomaly report, the system returns to continue executing the aforementioned step of "structural parsing the configuration information to obtain a semantic tree" and its subsequent steps. This process is iterated and looped until it is determined that there are no anomalies in the configuration information or that the anomalies in the configuration information can be ignored, thereby improving the configuration effect.

[0102] As can be seen, in one scenario, let's assume the current application is configuring an activity offering a certain discount to new users. Based on this assumption, the user first sets initial configuration information for the activity using the configuration page displayed by the configuration platform's front-end. This initial configuration information should represent the characteristics of the activity, such as its name, scope of application, duration, and discount level. When the configuration platform's front-end detects that the user has triggered a save or submit operation, it asynchronously calls FaaS to obtain the DOM of the configuration page, which serves as the initial configuration information for the activity. Then, the configuration platform's front-end extracts key-value pairs (such as field information) from this initial configuration information to obtain the final configuration. A structured semantic tree is generated so that it can represent the semantics carried by the first configuration information in a way that the model can understand. Then, the configuration platform front-end executes the following steps by calling the anomaly analysis service: the semantic tree is input into LLM1, so that LLM1, guided by its corresponding prompt word 1 (such as prompt words including "Please determine whether the following configuration meets the business rule 'New customer activities must include new customer restrictions'"), compares the semantic tree with the knowledge information recorded in the knowledge base to identify whether there are any semantic anomalies (such as illogical logic) in the semantic tree to obtain the first information, and also uses the semantic tree as... The configuration constraints extracted from the configuration information of similar historical activities (such as the constraint "discount is less than 10%") are input into LLM2. LLM2, guided by its corresponding prompt word 2, compares the semantic tree with the configuration constraint to identify whether the corresponding content in the semantic tree (such as "discount is 20%") deviates from the configuration constraint to obtain second information. Next, the first information and the second information are fused (e.g., anomaly merging, duplicate removal, and deletion of unnecessary anomalies) to obtain a fusion result. Finally, the fusion result is input into LLM3 so that LLM3 can, under its corresponding prompt word 3, [further processing is required]. Under guidance, an anomaly report is generated so that it presents the anomalies in the first configuration information, their related information, and solutions in a structured form of a four-tuple: {anomaly name, anomaly problem description, anomaly potential impact, and anomaly modification suggestion}. This allows the anomaly report to be fed back to the user in a certain way (such as displaying the anomaly information on the front end of the configuration platform or sending it to the user via instant messaging service), enabling the user to make decisions based on the anomaly report to modify the first configuration information, ensuring that the modified first configuration information does not contain these anomalies, thereby improving the configuration effect.

[0103] Based on the above, in one scenario, the above-mentioned abnormal information processing method may include the following steps: receiving a configuration operation, which indicates first configuration information, and the configuration operation is received via a configuration page; obtaining the first configuration information, which includes page information (such as DOM) of the configuration page; generating first semantic information (such as a semantic tree) based on the first configuration information; inputting the first semantic information into a semantic detection model (such as a first model configured to identify whether the first semantic information satisfies semantic constraints, and / or a second model configured to identify whether different semantic contents in the first semantic information have semantic conflicts, etc.), and obtaining first information output by the semantic detection model, which indicates the first configuration information. The configuration information contains semantic anomalies, such as unreasonable logical errors or semantic conflicts. This first semantic information, along with the configuration constraints, is input into a constraint detection model (such as LLM) to obtain second information output by the model. This second information indicates that the first configuration information does not meet the configuration constraints, which are determined based on the second configuration information. The first object corresponding to the first configuration information is similar to the second object corresponding to the second configuration information. Based on the first and second information, anomaly information is obtained, indicating the presence of anomalies in the first configuration information and their associated information. This anomaly information is then displayed so that the user can decide whether to modify the first configuration information to improve configuration effectiveness. For details on the semantic detection model and the constraint detection model, please refer to the above.

[0104] Based on the above-mentioned abnormal information processing methods and related content Figure 4 This section explains and clarifies the relevant content of the abnormal information processing device. Among other things, Figure 4 This is a schematic diagram of an anomaly information processing device provided in this specification. It should be noted that for technical details of this anomaly information processing device, please refer to the relevant content of the anomaly information processing method above.

[0105] like Figure 4 As shown, the anomaly information processing device 400 includes: Receiving unit 401 is used to receive a configuration operation, wherein the configuration operation indicates first configuration information; Display unit 402 is used to display abnormal information, the abnormal information including at least one of first information and second information, the first information indicating that the first configuration information has a semantic abnormality, the second information indicating that the first configuration information does not meet the configuration constraints, the configuration constraints being determined based on the second configuration information, and the first object corresponding to the first configuration information being similar to the second object corresponding to the second configuration information.

[0106] In one scenario, the semantic anomaly includes the first configuration information not satisfying semantic constraints; the first information is obtained by analyzing the first semantic information and knowledge information using at least a first model, the knowledge information including the semantic constraints, and the first semantic information indicating the semantics carried by the first configuration information.

[0107] In one scenario, the knowledge information is determined based on first intent information, which is obtained by performing intent recognition on the first semantic information.

[0108] In one scenario, the first configuration information includes multiple field information; the first semantic information includes the semantic content corresponding to each of the field information; the semantic anomaly includes at least two of the field information having semantic conflicts; and the first information is obtained by analyzing different semantic content in the first semantic information using at least a second model.

[0109] In one scenario, the configuration operation is received via a configuration page; The anomaly information processing device 400 further includes: The processing unit includes obtaining the first configuration information, the first configuration information including page information of the configuration page; generating first semantic information based on the first configuration information; inputting the first semantic information into a semantic detection model to obtain the first information output by the semantic detection model; inputting the first semantic information and the configuration constraints into a constraint detection model to obtain the second information output by the constraint detection model; and obtaining the abnormal information based on the first information and the second information.

[0110] In one scenario, the anomaly information processing device 400 further includes one or more of the following: the display time corresponding to the anomaly information is earlier than the effective time corresponding to the first configuration information; a modification suggestion corresponding to the anomaly information is displayed; at least a portion of the information in the anomaly information is determined based on first semantic information, the first semantic information indicating the semantics carried by the first configuration information, the first semantic information being generated based on page information of the configuration page, and the configuration operation being received via the configuration page; at least a portion of the anomaly information is determined using a large language model; the effective time corresponding to the second configuration information is earlier than the effective time corresponding to the first configuration information; the configuration constraint is obtained by statistical analysis of multiple pieces of second configuration information, and the second object corresponding to different pieces of second configuration information is different.

[0111] Based on the aforementioned content regarding the anomaly information processing device 400, its working principle includes: receiving a configuration operation that indicates first configuration information corresponding to a first object (such as a new activity); and displaying anomaly information, which includes at least one of first information and second information. Specifically, because the first information indicates a semantic anomaly in the first configuration information (e.g., a new activity is applicable to a new user, but the restriction "the activity is only applicable to new users" is not configured, resulting in a semantic anomaly), the first information can indicate a semantic configuration error (such as a logical error, semantic conflict, etc.) in the first configuration information. Therefore, the anomaly information including the first information can at least indicate a semantic configuration error in the first configuration information. Furthermore, because the second information indicates that the first configuration information does not meet configuration constraints (such as the general practice of "configuring frequency limits when the discount exceeds a threshold"), and these constraints are determined based on the second configuration information corresponding to a second object similar to the first object (such as historical activities), the second information can indicate that the first configuration information deviates significantly from the configuration constraints determined from the configuration information of similar objects. Therefore, the anomaly information including the second information can at least indicate configuration errors in the first configuration information based on some general practices. Thus, the anomaly information can comprehensively indicate various anomalies in the first configuration information (such as semantic contradictions, non-compliance with historical patterns, etc.), enabling users to recognize these anomalies and avoid consequences caused by users' lack of awareness of these anomalies.

[0112] In addition, in some cases, an electronic device is provided, the device including a processor and a memory: the memory for storing instructions or computer programs; the processor for executing the instructions or computer programs in the memory to cause the electronic device to perform any of the above-described embodiments of the abnormal information processing method.

[0113] See Figure 5 The diagram illustrates a structural schematic of an electronic device 500 suitable for implementing the embodiments of the technical solution. The electronic device 500 may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital broadcast receivers, personal digital assistants (PDAs), tablets, portable media players (PMPs), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as television (TV) and desktop computers. Figure 5 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments involved in the technical solution.

[0114] like Figure 5 As shown, the electronic device 500 may include a processing unit (such as a central processing unit, graphics processor, etc.) 501, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 502 or a program loaded from storage device 508 into random access memory (RAM) 503. The RAM 503 also stores various programs and data required for the operation of the electronic device 500. The processing unit 501, ROM 502, and RAM 503 are interconnected via a bus 504. An input / output (I / O) interface 505 is also connected to the bus 504.

[0115] Typically, the following devices can be connected to I / O interface 505: input devices 506 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 507 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 508 including, for example, magnetic tapes, hard disks, etc.; and communication devices 509. Communication device 509 allows electronic device 500 to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 5 An electronic device 500 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.

[0116] Specifically, according to the embodiments of the technical solution, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments of the technical solution include a computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing any embodiment of the above-described exception information processing method. In such an embodiment, the computer program can be downloaded and installed from a network via a communication device 509, or installed from a storage device 508, or installed from a ROM 502. When the computer program is executed by the processing device 501, it performs the functions defined in the method of the embodiments of the technical solution.

[0117] The aforementioned electronic device and the aforementioned abnormal information processing method belong to the same inventive concept. For technical details not described in detail, please refer to the relevant content of the abnormal information processing method. Furthermore, the electronic device and the abnormal information processing method have the same beneficial effects.

[0118] In some cases, a computer-readable medium is provided that stores instructions or a computer program that, when executed on a device, causes the device to perform any of the embodiments of the above-described exception information processing method.

[0119] It should be noted that the aforementioned computer-readable medium can be a computer-readable signal medium, a computer-readable storage medium, or any combination thereof. A computer-readable storage medium can 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, RAM, ROM, an erasable programmable read-only memory (EPROM), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination thereof. In some cases, a computer-readable storage medium can 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 other cases, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium may 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, radio frequency (RF), or any suitable combination thereof.

[0120] In some cases, clients and servers can communicate using any currently known or future network protocol, such as Hypertext Transfer Protocol (HTTP), 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 (such as the Internet), and peer-to-peer networks, as well as any currently known or future networks.

[0121] 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.

[0122] The aforementioned computer-readable medium carries one or more programs, which, when executed by the electronic device, enable the electronic device to perform the aforementioned methods.

[0123] Computer program code for performing the operations involved in the above-described technical solutions can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include, but are not limited to, object-oriented programming languages—such as Java, Smalltalk, and C++—as well as 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 LANs or WANs—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0124] 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 the technical solutions. 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 the 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.

[0125] The units described in the technical solution can be implemented in software or hardware. The name of the unit / module does not, in some cases, constitute a limitation on the unit itself.

[0126] The functions described above 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 Array (FPGA), Application Specific Integrated Circuit (ASIC), Application Specific Standard Product (ASSP), System on Chip (SOC), Complex Programmable Logic Device (CPLD), etc.

[0127] In this specification, the various embodiments are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems or apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to in the method section.

[0128] It should be understood that "at least one (item)" refers to one or more, while "more than" refers to two or more. "And / or" describes the relationship between related objects, indicating that three relationships can exist. For example, "A and / or B" can represent three cases: only A exists, only B exists, and both A and B exist simultaneously, where A and B can be singular or plural. The character " / " generally indicates that the preceding and following related objects are in an "or" relationship. "At least one (item) of the following" or similar expressions refer to any combination of these items, including any combination of single or plural items. For example, at least one (item) of a, b, or c can represent: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, and c can be single or multiple.

[0129] The steps of the methods or algorithms described in conjunction with the embodiments can be implemented directly by hardware, a software module executed by a processor, or a combination of both. The software module can be located in RAM, memory, ROM, electrically programmable ROM, electrically erasable programmable ROM, registers, hard disk, removable disk, CD-ROM, or any other form of storage medium known in the art.

[0130] Based on the content recorded in this specification, those skilled in the art will be able to implement or use this application. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of this application. Therefore, the technical solutions will not be limited to the embodiments shown herein, but are to be accorded the widest scope consistent with the principles and novel features recorded in this specification.

Claims

1. An abnormal information processing method, comprising: Receive a configuration operation, wherein the configuration operation indicates first configuration information; Displaying abnormal information, the abnormal information including at least one of first information and second information, the first information indicating that the first configuration information has a semantic abnormality, the second information indicating that the first configuration information does not meet the configuration constraints, the configuration constraints being determined based on the second configuration information, and the first object corresponding to the first configuration information being similar to the second object corresponding to the second configuration information.

2. The method according to claim 1, wherein the semantic anomaly includes the first configuration information not satisfying semantic constraints; The first information is obtained by analyzing the first semantic information and knowledge information using at least the first model. The knowledge information includes the semantic constraints, and the first semantic information indicates the semantics carried by the first configuration information.

3. The method according to claim 2, wherein the knowledge information is determined based on first intent information, and the first intent information is obtained by performing intent recognition on the first semantic information.

4. The method according to claim 1, wherein the first configuration information includes multiple field information; The first semantic information includes the semantic content corresponding to each of the aforementioned field information; The semantic anomaly includes at least two of the field information having semantic conflicts; The first information is obtained by analyzing the different semantic contents in the first semantic information using at least the second model.

5. The method according to claim 1, wherein the configuration operation is received via a configuration page; The method further includes: Obtain the first configuration information, which includes the page information of the configuration page; Based on the first configuration information, first semantic information is generated; The first semantic information is input into the semantic detection model to obtain the first information output by the semantic detection model; The first semantic information is combined with the configuration constraint input constraint detection model to obtain the second information output by the constraint detection model; Based on the first information and the second information, the abnormal information is obtained.

6. The method according to any one of claims 1-5, wherein the method further comprises one or more of the following: The display time corresponding to the abnormal information is earlier than the effective time corresponding to the first configuration information; Display the suggested modifications corresponding to the anomaly information; At least a portion of the abnormal information is determined based on first semantic information, which indicates the semantics carried by the first configuration information. The first semantic information is generated based on the page information of the configuration page, and the configuration operation is received via the configuration page. At least part of the anomaly information was determined using a large language model; The effective time of the second configuration information is earlier than the effective time of the first configuration information; The configuration constraints are obtained by analyzing multiple sets of the second configuration information, and the second objects corresponding to different sets of the second configuration information are different.

7. An anomaly information processing device, comprising: A receiving unit is configured to receive a configuration operation, wherein the configuration operation indicates first configuration information; A display unit is used to display abnormal information, the abnormal information including at least one of first information and second information, the first information indicating that the first configuration information has a semantic abnormality, the second information indicating that the first configuration information does not meet the configuration constraints, the configuration constraints being determined based on the second configuration information, and the first object corresponding to the first configuration information being similar to the second object corresponding to the second configuration information.

8. An electronic device, the device comprising: Processor and memory; The memory is used to store instructions or computer programs; The processor is configured to execute the instructions or computer program in the memory to cause the electronic device to perform the method according to any one of claims 1-6.

9. A computer-readable medium storing instructions or a computer program that, when executed on a device, cause the device to perform the method of any one of claims 1-6.

10. A computer program product comprising a computer program carried on a non-transitory computer-readable medium, the computer program containing program code for performing the method of any one of claims 1-6.

Images