A method, system, device and medium for processing alarm information
By matching the alarm data and historical feature information and inputting the target model, the problem of low alarm handling efficiency and accuracy in the prior art is solved, and more efficient and accurate alarm processing is achieved.
Patent Information
- Application Number
- CN202410382432.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-03-29
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2044-03-29
AI Technical Summary
Existing security protection products have low efficiency and accuracy when processing alarm information. Especially in complex dynamic environments, the alarm handling information generated by the model is prone to false alarms, which makes operational personnel time-consuming and labor-consuming.
By obtaining the alarm data generated by the security detection module, matching it with the historical alarm feature information, obtaining the target historical alarm data and handling information, generating prompt information to send to the target model, and using the target model to generate more accurate alarm handling information.
It improves the efficiency and accuracy of security protection products in alarm handling, reduces the false alarm rate, and improves the work efficiency of operation personnel.
Smart Images

Figure CN118337422B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of computer technology, and in particular to an alarm information processing method, system, electronic device and computer-readable storage medium. Background Art
[0002] With the continuous development of computer technology, security protection products for security detection have emerged. Security protection products can detect computing devices such as computers and hosts or virtual operating environments such as containers to ensure operational safety. In actual applications, security protection products can include cloud workload protection platform (CWPP), host-based intrusion detection system (HIDS), endpoint detection and response (EDR), high container security platform (CSP), etc.
[0003] When a security protection product detects an alarm event, it will generate corresponding alarm information. Users (such as security operators) can analyze and judge the alarm information to form alarm handling information, thereby completing the alarm handling.
[0004] Considering the large number of alarm messages generated by security protection products, it is time-consuming and labor-intensive for security operators to handle alarms, which is inefficient. In related technologies, alarm messages can be analyzed with the help of models (for example, language models with natural language processing capabilities).
[0005] However, since security protection products can detect complex and dynamic operating environments, the accuracy of using models for alarm handling is relatively low. Summary of the invention
[0006] The present application provides a method for processing alarm information. The method can improve the operating efficiency and accuracy of security protection products when handling alarms. The present application also provides a system, electronic device, computer-readable storage medium, and computer program product corresponding to the above method.
[0007] In a first aspect, the present application provides a method for processing alarm information, the method comprising:
[0008] Acquire first alarm data generated by the security detection module;
[0009] Matching the first alarm data with at least one piece of alarm characteristic information; wherein one piece of alarm characteristic information corresponds to at least one piece of historical alarm data;
[0010] In response to the first alarm data hitting the target alarm characteristic information in the at least one alarm characteristic information, acquiring at least one target historical alarm data corresponding to the target alarm characteristic information and alarm handling information of the at least one target historical alarm data;
[0011] A first prompt message is generated based at least on the first alarm data, the at least one target historical alarm data and the alarm handling information of the at least one target historical alarm data, the first prompt message is sent to the target model, and the first alarm handling information for the first alarm data returned by the target model is received, wherein the first prompt message is used to instruct the target model to perform alarm handling for the first alarm data.
[0012] In a second aspect, the present application provides an alarm information processing system, the system comprising:
[0013] A first acquisition module, used to acquire first alarm data generated by the security detection module;
[0014] A matching module, used to match the first alarm data with at least one piece of alarm characteristic information; wherein one piece of alarm characteristic information corresponds to at least one piece of historical alarm data;
[0015] A second acquisition module is used for acquiring at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data in response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information;
[0016] A handling module is used to generate a first prompt information based at least on the first alarm data, the at least one target historical alarm data and the alarm handling information of the at least one target historical alarm data, send the first prompt information to the target model, and receive the first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling for the first alarm data.
[0017] In a third aspect, the present application provides an electronic device, the electronic device comprising a processor and a memory. The processor and the memory communicate with each other. The processor is used to execute instructions stored in the memory so that the electronic device executes the alarm information processing method in the first aspect or any implementation of the first aspect.
[0018] In a fourth aspect, the present application provides a computer-readable storage medium, wherein the computer-readable storage medium stores instructions, wherein the instructions instruct an electronic device to execute the alarm information processing method described in the first aspect or any one of the implementations of the first aspect.
[0019] In a fifth aspect, the present application provides a computer program product comprising instructions, which, when executed on an electronic device, enables the electronic device to execute the alarm information processing method described in the first aspect or any one of the implementations of the first aspect.
[0020] Based on the implementations provided in the above aspects, this application can also be further combined to provide more implementations.
[0021] It can be seen from the above technical solutions that this application has the following advantages:
[0022] The present application provides an alarm information processing method, which first obtains first alarm data generated by a security detection module, and then matches the first alarm data with at least one alarm feature information, wherein one alarm feature information corresponds to at least one historical alarm data, and in response to the first alarm data hitting target alarm feature information in at least one alarm feature information, obtains at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of at least one target historical alarm data, and then, generates first prompt information based on at least the first alarm data, at least one target historical alarm data and the alarm handling information of at least one target historical alarm data, sends the first prompt information to a target model, and receives first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling on the first alarm data.
[0023] In the process of handling alarm data with the help of the target model, the method matches the alarm data to be handled with the alarm feature information. If the alarm data to be handled hits the target alarm feature information, since the target alarm feature information corresponds to the target historical alarm data, and the target historical alarm data has been handled by alarm, by inputting the target historical alarm data and the alarm handling information for the target historical alarm data into the target model, the target model can generate more accurate handling information for the current alarm data to be handled with the help of information related to the target historical alarm data. When this method is applied to security protection products, the security protection products can handle alarms accurately, effectively improving the operating efficiency and accuracy of the security protection products. BRIEF DESCRIPTION OF THE DRAWINGS
[0024] In order to more clearly illustrate the technical method of the embodiments of the present application, the drawings required for use in the embodiments are briefly introduced below.
[0025] Figure 1 A flowchart of a method for processing alarm information provided by an embodiment of the present application;
[0026] Figure 2 A flowchart of another method for processing alarm information provided in an embodiment of the present application;
[0027] Figure 3 A schematic diagram of the structure of an alarm information processing system provided in an embodiment of the present application;
[0028] Figure 4 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION
[0029] The terms "first" and "second" in the embodiments of the present application are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features.
[0030] First, some technical terms involved in the embodiments of the present application are introduced.
[0031] With the continuous development of computer technology, security protection products for security detection and ensuring the safe operation of computing devices such as computers and hosts or virtual operating environments such as containers have emerged. Security protection products can perform multi-faceted security detection for a variety of operating scenarios. For example, the security protection product can be a cloud workload protection platform (CWPP), which can detect host security and network security. For another example, the security protection product can be a host-based intrusion detection system (HIDS), which can perform security detection on the behavior and status of the computer system. For another example, the security protection product can be an endpoint detection and response (EDR), which can perform security detection on the system-level behavior of the endpoint. For another example, the security protection product can be a container security platform (CSP), which can provide multi-faceted security detection for containers.
[0032] When a security protection product detects an alarm event, it will generate corresponding alarm information. Among them, the alarm information may include the alarm data that generated the alarm event. Users (such as security operators) can analyze and judge the alarm data to form alarm handling information, thereby completing the alarm handling. For example, the security operator can determine whether the alarm event is a false alarm based on the alarm data. If it is not a false alarm, the security operator can also provide a solution for handling the alarm event.
[0033] Since security protection products can generate a large amount of alarm data, security operations personnel need to spend a lot of time to screen the alarm data. For example, security operations personnel need to spend a lot of time to verify the alarm data and record the handling information of each alarm data, which is time-consuming and labor-intensive.
[0034] In the related art, the alarm data can be analyzed with the help of a model (for example, a language model with natural language processing capabilities). Specifically, by sending the alarm data to the model, the model can identify and analyze the alarm data and give a result on whether the alarm event is a false alarm.
[0035] However, since security protection products can detect complex and dynamic operating environments, and the model will inevitably have the "hallucination problem", that is, the model may give some seemingly reasonable but actually wrong output results, resulting in low accuracy of the alarm handling information generated by the model.
[0036] In view of this, the present application provides a method for processing alarm information. The method first obtains the first alarm data generated by the security detection module, then matches the first alarm data with at least one alarm feature information, wherein one alarm feature information corresponds to at least one historical alarm data, and in response to the first alarm data hitting the target alarm feature information in at least one alarm feature information, obtains at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of at least one target historical alarm data, and then, generates first prompt information based on at least the first alarm data, at least one target historical alarm data and alarm handling information of at least one target historical alarm data, and sends the first prompt information to the target model, and receives the first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling for the first alarm data.
[0037] In the process of handling alarm data with the help of the target model, the method matches the alarm data to be handled with the alarm feature information. If the alarm data to be handled hits the target alarm feature information, since the target alarm feature information corresponds to the target historical alarm data, and the target historical alarm data has been handled by alarm, by inputting the target historical alarm data and the alarm handling information for the target historical alarm data into the target model, the target model can generate more accurate handling information for the current alarm data to be handled with the help of information related to the target historical alarm data. When this method is applied to security protection products, the security protection products can handle alarms accurately, effectively improving the operating efficiency and accuracy of the security protection products.
[0038] To facilitate understanding of the technical solution provided by the embodiments of the present application, it will be described below with reference to the accompanying drawings.
[0039] See also Figure 1 The flowchart of a method for processing alarm information provided by an embodiment of the present application is shown, and the method specifically includes:
[0040] S101: Acquire first alarm data generated by a security detection module.
[0041] The first alarm data may be understood as the alarm data to be processed currently, and the first alarm data may lead to the occurrence of an alarm event, wherein the alarm event may be understood as an event that may generate an operational risk.
[0042] The first alarm data may be composed of multiple alarm fields. For example, the alarm data may include a command line field, a parent process command line field, a process group command line field, a process tree information field, a runtime link field, an execution directory field, etc.
[0043] In an embodiment of the present application, the first alarm data may be generated by a security detection module. Specifically, the security detection module may be a module in a security protection product, such as a module in a security protection product such as CWPP, HIDS, EDR, CSP, etc. The security detection module may perform security detection on the operation data generated during the operation of a computing device or a virtual operation environment, such as performing security detection on operation data such as operation logs and network traffic. When the security detection module detects that there is abnormal data in the operation data, and the abnormal data may cause an alarm event to occur, the security detection module may determine the abnormal data as the first alarm data. In this way, after the security detection module detects the first alarm data, the server may obtain the first alarm data.
[0044] S102: Match the first alarm data with at least one piece of alarm characteristic information.
[0045] Alarm characteristic information may also be referred to as alarm pattern information. In the embodiment of the present application, one alarm characteristic information may correspond to at least one historical alarm data. In other words, the alarm characteristic information may be used to characterize a certain characteristic of at least one historical alarm data corresponding thereto, or to characterize a certain pattern that at least one historical alarm data corresponding thereto conforms to. The historical alarm data may be alarm data generated within a historical time period.
[0046] That is, the alarm characteristic information may be information common to a certain characteristic or a certain mode extracted from at least one historical alarm data corresponding thereto. As an example, at least one or more historical alarm data may be https: / / dl.influxdata.com / influxdb / releases / influ xdb2-2.0.7-amd64.deb and https: / / dl.influxdata.com / influxdb / releases / influxdb-1.8.9.x86_64.dev, and the extracted alarm characteristic information may be https: / / dl.influxdata.com / influxdb / releases / influxdb-.*.
[0047] In the embodiment of the present application, in order to improve the effect of using the alarm feature information to assist the target model in the subsequent process, at least one historical alarm data corresponding to the alarm feature information may be processed by the alarm. For example, the historical alarm data may be manually processed by a user (e.g., a security operator), or may be automatically processed by the target model.
[0048] At least one piece of alarm feature information may be stored in a feature information library. Specifically, the feature information library may store, for each piece of alarm feature information, a correspondence between the alarm feature information and at least one piece of historical alarm data used to generate the alarm feature information, and a correspondence between the alarm feature information and the alarm handling information of at least one piece of historical alarm data. In this way, the corresponding at least one piece of historical alarm data and the alarm handling information of at least one piece of historical alarm data may be queried through the alarm feature information.
[0049] The alarm handling information may be understood as information used to handle alarm data. Specifically, the alarm handling information may include at least one of alarm tag information and alarm attribution information.
[0050] The alarm tag information can be used to indicate the handling label of the alarm data, that is, the handling result of the alarm data can be obtained through the alarm tag information. For example, the alarm tag information can be a false alarm, a real intrusion, a violation, etc.
[0051] The alarm attribution information can be used to characterize the reason for generating the disposal tag, that is, the reason for the disposal result of the alarm data can be obtained through the alarm attribution information. For example, when the alarm tag information is "false alarm", the alarm attribution information can be "the downloaded file is the installation package of influxdb, does not contain other abnormal behaviors, and no virus detection related alarms are triggered. Therefore, it is determined that the download of the database installation package is a normal business behavior and is a false alarm."
[0052] After obtaining the first alarm data, the server can match the first alarm data with at least one alarm feature information in the feature information library, thereby determining whether the first alarm data to be processed currently has certain features of the historical alarm data, or whether it has a certain pattern that the historical alarm data conforms to.
[0053] S103: In response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information, acquiring at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data.
[0054] In specific implementation, the first alarm data can be matched with at least one alarm characteristic information in different ways. In some embodiments, the server can match the first alarm data with a regular expression corresponding to at least one alarm characteristic information. In other words, the alarm characteristic information can be stored in the form of a regular expression, and the first alarm data can be matched with the alarm characteristic information through regular matching rules.
[0055] In this case, the server can obtain at least one target historical alarm data corresponding to the target alarm characteristic information and alarm handling information of at least one target historical alarm data in response to the presence of an alarm field in the first alarm data matching a regular expression corresponding to the target alarm characteristic information in at least one alarm characteristic information.
[0056] It is understandable that, since the alarm data may include multiple alarm fields, in order to improve matching efficiency and matching accuracy, the alarm feature information may be information extracted from an alarm field in at least one historical alarm data. In this way, by matching each alarm field of the first alarm data with the alarm feature information, at least one target historical alarm data and alarm handling information of at least one target historical alarm data can be obtained.
[0057] In other embodiments, the server may calculate the text similarity with at least one piece of alarm characteristic information for each alarm field of the first alarm data. That is, when the alarm characteristic information is not stored in the form of a regular expression, the first alarm data and the alarm characteristic information are matched by similarity calculation.
[0058] In this case, the server can obtain at least one target historical alarm data and alarm handling information of at least one target historical alarm data corresponding to the target alarm feature information in response to the existence of target alarm feature information whose text similarity satisfies the first set condition. For example, the first set condition can be that the text similarity is greater than the similarity threshold. In this way, by matching each alarm field of the first alarm data with the alarm feature information, at least one target historical alarm data and alarm handling information of at least one target historical alarm data can be obtained.
[0059] Since the first alarm data hits the target alarm feature information, and the target alarm feature information corresponds to at least one target historical alarm data, the first alarm data is similar to the at least one target historical alarm data.
[0060] S104: Generate first prompt information based on at least the first alarm data, at least one target historical alarm data and alarm handling information of at least one target historical alarm data, send the first prompt information to the target model, and receive first alarm handling information for the first alarm data returned by the target model.
[0061] In an embodiment of the present application, the first alarm data is processed by means of a target model, thereby realizing automatic generation of alarm processing information. The target model may be a language model with natural language processing capabilities, and the target model may be able to perform different types of natural language processing tasks. For example, the target model may be a deep learning model trained based on text data.
[0062] In specific implementation, the target model can be used to generate alarm handling information based on the prompt learning technology. The target model is instructed to generate the alarm handling information by generating the first prompt information and utilizing the prompt capability of the first prompt information. The first prompt information can be used to instruct the target model to perform alarm handling for the first alarm data.
[0063] In other words, the first prompt information can inform the target model that the first alarm data is similar to at least one target historical alarm data, and instruct the target model to combine at least one target historical alarm data and at least one historical alarm handling information of the target historical alarm data to perform alarm handling on the first alarm data and generate first alarm handling information.
[0064] For example, the first prompt information may be "The current alarm data is XXX, and the historical alarm situations that meet the same alarm characteristic information as the current alarm data are as follows: the historical alarm data is XXX, of which X alarm mark information are false alarms, and X alarm mark information are real intrusions, and the alarm attribution information are XXX respectively. Please give the alarm mark information and alarm attribution information of the current alarm data based on the historical alarm data, the alarm mark information of the historical alarm data, and the alarm attribution information of the historical alarm data."
[0065] In this way, when generating the first alarm handling information, the target model can refer to the alarm handling information of the target historical alarm data similar to the first alarm data, and thus generate the first alarm handling information with higher accuracy.
[0066] The foregoing describes the case where the first alarm data hits the target alarm feature information in at least one alarm feature information. Next, the following describes the case where the first alarm data does not hit at least one alarm feature information.
[0067] In specific implementation, in response to the first alarm data not matching at least one alarm feature information, the server may send the second prompt information generated at least based on the first alarm data to the target model multiple times, and receive multiple second alarm handling information for the first alarm data returned by the target model.
[0068] The fact that the first alarm data does not match at least one alarm feature information indicates that there is no alarm feature information matching the feature or pattern of the first alarm data in the feature information library. At this time, the target model cannot analyze the first alarm data with the help of the target historical alarm data. In order to avoid the target model generating second alarm handling information with poor accuracy in a single time, the server can use the target model to generate multiple second alarm handling information multiple times.
[0069] It can be understood that the target model may have parameters for controlling the randomness of the model. When the parameters for controlling the randomness of the model are not 0, the same second prompt information is sent to the target model multiple times, and the target model can generate multiple different outputs, namely the multiple second alarm handling information mentioned above.
[0070] In some possible implementations, the server may first obtain auxiliary information, then generate second prompt information based at least on the first alarm data and the auxiliary information, and send the second prompt information to the target model multiple times, and receive multiple second alarm handling information for the first alarm data returned by the target model.
[0071] The auxiliary information can be used to assist the target model in generating the second alarm handling information. In the embodiment of the present application, the auxiliary information can include at least one of the following: first information obtained by querying the first alarm data as an index, and second information related to the alarm event that generates the first alarm data.
[0072] For example, the server can use the first alarm data as an index to query the alarm knowledge base to obtain first information related to the first alarm data, wherein the alarm knowledge base can store information related to security operations. For another example, the server can query the alarm database to obtain second information related to the alarm event that generates the first alarm data, such as alarm rule information, alarm context information, related asset information, vulnerability information, etc., wherein the alarm database can store information related to the alarm event.
[0073] The second prompt information can be used to instruct the target model to perform an alarm treatment on the first alarm data. In other words, the second prompt information can inform the target model that the auxiliary information is related to the first alarm data, and instruct the target model to perform an alarm treatment on the first alarm data in combination with the auxiliary information to generate second alarm treatment information.
[0074] For example, the second prompt information can be "The current alarm data is XXX, the alarm context information is XXX, the alarm rule information is XXX, the related assets and vulnerability information is XXX, it cannot be determined as a false alarm under XXX circumstances, please provide the alarm mark information and alarm attribution information of the current alarm data."
[0075] In this way, by obtaining auxiliary information related to the first alarm data and inputting the second prompt information into the target model multiple times, the accuracy of the second alarm data generated when the first alarm data does not hit the alarm feature information can be improved to a certain extent.
[0076] Based on the above description, an embodiment of the present application provides an alarm information processing method. The method first obtains the first alarm data generated by the security detection module, then matches the first alarm data with at least one alarm feature information, wherein one alarm feature information corresponds to at least one historical alarm data, and in response to the first alarm data hitting the target alarm feature information in at least one alarm feature information, obtains at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of at least one target historical alarm data, and then, generates a first prompt information based on at least the first alarm data, at least one target historical alarm data and the alarm handling information of at least one target historical alarm data, and sends the first prompt information to the target model, and receives the first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling for the first alarm data.
[0077] In the process of handling alarm data with the help of the target model, the method matches the alarm data to be handled with the alarm feature information. If the alarm data to be handled hits the target alarm feature information, since the target alarm feature information corresponds to the target historical alarm data, and the target historical alarm data has been handled by alarm, by inputting the target historical alarm data and the alarm handling information for the target historical alarm data into the target model, the target model can generate more accurate handling information for the current alarm data to be handled with the help of information related to the target historical alarm data. When this method is applied to security protection products, the security protection products can handle alarms accurately, effectively improving the operating efficiency and accuracy of the security protection products.
[0078] In the foregoing, by matching the first alarm data with the alarm feature information, the target model can refer to the alarm handling information of the target historical alarm data to generate the first alarm handling information for the first alarm data. In an embodiment of the present application, the user is supported to provide feedback on the alarm handling information generated by the target model, and then the target model is optimized using the feedback information.
[0079] In specific implementation, the server may receive feedback information for the first alarm handling information, and adjust the parameters of the target model according to the feedback information, wherein the feedback information represents the judgment result of the user for the first alarm handling information.
[0080] That is to say, during the user's alarm handling process, when the first alarm data hits the alarm feature information, after the first alarm handling information for the first alarm data is automatically generated with the help of the target model, the user can provide feedback on the first alarm handling information. For example, the user can provide feedback on whether the first alarm handling information generated by the target model is correct. If it is incorrect, the user can further provide feedback on the correct alarm handling information.
[0081] In a specific implementation, the server may provide a first control and a second control, wherein the first control indicates that the first alarm handling information is correct, and the second control indicates that the first alarm handling information is incorrect. In other words, after the target model generates the first alarm handling information for the first alarm data, the first alarm handling information may be presented to the user, and a control may be provided to receive user feedback indicating whether the first alarm handling information is correct.
[0082] For example, when the alarm mark information in the first alarm handling information is "real intrusion", but the user judges that the actual alarm mark information of the first alarm data should be "false alarm", the second control can be triggered to feedback to the server that the target model has generated incorrect alarm handling information.
[0083] In response to the triggering operation on the second control, the server may receive first modification information for the first alarm handling information, and then determine feedback information according to the first modification information.
[0084] That is to say, when the user feedback target model generates incorrect first alarm handling information, the user can correct the incorrect first alarm handling information. At this time, the feedback information can be information indicating that the first alarm handling information is incorrect, and first modification information for the first alarm handling information. In addition, the server can also update the incorrect first alarm handling information according to the first modification information, for example, by adding the first modification information to the first alarm handling information.
[0085] It can be understood that if the user triggers the first control, it indicates that the first alarm handling information generated by the target model fed back to the server is correct. At this time, the server can store the first alarm handling information and the user does not need to perform subsequent operations.
[0086] For the case where the first alarm data does not match at least one piece of alarm feature information, the embodiment of the present application also supports the user to provide feedback on multiple pieces of second alarm handling information generated by the target model, and then optimize the target model using the feedback information.
[0087] In a specific implementation, the server may receive feedback information for multiple pieces of second alarm handling information, and adjust the parameters of the target model according to the feedback information, wherein the feedback information represents the judgment result of the user for the second alarm handling information.
[0088] In some possible implementations, the server may receive sorting information for the second alarm handling information that is confirmed to be correct among the multiple second alarm handling information, and / or receive information for at least part of the multiple second alarm handling information that is confirmed to be incorrect. The sorting information may represent the user's preference for the second alarm handling information that is confirmed to be correct among the multiple second alarm handling information. In this way, the feedback information may include the sorting information and / or the information that is confirmed to be incorrect.
[0089] That is to say, for the multiple second alarm handling information generated by the target model, since the target model cannot generate alarm handling information with the help of the target historical alarm data and the alarm handling information of the target historical alarm data, the accuracy of the second alarm handling information may be lower than that of the first alarm data hitting the target alarm feature information in the alarm feature information. Therefore, the user can first feedback whether the second alarm handling information is correct, and further, the user can sort the correct second alarm handling information and modify the incorrect second alarm handling information.
[0090] In specific implementation, the server may provide a third control and a fourth control for each second alarm handling information, wherein the third control indicates that the second alarm handling information is correct, and the fourth control indicates that the second alarm handling information is incorrect. In other words, after the target model generates multiple second alarm handling information for the first alarm data, multiple second alarm handling information may be presented to the user (for example, in the form of a list), and a control may be provided to receive user feedback indicating whether the multiple second alarm handling information is correct.
[0091] For at least one second alarm handling information that triggers the third control, it indicates that the at least one second alarm handling information generated by the target model fed back by the user to the server is correct. At this time, the user can further sort the at least one correct second alarm handling information, and sort the second alarm handling information that is most consistent with the facts and has the most complete interpretation in front, thereby forming sorting information.
[0092] The embodiment of the present application does not limit the way in which the user sorts the at least one correct second alarm handling information. For example, the user can sort the at least one correct second alarm handling information by dragging the second alarm handling information.
[0093] At least one second alarm handling information that triggers the fourth control indicates that the user feedback to the server that at least one second alarm handling information generated by the target model is incorrect. At this time, the user can correct the at least one incorrect second alarm handling information to form second modification information.
[0094] Then, the server may receive the sorting information and / or the second modification information, and then determine the feedback information. In addition, the server may also update the at least one incorrect second alarm handling information according to the second modification information, for example, by adding the second modification information to the second alarm handling information.
[0095] By receiving feedback information during the user's alarm handling process, the alarm handling can be completed with the help of the target model without the user having to do additional work, and the server can perform human feedback reinforcement learning (RLHF) on the target model based on the feedback information.
[0096] In this way, RLHF can be performed on the target model throughout the entire life cycle of the target model. Compared with the traditional method of performing RLHF only in the pre-training stage of the target model, it can improve the reasoning ability of the target model to generate alarm handling information in a long-term and continuous manner, and enhance the reasoning stability of the target model.
[0097] In specific implementation, the server can use the feedback information as training data to train the reward model, where the reward model can be used to measure the inference accuracy of the target model.
[0098] In an embodiment of the present application, in combination with a scenario in which the target model is used to generate alarm handling information, the reward model may include at least one of the following: a classification model for determining whether the alarm handling information is correct, and a sorting model for sorting multiple alarm handling information.
[0099] It can be understood that since the feedback information carries the user's judgment on whether the alarm handling information generated by the target model is correct, the user's modification of incorrect alarm handling information generated by the target model, and the sorting of correct alarm handling information generated by the target model for the alarm data of missed alarm feature information, using the feedback information as training data can enable the trained reward model to have the ability to judge whether the alarm handling information is correct, or the ability to sort multiple alarm handling information according to user preferences.
[0100] The trained reward model can be used to adjust the parameters of the target model. For example, the server can use the proximal policy optimization (PPO) algorithm to perform reinforcement learning on the target model, and then adjust the parameters of the target model to improve the inference accuracy of the target model.
[0101] In some embodiments, the server can also generate alarm feature information with the help of feedback information. Taking the example that at least one existing alarm feature information is stored in the feature information database, the generation process of the alarm feature information is described.
[0102] The server may first obtain multiple pieces of historical alarm data, wherein the multiple pieces of historical alarm data do not match at least one piece of existing alarm feature information.
[0103] It is understandable that in order to enrich the amount of alarm feature information in the feature information database, the server may generate new alarm feature information by using multiple pieces of historical alarm data that do not match the existing alarm feature information.
[0104] Next, the server may determine multiple pieces of historical alarm handling information for the multiple pieces of historical alarm data, wherein the multiple pieces of historical alarm handling information include at least one of the following: third alarm handling information for the multiple pieces of historical alarm data returned by the target model, and fourth alarm handling information after updating the third alarm handling information according to feedback information of the third alarm handling information.
[0105] That is to say, the server can continuously obtain the alarm handling information generated by the target model and the feedback information of the user with respect to the alarm handling information generated by the target model, and then determine multiple pieces of historical alarm handling information with respect to multiple pieces of historical alarm data.
[0106] Furthermore, the server can classify multiple historical alarm data according to at least one alarm event that generates multiple historical alarm data, obtain at least one category of historical alarm data set, and cluster the historical alarm data set of each category according to at least one alarm field to obtain clustering results of the historical alarm data set of each category.
[0107] In other words, the server can perform group processing on multiple historical alarm data, group the historical alarm data under the same alarm event into one group, and perform cluster processing on different fields of the historical alarm data in each group to obtain a clustering result.
[0108] In response to the existence of a target clustering result in the clustering results satisfying the second set condition, the server may generate second alarm feature information according to the target clustering result, and store the generated alarm feature information in the feature information library.
[0109] In some possible implementations, the server can determine the number of samples in each cluster in the clustering result for each clustering result, and in response to the presence of a cluster in the target clustering result whose number of samples is greater than a set number threshold, generate alarm feature information based on the target clustering result, and store the generated alarm feature information in a feature information library.
[0110] Specifically, the server can generate a third prompt information based on at least one historical alarm data included in the target clustering result, send the third prompt information to the target model, and receive the alarm feature information returned by the target model. The third prompt information can be used to instruct the target model to generate alarm feature information that matches the at least one historical alarm data included in the target clustering result.
[0111] For example, the third prompt information may be "Similar content appears XXX times in the XX alarm field in the historical alarm data, which are XXX respectively. Please construct a feature pattern / regular expression from it, which is required to match the similar content appearing in the above XX alarm field. It must have enough features to avoid too many false matches causing alarm omissions." In this way, with the help of the target model, new alarm feature information is automatically extracted.
[0112] Next, the above alarm information processing method will be used as an example to illustrate the application of the alarm information processing method to security protection products. Figure 2The flowchart of another alarm information processing method shown is as follows. In this case, the server can provide a server for security protection products such as CWPP, HIDS, EDR, and CSP.
[0113] The security protection product has a security detection module. When the user uses the security protection product, the security detection module can detect the operating environment. When an alarm event is detected, the first alarm data that generates the alarm event is obtained.
[0114] By matching the first alarm data with at least one alarm feature information, when the match is successful, the target historical alarm data corresponding to the hit target alarm feature information and the alarm handling information of the target historical alarm data are obtained. When the match is not successful, the auxiliary information is obtained from the alarm knowledge base and alarm database of the security protection product.
[0115] Next, the target model is used to handle the alarm. When the match is successful, the first alarm data, the target historical alarm data, and the alarm handling information of the target historical alarm data are combined to generate the first prompt information, and the first prompt information is sent to the target model to receive the first alarm handling information returned by the target model. When the match is not successful, the first alarm data and the auxiliary information are combined to generate the second prompt information, and the second prompt information is sent to the target model multiple times to receive multiple second alarm handling information returned by the target model.
[0116] When a user uses a security protection product to handle an alarm, the security protection product can directly receive feedback from the user. For example, by providing controls, the user can judge the correctness of the alarm handling information generated by the target model. For another example, the user can also modify incorrect alarm handling information. For another example, the user can also sort the correct at least one second alarm handling information according to the accuracy.
[0117] The received feedback information can be used to train a reward model, for example, a classification model for determining whether the alarm handling information is correct, or a sorting model for sorting multiple alarm handling information, and then the trained reward model can be used to adjust the parameters of the target model. In this way, during the use of the safety protection product, the target model is continuously subjected to RLHF for a long time to enhance the stability of the safety protection product in handling alarms.
[0118] Furthermore, when users use security protection products to handle alarms, if no match is successful, they can also group and cluster the third alarm handling information generated by the target model and the fourth alarm handling information updated according to the feedback information of the third alarm handling information, extract and generate new alarm feature information, thereby enriching the feature information library of the security protection product and improving the comprehensiveness of the security protection product in alarm handling.
[0119] Combination of the above Figure 1 and Figure 2 The alarm information processing method provided in the embodiment of the present application is introduced in detail. The system and equipment provided in the embodiment of the present application will be introduced in conjunction with the accompanying drawings.
[0120] See also Figure 3 The structural diagram of the alarm information processing system shown in FIG. 30 includes:
[0121] A first acquisition module 301 is used to acquire first alarm data generated by a security detection module;
[0122] A matching module 302, configured to match the first alarm data with at least one piece of alarm characteristic information; wherein one piece of alarm characteristic information corresponds to at least one piece of historical alarm data;
[0123] The second acquisition module 303 is used for acquiring at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data in response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information;
[0124] The handling module 304 is used to generate a first prompt information based on at least the first alarm data, the at least one target historical alarm data and the alarm handling information of the at least one target historical alarm data, send the first prompt information to the target model, and receive the first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling for the first alarm data.
[0125] In some possible implementations, the alarm handling information includes at least one of alarm mark information and alarm attribution information; wherein the alarm mark information indicates a handling label of the alarm data, and the alarm attribution information represents a reason for generating the handling label.
[0126] In some possible implementations, the matching module 302 is specifically configured to:
[0127] Matching the first alarm data with a regular expression corresponding to at least one piece of alarm characteristic information;
[0128] The second acquisition module 303 is specifically used for:
[0129] In response to the alarm field in the first alarm data matching the regular expression corresponding to the target alarm feature information in the at least one alarm feature information, at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data are obtained.
[0130] In some possible implementations, the matching module 302 is specifically configured to:
[0131] For each alarm field of the first alarm data, respectively calculating a text similarity with at least one piece of alarm feature information;
[0132] The second acquisition module 303 is specifically used for:
[0133] In response to the existence of target alarm feature information whose text similarity satisfies a first set condition, at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data are acquired.
[0134] In some possible implementations, the processing module 304 is further configured to:
[0135] In response to the first alarm data not hitting the at least one alarm feature information, second prompt information generated at least based on the first alarm data is sent to the target model multiple times, and multiple second alarm handling information for the first alarm data returned by the target model are received, wherein the second prompt information is used to instruct the target model to perform alarm handling for the first alarm data.
[0136] In some possible implementations, the handling module 304 is specifically configured to:
[0137] Acquire auxiliary information, where the auxiliary information includes at least one of the following: first information obtained by querying with the first alarm data as an index, and second information related to an alarm event that generates the first alarm data;
[0138] generating second prompt information based at least on the first warning data and the auxiliary information;
[0139] The second prompt information is sent to the target model multiple times, and multiple pieces of second alarm handling information for the first alarm data returned by the target model are received.
[0140] In some possible implementations, the system 30 further includes an adjustment module, and the adjustment module is configured to:
[0141] receiving feedback information regarding the first alarm handling information, wherein the feedback information represents a judgment result of a user regarding the first alarm handling information;
[0142] According to the feedback information, adjusting the parameters of the target model;
[0143] The receiving of feedback information for the first alarm handling information includes: providing a first control and a second control, the first control indicating that the first alarm handling information is correct, and the second control indicating that the first alarm handling information is incorrect; in response to a trigger operation on the second control, receiving first modification information for the first alarm handling information; and determining feedback information based on the first modification information.
[0144] In some possible implementations, the adjustment module is further configured to:
[0145] receiving feedback information for the plurality of pieces of second alarm handling information, the feedback information representing a user's judgment result for the second alarm handling information;
[0146] According to the feedback information, the parameters of the target model are adjusted.
[0147] In some possible implementations, the adjustment module is specifically configured to:
[0148] receiving sorting information of the second alarm handling information confirmed to be correct among the plurality of second alarm handling information; and / or receiving information confirming that at least part of the plurality of second alarm handling information is incorrect;
[0149] The feedback information includes the ranking information and / or the information confirmed to be incorrect, and the ranking information represents the user's preference for the second alarm handling information confirmed to be correct among the multiple pieces of second alarm handling information.
[0150] In some possible implementations, the adjustment module is specifically configured to:
[0151] For each piece of the second alarm handling information, a third control and a fourth control are provided, wherein the third control indicates that the second alarm handling information is correct, and the fourth control indicates that the second alarm handling information is incorrect;
[0152] Receive sorting information for at least one second alarm handling information triggering the third control, and / or receive second modification information for at least one second alarm handling information triggering the fourth control.
[0153] In some possible implementations, the adjustment module is specifically configured to:
[0154] The feedback information is used as training data to train a reward model, wherein the reward model includes at least one of the following: a classification model for determining whether the alarm handling information is correct, and a ranking model for ranking multiple alarm handling information, and the reward model is used to measure the inference accuracy of the target model;
[0155] The trained reward model is used to adjust the parameters of the target model.
[0156] In some possible implementations, the feature information database stores at least one existing alarm feature information, and the system 30 further includes a generation module, which is used to:
[0157] Acquire multiple pieces of historical alarm data, wherein the multiple pieces of historical alarm data do not match the at least one piece of existing alarm feature information;
[0158] Determine multiple pieces of historical alarm handling information for the multiple pieces of historical alarm data, wherein the multiple pieces of historical alarm handling information include at least one of the following: third alarm handling information for the multiple pieces of historical alarm data returned by the target model, and fourth alarm handling information after updating the third alarm handling information according to feedback information of the third alarm handling information;
[0159] Classifying the plurality of pieces of historical alarm data according to at least one alarm event that generates the plurality of pieces of historical alarm data to obtain a historical alarm data set of at least one category;
[0160] For each category of historical alarm data set, clustering is performed according to at least one alarm field to obtain a clustering result of the historical alarm data set of each category;
[0161] In response to the existence of a target clustering result in the clustering results satisfying a second set condition, alarm feature information is generated according to the target clustering result, and the generated alarm feature information is stored in the feature information library.
[0162] In some possible implementations, the generating module is specifically used to:
[0163] For each of the clustering results, determining the number of samples in each cluster in the clustering result;
[0164] In response to the existence of a cluster whose sample quantity is greater than a set quantity threshold in the target clustering result, alarm feature information is generated according to the target clustering result, and the generated alarm feature information is stored in the feature information library.
[0165] In some possible implementations, the generating module is specifically used to:
[0166] generating third prompt information based at least on at least one piece of historical alarm data included in the target clustering result;
[0167] The third prompt information is sent to the target model, and the alarm feature information returned by the target model is received, wherein the third prompt information is used to instruct the target model to generate alarm feature information matching at least one historical alarm data included in the target clustering result.
[0168] The alarm information processing system 30 according to the embodiment of the present application may correspond to executing the method described in the embodiment of the present application, and the above and other operations and / or functions of each module / unit of the alarm information processing system 30 are respectively to implement Figure 1 or Figure 2 For the sake of brevity, the corresponding processes of each method in the illustrated embodiment are not described in detail here.
[0169] The present application also provides an electronic device. The electronic device is specifically used to implement Figure 3 The functions of the alarm information processing system 30 in the illustrated embodiment.
[0170] Figure 4 A schematic diagram of the structure of an electronic device 400 is provided. Figure 4 As shown, the electronic device 400 includes a bus 401, a processor 402, a communication interface 403 and a memory 404. The processor 402, the memory 404 and the communication interface 403 communicate with each other via the bus 401.
[0171] The bus 401 may be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, Figure 4 The bus is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.
[0172] The processor 402 may be any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP).
[0173] The communication interface 403 is used for communicating with the outside. For example, the communication interface 403 can be used for communicating with a terminal.
[0174] The memory 404 may include a volatile memory, such as a random access memory (RAM). The memory 404 may also include a non-volatile memory, such as a read-only memory (ROM), a flash memory, a hard disk drive (HDD), or a solid state drive (SSD).
[0175] The memory 404 stores executable codes, and the processor 402 executes the executable codes to perform the aforementioned alarm information processing method.
[0176] Specifically, in implementing Figure 3 In the case of the embodiment shown, and Figure 3 When each module or unit of the alarm information processing system 30 described in the embodiment is implemented by software, the execution Figure 3 The software or program code required for the functions of each module / unit in the system may be partially or completely stored in the memory 404. The processor 402 executes the program code corresponding to each unit stored in the memory 404 to perform the above-mentioned alarm information processing method.
[0177] The embodiment of the present application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that can be stored by the computing device or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a tape), an optical medium (such as a DVD), or a semiconductor medium (such as a solid-state hard disk). The computer-readable storage medium includes instructions that instruct the computing device to execute the above-mentioned alarm information processing method applied to the alarm information processing system 30.
[0178] The embodiment of the present application further provides a computer program product, which includes one or more computer instructions. When the computer instructions are loaded and executed on a computing device, the process or function described in the embodiment of the present application is generated in whole or in part.
[0179] The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions may be transmitted from one website, computer or data center to another website, computer or data center via wired (e.g., coaxial cable, optical fiber, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means.
[0180] When the computer program product is executed by a computer, the computer executes any of the aforementioned alarm information processing methods. The computer program product may be a software installation package, and when any of the aforementioned alarm information processing methods is needed, the computer program product may be downloaded and executed on a computer.
[0181] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to each embodiment of the present application. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0182] The units involved in the embodiments described in this application may be implemented by software or hardware, wherein the name of the unit / module does not, in some cases, constitute a limitation on the unit itself.
[0183] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.
[0184] In the context of the present application embodiment, machine-readable medium can be a tangible medium that can contain or store a program for use by an instruction execution system, device or equipment or used in combination with an instruction execution system, device or equipment. Machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable medium can include but is not limited to electronic, magnetic, optical, electromagnetic, infrared or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of machine-readable storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.
[0185] It should be noted that the various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments, and the same or similar parts between the various embodiments can be referred to each other. For the system or device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part description.
[0186] It should be understood that in the present application, "at least one (item)" means one or more, and "plurality" means two or more. "And / or" is used to describe the association relationship of associated objects, indicating that three relationships may exist. For example, "A and / or B" can mean: only A exists, only B exists, and A and B exist at the same time, where A and B can be singular or plural. The character " / " generally indicates that the objects associated before and after are in an "or" relationship. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, "a and b", "a and c", "b and c", or "a and b and c", where a, b, c can be single or multiple.
[0187] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.
[0188] The steps of the method or algorithm described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, a software module executed by a processor, or a combination of the two. The software module may be placed in a random access memory (RAM), a memory, a read-only memory (ROM), an electrically programmable ROM, an electrically erasable programmable ROM, a register, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.
[0189] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be 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 the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.
Claims
1. A method for processing alarm information, It is characterized in that The method comprises: Acquire first alarm data generated by the security detection module; Matching the first alarm data with at least one piece of alarm characteristic information; wherein one piece of alarm characteristic information corresponds to at least one piece of historical alarm data; In response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information, obtaining at least one target historical alarm data corresponding to the target alarm feature information and the alarm handling information of the at least one target historical alarm data; generating first prompt information based at least on the first alarm data, the at least one target historical alarm data and the alarm handling information of the at least one target historical alarm data, sending the first prompt information to the target model, receiving the first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling for the first alarm data; In response to the first alarm data not hitting the at least one alarm feature information, obtaining auxiliary information, the auxiliary information comprising at least one of the following: first information obtained by querying with the first alarm data as an index, and second information related to the alarm event that generates the first alarm data; generating second prompt information based at least on the first alarm data and the auxiliary information, the second prompt information being used to instruct the target model to perform alarm handling for the first alarm data; sending the second prompt information to the target model multiple times, and receiving multiple second alarm handling information for the first alarm data returned by the target model.
2. The method according to claim 1, It is characterized in that The alarm handling information includes at least one of alarm mark information and alarm attribution information; wherein the alarm mark information indicates a handling label of the alarm data, and the alarm attribution information represents a reason for generating the handling label.
3. The method according to claim 1, It is characterized in that The matching of the first alarm data with at least one piece of alarm characteristic information includes: Matching the first alarm data with a regular expression corresponding to at least one piece of alarm characteristic information; The step of acquiring, in response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information, at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data, comprises: In response to the alarm field in the first alarm data matching the regular expression corresponding to the target alarm feature information in the at least one alarm feature information, at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data are obtained.
4. The method according to claim 1, It is characterized in that The matching of the first alarm data with at least one piece of alarm characteristic information includes: For each alarm field of the first alarm data, respectively calculating a text similarity with at least one piece of alarm feature information; The step of acquiring, in response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information, at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data, comprises: In response to the existence of target alarm feature information whose text similarity satisfies a first set condition, at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data are acquired.
5. The method according to claim 1, It is characterized in that The method further comprises: receiving feedback information regarding the first alarm handling information, wherein the feedback information represents a judgment result of a user regarding the first alarm handling information; According to the feedback information, adjusting the parameters of the target model; The receiving of feedback information for the first alarm handling information includes: providing a first control and a second control, the first control indicating that the first alarm handling information is correct, and the second control indicating that the first alarm handling information is incorrect; in response to a trigger operation on the second control, receiving first modification information for the first alarm handling information; and determining feedback information based on the first modification information.
6. The method according to claim 1, It is characterized in that The method further comprises: receiving feedback information for the plurality of pieces of second alarm handling information, the feedback information representing a judgment result of a user for the second alarm handling information; According to the feedback information, the parameters of the target model are adjusted.
7. The method according to claim 6, It is characterized in that The receiving feedback information for the plurality of pieces of second alarm handling information includes: receiving sorting information of the second alarm handling information confirmed to be correct among the plurality of second alarm handling information; and / or receiving information confirming that at least part of the plurality of second alarm handling information is incorrect; The feedback information includes the ranking information and / or the information confirmed to be incorrect, and the ranking information represents the user's preference for the second alarm handling information confirmed to be correct among the multiple pieces of second alarm handling information.
8. The method according to claim 7, It is characterized in that The receiving of sorting information of the second alarm handling information confirmed to be correct among the plurality of second alarm handling information; And / or, receiving information confirming that at least part of the plurality of second alarm handling information is incorrect, comprising: For each piece of the second alarm handling information, a third control and a fourth control are provided, wherein the third control indicates that the second alarm handling information is correct, and the fourth control indicates that the second alarm handling information is incorrect; Receive sorting information for at least one second alarm handling information triggering the third control, and / or receive second modification information for at least one second alarm handling information triggering the fourth control.
9. The method according to claim 5 or 6, It is characterized in that The adjusting the parameters of the target model according to the feedback information includes: The feedback information is used as training data to train a reward model, wherein the reward model includes at least one of the following: a classification model for determining whether the alarm handling information is correct, and a ranking model for ranking multiple alarm handling information, and the reward model is used to measure the inference accuracy of the target model; The trained reward model is used to adjust the parameters of the target model.
10. The method according to any one of claims 1 to 9, It is characterized in that The feature information database stores at least one existing alarm feature information, and the method further includes: Acquire multiple pieces of historical alarm data, wherein the multiple pieces of historical alarm data do not match the at least one piece of existing alarm feature information; Determine multiple pieces of historical alarm handling information for the multiple pieces of historical alarm data, wherein the multiple pieces of historical alarm handling information include at least one of the following: third alarm handling information for the multiple pieces of historical alarm data returned by the target model, and fourth alarm handling information after updating the third alarm handling information according to feedback information of the third alarm handling information; Classifying the plurality of pieces of historical alarm data according to at least one alarm event that generates the plurality of pieces of historical alarm data to obtain a historical alarm data set of at least one category; For each category of historical alarm data set, clustering is performed according to at least one alarm field to obtain a clustering result of the historical alarm data set of each category; In response to the existence of a target clustering result in the clustering results satisfying a second set condition, alarm feature information is generated according to the target clustering result, and the generated alarm feature information is stored in the feature information library.
11. The method according to claim 10, It is characterized in that In response to the existence of a target clustering result in the clustering results satisfying a second set condition, generating alarm feature information according to the target clustering result, and storing the generated alarm feature information in the feature information library, including: For each of the clustering results, determining the number of samples in each cluster in the clustering result; In response to the existence of a cluster whose sample quantity is greater than a set quantity threshold in the target clustering result, alarm feature information is generated according to the target clustering result, and the generated alarm feature information is stored in the feature information library.
12. The method according to claim 10, It is characterized in that The generating of alarm feature information according to the target clustering result includes: generating third prompt information based at least on at least one piece of historical alarm data included in the target clustering result; The third prompt information is sent to the target model, and the alarm feature information returned by the target model is received, wherein the third prompt information is used to instruct the target model to generate alarm feature information matching at least one historical alarm data included in the target clustering result.
13. An alarm information processing system, It is characterized in that The system comprises: A first acquisition module, used to acquire first alarm data generated by the security detection module; A matching module, used to match the first alarm data with at least one piece of alarm characteristic information; wherein one piece of alarm characteristic information corresponds to at least one piece of historical alarm data; A second acquisition module is used for acquiring at least one target historical alarm data corresponding to the target alarm feature information and alarm handling information of the at least one target historical alarm data in response to the first alarm data hitting the target alarm feature information in the at least one alarm feature information; a handling module, configured to generate first prompt information based at least on the first alarm data, the at least one target historical alarm data and the alarm handling information of the at least one target historical alarm data, send the first prompt information to a target model, and receive the first alarm handling information for the first alarm data returned by the target model, wherein the first prompt information is used to instruct the target model to perform alarm handling for the first alarm data; The handling module is also used to obtain auxiliary information in response to the first alarm data not hitting the at least one alarm feature information, the auxiliary information including at least one of the following: first information obtained by querying with the first alarm data as an index, and second information related to the alarm event that generates the first alarm data; generating second prompt information based at least on the first alarm data and the auxiliary information, the second prompt information being used to instruct the target model to perform alarm handling for the first alarm data; sending the second prompt information to the target model multiple times, and receiving multiple second alarm handling information for the first alarm data returned by the target model.
14. An electronic device, It is characterized in that The electronic device comprises a processor and a memory; The processor is configured to execute instructions stored in the memory, so that the electronic device performs the method according to any one of claims 1 to 12.
15. A computer-readable storage medium, It is characterized in that The method comprises instructions, wherein the instructions instruct an electronic device to execute the method according to any one of claims 1 to 12.
