Abnormal behavior data detection method and device

By constructing target behavior examples and performing multi-dimensional calculations under different dimension labels, it judges its anomalies in high-dimensional space, and solves the problem that single-point detection analysis in the existing technology is difficult to deal with unknown threats, and improves abnormal behavior data detection accuracy.

CN119011159APending Publication Date: 2024-11-22BEIJING GRIDSUM TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202310552447.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2023-05-16
Publication Date
2024-11-22

AI Technical Summary

Technical Problem

In the prior art, abnormal behavior data detection is mainly based on single-point detection analysis, and it is difficult to achieve multi-dimensional analysis and respond to unknown threats, resulting in low detection accuracy.

Method used

By constructing target behavior examples, using the multi-dimensional calculation results of target behavior characteristics under different dimension labels, we can judge whether its spatial position in high-dimensional space is abnormal, and generate abnormal alarm information.

Benefits of technology

Multi-dimensional analysis of abnormal behavior data is realized, the accuracy of detection of unknown threats is improved, and the accuracy of detection of abnormal behavior data is improved.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119011159A_ABST
    Figure CN119011159A_ABST
Patent Text Reader

Abstract

The invention discloses an abnormal behavior data detection method and device, relates to the technical field of network security detection, and mainly aims to analyze behavior data in multiple dimensions to cope with unknown threats so as to improve the detection accuracy of the abnormal behavior data. According to the main technical scheme, a target behavior sample is constructed based on target behavior characteristics in target behavior data, and the target behavior characteristics are used for representing behavior characteristics which belong to different dimension labels and are subjected to optimization processing; calculating a multi-dimensional calculation result of the target behavior sample under different dimension labels by using a specified weight corresponding to the target behavior feature; judging whether the target behavior sample is abnormal or not according to the behavior area label of the corresponding spatial position of the multi-dimensional calculation result in the high-dimensional space; and if yes, generating behavior abnormity alarm information corresponding to the target behavior data. The method is used for detecting the abnormal behavior data.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application relates to the field of network security detection technology, and in particular to a method and device for detecting abnormal behavior data. Background Art

[0002] With the rapid development of Internet technology, the interconnection and data interaction between networks have made the relationship between people and information closer, and network security protection is particularly important. As network attack technology develops faster and faster, people have more and more ways and channels to collect data, resulting in a significant increase in the generation of abnormal behavior data. Therefore, it is increasingly important to detect abnormal behavior data in order to improve the corresponding network security protection measures.

[0003] At present, abnormal behavior data detection in the prior art is generally based on single-point detection analysis, that is, detecting behavior data from a single dimension based on known threats. However, single-dimensional detection is difficult to achieve correlation analysis of behavior data and cannot cope with unknown threats, resulting in low accuracy in detecting abnormal behavior data. Summary of the invention

[0004] In view of the above problems, the present application provides a method and device for detecting abnormal behavior data, the main purpose of which is to analyze the behavior data in multiple dimensions to deal with unknown threats, thereby improving the detection accuracy of abnormal behavior data.

[0005] In order to solve the above technical problems, this application proposes the following solutions:

[0006] In a first aspect, the present application provides a method for detecting abnormal behavior data, the method comprising:

[0007] Constructing a target behavior sample based on the target behavior features in the target behavior data, wherein the target behavior features are used to characterize the optimized behavior features belonging to different dimensional labels;

[0008] Calculate the multi-dimensional calculation results of the target behavior sample under different dimensional labels using the specified weights corresponding to the target behavior features;

[0009] Determining whether the target behavior sample is abnormal based on the behavior region label of the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space;

[0010] If so, abnormal behavior alarm information corresponding to the target behavior data is generated.

[0011] In a second aspect, the present application provides an abnormal behavior data detection device, the device comprising:

[0012] A construction unit, configured to construct a target behavior sample based on target behavior features in the target behavior data, wherein the target behavior features are used to characterize optimized behavior features belonging to different dimensional labels;

[0013] A calculation unit, configured to calculate the multi-dimensional calculation results of the target behavior sample obtained by the construction unit under different dimensional labels by using the specified weights corresponding to the target behavior features;

[0014] A judging unit, configured to judge whether the target behavior sample is abnormal according to the behavior region label of the spatial position corresponding to the multi-dimensional calculation result obtained by the calculating unit in the high-dimensional space;

[0015] A generating unit is used to generate abnormal behavior alarm information corresponding to the target behavior data if the judging unit judges that the target behavior sample is abnormal. In order to achieve the above-mentioned purpose, according to the third aspect of the present application, a storage medium is provided, wherein the storage medium includes a stored program, wherein when the program is run, the device where the storage medium is located is controlled to execute the abnormal behavior data detection method of the first aspect.

[0016] In order to achieve the above-mentioned purpose, according to the fourth aspect of the present application, an electronic device is provided, which includes at least one processor, and at least one memory and a bus connected to the processor; wherein the processor and the memory communicate with each other through the bus; the processor is used to call the program instructions in the memory to execute the abnormal behavior data detection method of the first aspect above.

[0017] By means of the above technical scheme, the present application provides an abnormal behavior data detection method and device, when it is necessary to detect abnormal behavior data, firstly, a target behavior sample is constructed based on the target behavior feature in the target behavior data, wherein the target behavior feature is used to characterize the behavior feature that belongs to different dimensional labels and has been optimized, and then the multi-dimensional calculation results of the target behavior sample under different dimensional labels are calculated using the specified weights corresponding to the target behavior feature, and then the target behavior sample is judged whether it is abnormal according to the behavior area label of the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space, and finally, if so, the behavior abnormality alarm information corresponding to the target behavior data is generated. Through the technical scheme provided by the present application, it is possible to construct the behavior sample based on the behavior feature of the behavior data under different dimensional labels, and the multi-dimensional weight results of the behavior sample under different dimensional labels, and judge whether the behavior sample is abnormal based on the behavior area label of the spatial position corresponding to the multi-dimensional weight result in the high-dimensional space, that is, to analyze the behavior data from multiple dimensions, and to effectively distinguish unknown threats, thereby improving the detection accuracy of abnormal behavior data.

[0018] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Various other advantages and benefits will become apparent to those of ordinary skill in the art by reading the detailed description of the preferred embodiments below. The accompanying drawings are only for the purpose of illustrating the preferred embodiments and are not to be considered as limiting the present application. Also, the same reference symbols are used throughout the accompanying drawings to represent the same components. In the accompanying drawings:

[0020] Figure 1 A flow chart of an abnormal behavior data detection method provided by an embodiment of the present application is shown;

[0021] Figure 2 Another abnormal behavior data detection method flow chart provided in an embodiment of the present application is shown;

[0022] Figure 3 A block diagram showing a composition of an abnormal behavior data detection device provided in an embodiment of the present application is shown;

[0023] Figure 4 A block diagram showing another abnormal behavior data detection device provided by an embodiment of the present application is shown;

[0024] Figure 5 A schematic diagram of an electronic device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0025] The exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although the exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0026] At present, abnormal behavior data detection in the prior art is generally based on single-point detection and analysis, that is, the behavior data is detected from a single dimension based on known threats. However, single-dimensional detection makes it difficult to achieve correlation analysis of behavior data, and cannot cope with unknown threats, resulting in low accuracy in detecting abnormal behavior data. The present application calculates the multidimensional weight results of behavior samples constructed based on the behavior characteristics of behavior data under different dimensional labels under different dimensional labels, and judges whether the behavior samples are abnormal based on the behavior area labels of the spatial positions corresponding to the multidimensional weight results in high-dimensional space, that is, analyzes the behavior data from multiple dimensions, which can effectively distinguish unknown threats, thereby improving the accuracy of detecting abnormal behavior data.

[0027] The embodiment of the present application provides a method for detecting abnormal behavior data, through which the behavior data can be analyzed in multiple dimensions to deal with unknown threats, thereby improving the detection accuracy of abnormal behavior data. The specific execution steps are as follows: Figure 1 As shown, including:

[0028] 101. Construct a target behavior sample based on the target behavior features in the target behavior data.

[0029] Among them, the target behavior feature is used to characterize the behavior features that belong to different dimensional labels and are optimized. It should be noted that in this embodiment, the target behavior sample is a complete description of the target behavior data based on the behavior features under different dimensional labels and after optimization, and the different dimensional labels include operator identity, operation time, operation category, access method, operation object and operation process, that is, what operator identity performs what operation category at what operation time with what access method on what operation object with what operation process, so as to accurately describe the target behavior feature, and the behavior features under different dimensional labels can be obtained from the traffic data and log data of the target behavior data, and the optimization process is to merge the behavior features with the same frequency under different dimensional labels or to manually clean the behavior features based on the PDCA model, so as to delete the invalid and redundant behavior features for subsequent calculations, so as to obtain effective target behavior features, so as to accurately describe the target behavior data and reduce the subsequent calculation burden based on the target behavior sample, and in this step, the target behavior data generally refers to the behavior data corresponding to the most recent traffic data and log data.

[0030] 102. Use the specified weights corresponding to the target behavior features to calculate the multi-dimensional calculation results of the target behavior samples under different dimensional labels.

[0031] It should be noted that, in this embodiment, by setting specified weights, it is possible to characterize the degree of attenuation or increase of each feature in the target behavior feature within its corresponding life cycle, so as to judge the suspiciousness of the target behavior sample based on this, that is, to judge whether the target behavior data is abnormal. The specified weight can be set manually or determined based on the attribute characteristics of the attribute information of the target behavior data. This embodiment does not limit this, and the attribute characteristics include but are not limited to business sensitivity, business level, business operation frequency, business operation time, etc. Since it is difficult to analyze the behavior data in a single dimension, in this step, the multi-dimensional calculation results of the target behavior sample under different dimensional labels can be calculated based on the specified weights to improve the judgment accuracy when executing step 103 later.

[0032] 103. Determine whether the target behavior sample is abnormal based on the behavior area label of the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space.

[0033] It should be noted that, in this embodiment, the high-dimensional space can be constructed in advance based on the historical multi-dimensional calculation results corresponding to the historical behavior data within the specified period, that is, the historical spatial position point of each historical behavior data in the high-dimensional space is determined, and since the historical behavior data within the specified period includes the historical abnormal behavior data and the historical normal behavior data that have been determined, and the distribution of multiple historical spatial position points of the historical abnormal behavior data and the historical normal behavior data in the high-dimensional space can be used to determine whether the target behavior sample is abnormal. Specifically, the high-dimensional space can be divided into high-dimensional cubes in advance, that is, high-dimensional cubes are constructed one by one from the origin of the space outward according to the equal length of the sides, and then behavior area labels are deployed for each high-dimensional cube according to the probability of occurrence of the historical spatial position point in each high-dimensional cube, such as a normal behavior area label or an abnormal behavior label, etc. Therefore, the behavior area label of the spatial position corresponding to the multi-dimensional calculation result of the target behavior sample in the high-dimensional space is a normal behavior area label or an abnormal behavior label, and it can be determined whether the target behavior sample is abnormal. If it is not abnormal, step 101 is continued to be executed when new target behavior data is generated. If it is abnormal, the subsequent step 104 is executed.

[0034] 104. Generate abnormal behavior alarm information corresponding to the target behavior data.

[0035] It should be noted that, in this embodiment, since the aforementioned step 103 has determined that the target behavior sample is abnormal, it means that the target behavior data is abnormal behavior data. Therefore, corresponding behavior abnormality alarm information can be generated at this time to alert maintenance personnel to the network security threat at this time, so as to promptly process the abnormal behavior data.

[0036] Based on the above Figure 1It can be seen from the implementation method that the abnormal behavior data detection method provided by the present application is to first construct a target behavior sample based on the target behavior features in the target behavior data when abnormal behavior data needs to be detected, wherein the target behavior features are used to characterize the behavior features that belong to different dimensional labels and are optimized, and then use the specified weights corresponding to the target behavior features to calculate the multidimensional calculation results of the target behavior samples under different dimensional labels, and then judge whether the target behavior sample is abnormal based on the behavior area label of the spatial position corresponding to the multidimensional calculation result in the high-dimensional space, and finally, if so, generate the behavior abnormality alarm information corresponding to the target behavior data. Through the technical solution provided by the present application, it is possible to construct the behavior samples based on the behavior features of the behavior data under different dimensional labels, and the multidimensional weight results of the behavior samples under different dimensional labels, and judge whether the behavior samples are abnormal based on the behavior area label of the spatial position corresponding to the multidimensional weight result in the high-dimensional space, that is, analyze the behavior data from multiple dimensions, and effectively distinguish unknown threats, thereby improving the detection accuracy of abnormal behavior data.

[0037] Further, the preferred embodiment of the present application is in the above Figure 1 Based on this, a detailed description of the abnormal behavior data detection process is given. The specific steps are as follows Figure 2 As shown, including:

[0038] 201. Extract specified behavior features from target behavior data based on preset dimension labels to obtain specified behavior features under different dimension labels.

[0039] It should be noted that, in this step, there are multiple preset dimension labels, including operator identity, operation time, operation category, access method, operation object and operation process, and the specified behavior characteristics are the behavior characteristics selected by the user that can or have a trend to affect the detection of abnormal behavior data. Specifically, they can be set according to human experience or determined based on historical data. This embodiment does not limit this. Among them, operator identity: that is, the user's attribute label, such as account, IP, region, terminal, role, authority, etc.; operation time: that is, the time when the operation behavior occurs, and the frequency of the operation behavior is statistically derived; operation category: mainly manifested as the access category of the operation, such as query category, processing category, etc. Different operation categories vary greatly. For example, query-type behaviors include login, browsing, searching, input (mobile phone number, ID number, name, etc.), while handling-type behaviors are further operations based on query, such as recharge, change of service, deactivation of service, input, etc.; access mode: that is, what channel is used to complete the operation, such as dedicated client software, middleware (middleware is usually used for database access), command line (CLI), internal login, external jump, remote desktop, etc.; operation object: that is, various resources, such as files, database tables, server hosts, data items, various application systems (BOSS system, CRM system, business analysis, operation management, VGOP / ESOP system), etc.; operation process: that is, the process corresponding to the operation behavior, such as customer query: login authentication>A module>B module>C module. Since there are invalid data such as dirty data and unreadable data in traffic data and log data, by extracting the specified behavior features under the above different dimensional labels, invalid data such as dirty data and unreadable data can be discarded, so that the target behavior data can be accurately described for the subsequent step 202.

[0040] 202. Optimize the specified behavior features under different dimension labels according to the optimization rules to obtain the target behavior features.

[0041] Among them, the optimization rules are used to characterize the rules for merging or cleaning the specified behavior features under different dimensional labels based on the same frequency situation and / or the PDCA model. It should be noted that in this step, since there are multiple different specified behavior features under each dimensional label, and different specified behavior features will have the same frequency situation, such as specified behavior features with strong correlation or similarity, that is, the application frequency increases or decreases at the same time, and for such specified behavior features, since their changes are the same, they can be merged to reduce the subsequent calculation burden, and there will be some specified behavior features whose application frequency is extremely low in stages, and for such specified behavior features, they can be cleaned. Specifically, they can be manually managed based on the PDCA model to obtain the target behavior features under each dimensional label, so as to form the target behavior features corresponding to the abnormal behavior data.

[0042] 203. Construct target behavior samples based on target behavior characteristics.

[0043] It should be noted that in this step, since the target behavior features are used to characterize the optimized behavior features belonging to different dimensional labels, that is, the merged and cleaned behavior features, it is possible to construct target behavior samples to accurately and completely describe the target behavior data so as to execute the subsequent step 204.

[0044] 204. Calculate the multi-dimensional calculation results of the target behavior sample under different dimensional labels using the specified weights corresponding to the target behavior features.

[0045] This step is combined with the description of step 102 in the above method, and the same content is not repeated here. It should be noted that in this step, the specific execution process for calculating the multidimensional calculation results of the target behavior sample under different dimensional labels is: obtain the specified weight corresponding to the target behavior feature; calculate the weight result corresponding to each dimensional label according to the specified weight corresponding to each target behavior feature; use multiple weight results as the multidimensional calculation results of the target behavior sample under different dimensional labels. Among them, the size of the specified weight is used to characterize the degree of attenuation or increase of each feature in the target behavior feature during its corresponding life cycle, so as to judge the suspicious degree of the target behavior sample based on this, that is, to judge whether the target behavior data is abnormal, and because the target behavior feature corresponds to different dimensional labels, the multidimensional calculation result is composed of the weight results corresponding to each dimensional label.

[0046] Furthermore, since the size of the specified weight can affect the accuracy of detecting abnormal behavior data, in order to adapt to the actual situation of actual business handling and make abnormal behavior more prominent, thereby improving the accuracy of detecting abnormal behavior data, specifically, obtain attribute information corresponding to the target behavior feature; determine the weight level corresponding to the target behavior feature according to the attribute feature corresponding to the attribute information, wherein the attribute feature includes at least one of business sensitivity, business level, business operation frequency, and business operation time; and configure the specified weight corresponding to the target behavior feature according to the weight level. Among them, business sensitivity reflects the level of sensitive information involved in the business handling process, and the higher the business sensitivity, the higher the corresponding weight level, for example: points redemption > customer information query > customer information browsing, and business operation frequency reflects the frequency of staged operations, and the higher the business operation frequency, the higher the corresponding weight level, for example: customer information query 5 minutes 50 times > 5 minutes 10 times, and business operation time reflects the time when the behavior occurs and ends. Since business handling generally occurs during working hours, that is, 8 am to 5 pm or 9 am to 6 pm on non-holidays, the more it is during non-working hours, the higher the corresponding weight level, for example: non-working hours > working hours, and business level reflects the importance of the business. The higher the business level, the higher the corresponding weight level, for example: important business > ordinary business, business handling > business query. The weight level corresponds to high, medium, low or first, second, third, etc. Specifically, a corresponding relationship table of attribute characteristics and weight levels can be maintained in advance, so as to determine the corresponding weight level by looking up the table.

[0047] 205. Determine the spatial position corresponding to the multidimensional array composed of multidimensional calculation results in the high-dimensional space.

[0048] It should be noted that, in this step, the number of dimensions of the high-dimensional space corresponds to the number of dimensions in the multidimensional calculation results. Since the high-dimensional space is pre-constructed based on the historical multidimensional calculation results corresponding to the historical behavior data within a specified period, and each high-dimensional cube in the high-dimensional space is deployed with a corresponding behavior area label, the spatial position of the target behavior sample in the high-dimensional space can be determined based on the multidimensional array composed of the multidimensional calculation results of the target behavior sample, so as to execute the subsequent step 206.

[0049] 206. Determine the target high-dimensional cube corresponding to the target behavior sample based on the spatial position.

[0050] It should be noted that in this step, since the high-dimensional space will also construct multiple high-dimensional cubes in sequence outward with specified side lengths based on the space origin in advance, the target high-dimensional cube corresponding to the target behavior sample can be determined in the multiple high-dimensional cubes based on the spatial position, and accordingly, each high-dimensional cube is deployed with a corresponding behavior area label, which includes a normal behavior area label or an abnormal behavior area label. Therefore, it can be determined whether the target behavior sample is abnormal based on whether the behavior area label corresponding to the target high-dimensional cube is a normal behavior area label or an abnormal behavior area label. If not, execute the subsequent step 207, otherwise execute the subsequent step 208.

[0051] 207. If the behavior region label corresponding to the target high-dimensional cube is a normal behavior region label, it is determined that the target behavior sample is not abnormal.

[0052] 208. If the behavior region label corresponding to the target high-dimensional cube is an abnormal behavior region label, the target behavior sample is determined to be abnormal.

[0053] Furthermore, since the high-dimensional space and high-dimensional cube of the execution process of steps 205-208 are pre-established, in order to ensure the accuracy of detecting whether the target behavior sample is abnormal during the execution of steps 205-208, it can be constructed based on the historical behavior data within the specified period, and the target behavior sample can be detected. Specifically, the historical multi-dimensional calculation results corresponding to the historical behavior data within the specified period are obtained; the historical space position corresponding to each historical multi-dimensional calculation result in the high-dimensional space is determined; the high-dimensional space is divided into high-dimensional cubes, and the occurrence probability of the historical space position in each high-dimensional cube is calculated; and an abnormal behavior area label or a normal behavior area label is deployed for each high-dimensional cube according to the comparison result between the occurrence probability and the preset probability threshold, wherein the preset probability threshold is determined according to the distribution corresponding to the historical space position. It should be noted that the specified period can be one month or half a month, etc., and this embodiment does not limit this, and can be manually customized, that is, the difference between the current target behavior data and the historical behavior data is used to determine whether the target behavior data is abnormal, and the high-dimensional cube can be divided outward from the spatial origin based on the specified side length, and the occurrence probability is used to characterize the proportion of the number of historical spatial positions contained in a high-dimensional cube relative to all historical spatial positions. Since there are relatively more historical normal behavior data and they are generally similar, their distribution is relatively concentrated, that is, historical abnormal behavior data is more dispersed than normal behavior data, and since the historical normal behavior data and historical abnormal behavior data included in the historical behavior data are known, a preset probability threshold can be determined based on the distribution, so as to deploy the behavior area label of each high-dimensional cube based on the comparison between the preset probability threshold and each occurrence probability, that is, if it is higher than the preset probability threshold, the abnormal behavior area label is deployed, otherwise the normal behavior area label is deployed, so as to detect anomalies in the target behavior data.

[0054] 209. Generate abnormal behavior alarm information corresponding to the target behavior data.

[0055] This step is combined with the description of step 104 in the above method, and the same contents are not repeated here.

[0056] Furthermore, in order to further ensure the accuracy of abnormal behavior data detection, specifically, the specified weights and behavior area labels in the high-dimensional space are updated periodically. That is, the specified weights and behavior area labels in the high-dimensional space in this embodiment are updated regularly, thereby ensuring the finiteness of the specified weights and behavior area labels in the high-dimensional space, so as to achieve the corresponding detection effect.

[0057] Furthermore, as a response to the above Figure 1-2The implementation of the method embodiment shown in the figure, the embodiment of the present application provides an abnormal behavior data detection device, which is used to analyze the behavior data in multiple dimensions to deal with unknown threats, thereby improving the detection accuracy of abnormal behavior data. The embodiment of the device corresponds to the aforementioned method embodiment. For ease of reading, this embodiment will no longer repeat the details of the aforementioned method embodiment one by one, but it should be clear that the device in this embodiment can correspond to all the contents of the aforementioned method embodiment. Specifically, Figure 3 As shown, the device comprises:

[0058] A construction unit 31 is used to construct a target behavior sample based on the target behavior features in the target behavior data, wherein the target behavior features are used to characterize the optimized behavior features belonging to different dimensional labels;

[0059] A calculation unit 32, configured to calculate the multi-dimensional calculation results of the target behavior sample obtained by the construction unit 31 under different dimensional labels by using the specified weights corresponding to the target behavior features;

[0060] A judging unit 33, configured to judge whether the target behavior sample is abnormal according to the behavior region label of the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space obtained by the calculating unit 32;

[0061] The generating unit 34 is configured to generate abnormal behavior alarm information corresponding to the target behavior data if the judging unit 33 judges that the target behavior sample is abnormal.

[0062] Further, such as Figure 4 As shown, the construction unit 31 includes:

[0063] An extraction module 311 is used to extract specified behavior features from the target behavior data based on preset dimension labels to obtain specified behavior features under different dimension labels;

[0064] The processing module 312 is used to optimize the specified behavior features under different dimensional labels obtained by the extraction module 311 according to the optimization rules to obtain the target behavior features, wherein the optimization rules are used to represent the rules for merging or cleaning the specified behavior features under different dimensional labels based on the same frequency situation and / or the PDCA model;

[0065] The construction module 313 is used to construct the target behavior sample based on the target behavior feature obtained by the processing module 312.

[0066] Further, such as Figure 4 As shown, the calculation unit 32 includes:

[0067] An acquisition module 321 is used to acquire a specified weight corresponding to the target behavior feature;

[0068] A calculation module 322, used to calculate the weight result corresponding to each dimension label according to the specified weight corresponding to each target behavior feature obtained by the acquisition module 321;

[0069] The first determination module 323 is used to use the multiple weight results obtained by the calculation module 322 as the multi-dimensional calculation results of the target behavior sample under different dimensional labels.

[0070] Further, such as Figure 4 As shown, the acquisition module 321 is specifically used for:

[0071] Acquire attribute information corresponding to the target behavior feature;

[0072] Determining a weight level corresponding to the target behavior feature according to the attribute feature corresponding to the attribute information, wherein the attribute feature includes at least one of business sensitivity, business level, business operation frequency, and business operation time;

[0073] The designated weight corresponding to the target behavior feature is configured according to the weight level.

[0074] Further, such as Figure 4 As shown, the judging unit 32 includes:

[0075] A second determination module 321 is used to determine the spatial position corresponding to the multidimensional array composed of the multidimensional calculation results in the high-dimensional space;

[0076] A third determination module 322, configured to determine a target high-dimensional cube corresponding to the target behavior example based on the spatial position obtained by the second determination module 321;

[0077] A fourth determination module 323 is configured to determine that the target behavior sample is not abnormal if the behavior region label corresponding to the target high-dimensional cube obtained by the third determination module 322 is a normal behavior region label;

[0078] The fourth determination module 323 is further configured to determine that the target behavior sample is abnormal if the behavior region label corresponding to the target high-dimensional cube obtained by the third determination module 322 is an abnormal behavior region label.

[0079] Further, such as Figure 4 As shown, the device also includes:

[0080] An acquisition unit 35, used for acquiring the historical multi-dimensional calculation results corresponding to the historical behavior data within a specified period before the judgment unit 32;

[0081] A determination unit 36, configured to determine a historical spatial position corresponding to each of the historical multi-dimensional calculation results obtained by the acquisition unit 35 in the high-dimensional space;

[0082] A processing unit 37 is used to divide the high-dimensional space obtained by the determining unit 36 ​​into high-dimensional cubes, and calculate the occurrence probability of the historical space position in each high-dimensional cube;

[0083] A deployment unit 38 is used to deploy the abnormal behavior area label or the normal behavior area label for each of the high-dimensional cubes according to the comparison result between the occurrence probability obtained by the processing unit 37 and the preset probability threshold, wherein the preset probability threshold is determined according to the distribution corresponding to the historical spatial position.

[0084] Further, such as Figure 4 As shown, the device also includes:

[0085] The updating unit 39 is used to periodically update the designated weights and the behavior region labels in the high-dimensional space.

[0086] Furthermore, an embodiment of the present application also provides an electronic device, the electronic device includes a processor and a memory, the construction unit 31, the calculation unit 32, the judgment unit 33 and the generation unit 34 are all stored in the memory as program units, and the processor executes the program units stored in the memory to implement corresponding functions. The processor includes a kernel, and the kernel retrieves the corresponding program unit from the memory.

[0087] The kernel can be set to one or more, and the behavior data can be analyzed in multiple dimensions by adjusting the kernel parameters to deal with unknown threats, thereby improving the detection accuracy of abnormal behavior data.

[0088] An embodiment of the present application provides a storage medium on which a program is stored. When the program is executed by a processor, the abnormal behavior data detection method is implemented.

[0089] An embodiment of the present application provides a processor, which is used to run a program, wherein the abnormal behavior data detection method is executed when the program is running.

[0090] The present application embodiment provides a device 40, such as Figure 5As shown, the device 40 includes at least one processor 401, at least one memory 402 connected to the processor 401, and a bus 403; wherein the processor 401 and the memory 402 communicate with each other through the bus 403; the processor 401 is used to call the program instructions in the memory 402 to execute the above-mentioned abnormal behavior data detection method. The device in this article can be a server, a PC, a PAD, a mobile phone, etc.

[0091] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program for initializing the following method steps:

[0092] Construct a target behavior sample based on the target behavior feature in the target behavior data, wherein the target behavior feature is used to characterize the behavior feature that belongs to different dimensional labels and has been optimized; calculate the multidimensional calculation results of the target behavior sample under different dimensional labels using the specified weights corresponding to the target behavior feature; determine whether the target behavior sample is abnormal based on the behavior area label of the spatial position corresponding to the multidimensional calculation result in the high-dimensional space;

[0093] If so, abnormal behavior alarm information corresponding to the target behavior data is generated.

[0094] The target behavior sample is constructed based on the target behavior features in the target behavior data, including: extracting specified behavior features from the target behavior data based on preset dimension labels to obtain specified behavior features under different dimension labels; optimizing the specified behavior features under different dimension labels according to optimization rules to obtain the target behavior features, wherein the optimization rules are used to characterize the rules for merging or cleaning the specified behavior features under different dimension labels based on the same frequency situation and / or the PDCA model; and constructing the target behavior sample based on the target behavior features.

[0095] The method of calculating the multi-dimensional calculation results of the target behavior samples under different dimensional labels by using the specified weights corresponding to the target behavior features includes: obtaining the specified weights corresponding to the target behavior features; calculating the weight results corresponding to each dimensional label according to the specified weights corresponding to each target behavior feature; and using multiple weight results as the multi-dimensional calculation results of the target behavior samples under different dimensional labels.

[0096] The obtaining of the designated weight corresponding to the target behavior feature comprises: obtaining attribute information corresponding to the target behavior feature; determining a weight level corresponding to the target behavior feature according to an attribute feature corresponding to the attribute information, wherein the attribute feature comprises at least one of business sensitivity, business level, business operation frequency, and business operation time; and configuring the designated weight corresponding to the target behavior feature according to the weight level.

[0097] The method of judging whether the target behavior sample is abnormal based on the behavior region label of the spatial position corresponding to the multidimensional calculation result in the high-dimensional space includes: determining the spatial position corresponding to the multidimensional array composed of the multidimensional calculation result in the high-dimensional space; determining the target high-dimensional cube corresponding to the target behavior sample based on the spatial position; if the behavior region label corresponding to the target high-dimensional cube is a normal behavior region label, determining that the target behavior sample is not abnormal;

[0098] If the behavior region label corresponding to the target high-dimensional cube is an abnormal behavior region label, it is determined that the target behavior sample is abnormal.

[0099] Before judging whether the target behavior sample is abnormal based on the behavior area label of the spatial position corresponding to the multidimensional calculation result in the high-dimensional space, the method also includes: obtaining historical multidimensional calculation results corresponding to historical behavior data within a specified period; determining the historical spatial position corresponding to each of the historical multidimensional calculation results in the high-dimensional space; dividing the high-dimensional space into high-dimensional cubes, and calculating the occurrence probability of the historical spatial position in each of the high-dimensional cubes; deploying the abnormal behavior area label or the normal behavior area label for each of the high-dimensional cubes according to the comparison result of the occurrence probability with a preset probability threshold, wherein the preset probability threshold is determined according to the distribution corresponding to the historical spatial position.

[0100] The designated weights and the behavior region labels in the high-dimensional space are updated periodically.

[0101] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1A device that provides the functions specified in a block or multiple blocks.

[0102] In a typical configuration, the device includes one or more processors (CPU), memory and bus. The device may also include input / output interface, network interface and the like.

[0103] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip. The memory is an example of a computer-readable medium.

[0104] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0105] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity 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, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0106] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0107] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for detecting abnormal behavior data, characterized in that: The method comprises: Constructing a target behavior sample based on the target behavior features in the target behavior data, wherein the target behavior features are used to characterize the optimized behavior features belonging to different dimensional labels; Calculate the multi-dimensional calculation results of the target behavior sample under different dimensional labels using the specified weights corresponding to the target behavior features; Determining whether the target behavior sample is abnormal based on the behavior region label of the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space; If so, abnormal behavior alarm information corresponding to the target behavior data is generated.

2. The method according to claim 1, characterized in that: The step of constructing a target behavior sample based on the target behavior features in the target behavior data includes: Extracting specified behavior features from the target behavior data based on preset dimension labels to obtain specified behavior features under different dimension labels; Optimizing the specified behavior features under different dimensional labels according to the optimization rules to obtain the target behavior features, wherein the optimization rules are used to represent the rules for merging or cleaning the specified behavior features under different dimensional labels based on the same frequency situation and / or the PDCA model; The target behavior sample is constructed based on the target behavior feature.

3. The method according to claim 1, characterized in that The calculating of the multi-dimensional calculation results of the target behavior example under different dimensional labels by using the specified weights corresponding to the target behavior features includes: Obtaining a specified weight corresponding to the target behavior feature; Calculate the weight result corresponding to each dimension label according to the specified weight corresponding to each target behavior feature; The multiple weight results are used as the multi-dimensional calculation results of the target behavior samples under different dimensional labels.

4. The method according to claim 3, characterized in that The obtaining of the specified weight corresponding to the target behavior feature includes: Acquire attribute information corresponding to the target behavior feature; Determining a weight level corresponding to the target behavior feature according to the attribute feature corresponding to the attribute information, wherein the attribute feature includes at least one of business sensitivity, business level, business operation frequency, and business operation time; The designated weight corresponding to the target behavior feature is configured according to the weight level.

5. The method according to claim 1, characterized in that The determining whether the target behavior sample is abnormal based on the behavior region label of the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space includes: Determining, in the high-dimensional space, the spatial position corresponding to the multidimensional array composed of the multidimensional calculation results; Determine a target high-dimensional cube corresponding to the target behavior sample based on the spatial position; If the behavior region label corresponding to the target high-dimensional cube is a normal behavior region label, then it is determined that the target behavior sample is not abnormal; If the behavior region label corresponding to the target high-dimensional cube is an abnormal behavior region label, it is determined that the target behavior sample is abnormal.

6. The method according to claim 1, characterized in that Before judging whether the target behavior sample is abnormal based on the behavior region label at the spatial position corresponding to the multi-dimensional calculation result in the high-dimensional space, the method further includes: Get the historical multi-dimensional calculation results corresponding to the historical behavior data within the specified period; Determine the historical spatial position corresponding to each of the historical multi-dimensional calculation results in the high-dimensional space; Dividing the high-dimensional space into high-dimensional cubes, and calculating the occurrence probability of the historical space position in each of the high-dimensional cubes; The abnormal behavior area label or the normal behavior area label is deployed for each of the high-dimensional cubes according to the comparison result between the occurrence probability and the preset probability threshold, wherein the preset probability threshold is determined according to the distribution corresponding to the historical spatial position.

7. The method according to claims 1-6, characterized in that: The method further comprises: The designated weights and the behavior region labels in the high-dimensional space are updated periodically.

8. An abnormal behavior data detection device, characterized in that: The device comprises: A construction unit, configured to construct a target behavior sample based on target behavior features in the target behavior data, wherein the target behavior features are used to characterize optimized behavior features belonging to different dimensional labels; A calculation unit, configured to calculate the multi-dimensional calculation results of the target behavior sample obtained by the construction unit under different dimensional labels by using the specified weights corresponding to the target behavior features; A judging unit, configured to judge whether the target behavior sample is abnormal according to the behavior region label of the spatial position corresponding to the multi-dimensional calculation result obtained by the calculating unit in the high-dimensional space; A generating unit is used to generate abnormal behavior alarm information corresponding to the target behavior data if the judging unit judges that the target behavior sample is abnormal.

9. A storage medium, characterized in that: The storage medium includes a stored program, wherein when the program is running, the device where the storage medium is located is controlled to execute the abnormal behavior data detection method according to any one of claims 1 to 7.

10. An electronic device, comprising at least one processor, and at least one memory and bus connected to the processor; wherein: The processor and the memory communicate with each other via the bus; The processor is used to call the program instructions in the memory to execute the abnormal behavior data detection method according to any one of claims 1 to 7.