Abnormal behavior user identification method, device, equipment and storage medium
By extracting the dimensional features of user behavior data, calculating feature distances and using an abnormal behavior recognition model, the problem of low recognition efficiency in the prior art is solved, and efficient identification and monitoring of abnormal behavior users within the enterprise is achieved.
Patent Information
- Application Number
- CN202111095865.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-09-18
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2041-09-18
AI Technical Summary
When identifying users of abnormal behavior within an enterprise, the recognition efficiency and reliability are low, and it is impossible to effectively identify users of unknown abnormal behavior.
By obtaining the behavioral data of the user querying business system data, extracting dimensional features, calculating the feature distance between users, and comparing the abnormal behavior recognition model of the feature distance input preset, to determine the abnormal behavior user. The method includes standardizing the processing of feature data and using density clustering algorithms to train the model to identify abnormal behavior.
It improves the recognition efficiency and accuracy of abnormal behavior users, enhances the ability to monitor user behavior, and effectively recognizes and prevents risky behaviors of sensitive information within the enterprise.
Smart Images

Figure CN113779568B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence, and in particular to a method, device, equipment and storage medium for identifying users with abnormal behavior. Background Art
[0002] There are two risks to enterprise data security. One is external risk, which is when external personnel enter the enterprise through various means, obtain enterprise data or destroy enterprise data, causing enterprise data loss or leakage, leading to enterprise data security incidents. The other is internal enterprise risk, which is when enterprise intranet users lose or leak enterprise data through various risky behaviors, causing enterprise data security incidents. And, in general, the number of enterprise data security incidents caused by the risky behaviors of enterprise intranet users is far greater than the number of enterprise data security incidents caused by external risks.
[0003] Under normal circumstances, the identification of users with abnormal behavior in the intranet can be achieved by setting up a blacklist. Specifically, the administrator adds the user names that need to be restricted to the blacklist. However, when identifying abnormal users in the intranet by setting up a blacklist, only abnormal users known to the current administrator can be identified. Users who are unknown to the current administrator and whose abnormal behavior cannot be found cannot be identified. Therefore, the identification efficiency and reliability of abnormal users are poor. Summary of the invention
[0004] The main purpose of the present invention is to solve the technical problem of poor efficiency in identifying users with abnormal behavior in the prior art.
[0005] A first aspect of the present invention provides a method for identifying users with abnormal behavior, and the method comprises: obtaining behavior data of each user querying business system data, and extracting dimensional features of each dimension in the behavior data; arbitrarily selecting a first target user from all the users, and calculating a characteristic distance between the first target user and other users based on the dimensional features of each user; inputting the characteristic distance into a preset abnormal behavior identification model, and comparing the characteristic distance with a preset radius; determining the abnormal behavior user based on the comparison result, wherein the abnormal behavior user is a user who abnormally queries sensitive information of the business system.
[0006] Optionally, in a first implementation method of the first aspect of the present invention, before arbitrarily selecting a first target user from all the users and calculating the feature distance between the first target user and other users based on the dimensional features of each user, it also includes: counting the number of features of the dimensional features corresponding to the users; calculating the mean and standard deviation of each of the dimensional features based on the number of features; and standardizing each of the dimensional features based on the mean and the standard deviation to obtain standard dimensional features.
[0007] Optionally, in a second implementation manner of the first aspect of the present invention, the arbitrarily selecting a first target user from all the users and calculating the characteristic distance between the first target user and the other users based on the dimensional characteristics of each user includes: arbitrarily selecting a first target user from all the users, and obtaining the characteristic value of each of the standard dimensional characteristics in the behavior data of each user; performing a perfect square difference operation on the characteristic values of the first target user and the other users belonging to the same dimension to obtain the characteristic difference value of each of the standard dimensional characteristics; performing a square root operation on the sum of each of the characteristic difference values to obtain the characteristic distance between the first target user and the other users.
[0008] Optionally, in a third implementation manner of the first aspect of the present invention, determining the abnormal behavior user based on the comparison result includes: extracting users whose characteristic distance from the first target user is not greater than a preset radius from all the users to obtain the first user; counting the number of users of the first users, and determining whether the number of users is less than a preset number threshold; if so, treating the first target user as an abnormal behavior user.
[0009] Optionally, in a fourth implementation manner of the first aspect of the present invention, when the number of users is not less than a preset number threshold, determining the abnormal behavior user based on the comparison result includes: marking the first target user as a core user, and assigning a cluster label to the first target user; eliminating the core user from all the users to obtain second users; calculating the characteristic distance of each of the second users, and comparing the characteristic distance of each of the second users with the radius; and determining the abnormal behavior user based on the comparison result.
[0010] Optionally, in the fifth implementation method of the first aspect of the present invention, before inputting the characteristic distance into the preset abnormal behavior recognition model and comparing the characteristic distance with a preset radius, it also includes: obtaining historical behavior data of each of the users; performing sample training on the historical behavior data using a preset density-based clustering algorithm to obtain abnormal behavior data samples; and performing model training on a preset abnormal behavior recognition tool using the abnormal behavior data samples to obtain an abnormal behavior recognition model.
[0011] The second aspect of the present invention proposes an abnormal behavior user identification device, which includes: an acquisition module, used to obtain the behavior data of each user querying business system data, and extract the dimensional features of each dimension in the behavior data; a calculation module, used to arbitrarily select a first target user from all the users, and calculate the characteristic distance between the first target user and other users according to the dimensional features of each user; a comparison module, used to input the characteristic distance into a preset abnormal behavior identification model, and compare the characteristic distance with a preset radius; a determination module, used to determine the abnormal behavior user according to the comparison result, wherein the abnormal behavior user is a user who abnormally queries the sensitive information of the business system.
[0012] Optionally, in a first implementation method of the second aspect of the present invention, the abnormal behavior user identification device also includes a standardization module, and the standardization module includes: a statistical unit, used to count the number of features of the dimensional features corresponding to the user; a calculation unit, used to calculate the mean and standard deviation of each of the dimensional features according to the number of features; and a standardization unit, used to standardize each of the dimensional features according to the mean and the standard deviation to obtain a standard dimensional feature.
[0013] Optionally, in a second implementation method of the second aspect of the present invention, the calculation module includes: a first acquisition unit, used to arbitrarily select a first target user from all the users, and obtain the feature values of each of the standard dimensional features in the behavior data of each of the users; a first operation unit, used to perform a complete square difference operation on the feature values of the first target user and other users belonging to the same dimension to obtain the feature difference values of each of the standard dimensional features; a second operation unit, used to perform a square root operation on the sum of each of the feature difference values to obtain the feature distance between the first target user and the other users.
[0014] Optionally, in a third implementation method of the second aspect of the present invention, the determination module is specifically used to: extract users from all the users whose characteristic distance with the first target user is not greater than a preset radius to obtain the first user; count the number of users of the first users, and determine whether the number of users is less than a preset number threshold; if the number of users is less than the preset number threshold, treating the first target user as an abnormal behavior user.
[0015] Optionally, in a fourth implementation manner of the second aspect of the present invention, when the number of users is not less than a preset number threshold, the determination module is further specifically used to: mark the first target user as a core user, and assign a cluster label to the first target user; eliminate the core user from all the users to obtain second users; calculate the characteristic distance of each of the second users, and compare the characteristic distance of each of the second users with the radius; and determine the abnormal behavior user based on the comparison result.
[0016] Optionally, in a fifth implementation of the second aspect of the present invention, the abnormal behavior user identification device also includes a model training module, and the model training module includes: a second acquisition unit, used to obtain historical behavior data of each of the users; a sample training unit, used to perform sample training on the historical behavior data using a preset density-based clustering algorithm to obtain abnormal behavior data samples; and a model training unit, used to perform model training on a preset abnormal behavior identification tool using the abnormal behavior data samples to obtain an abnormal behavior identification model.
[0017] A third aspect of the present invention provides an abnormal behavior user identification device, which includes: a memory and at least one processor, wherein a computer program is stored in the memory, and the memory and the at least one processor are interconnected via a line; the at least one processor calls the computer program in the memory so that the abnormal behavior user identification device performs the steps of the above-mentioned abnormal behavior user identification method.
[0018] A fourth aspect of the present invention provides a computer-readable storage medium, on which a computer program is stored, which, when executed on a computer, enables the computer to execute the steps of the above-mentioned abnormal behavior user identification method.
[0019] In the technical solution provided by the present invention, by obtaining the dimensional features of the behavioral data of each user querying the business system data, the target user is selected, the characteristic distance between the target user and other users is calculated according to the dimensional features, and the characteristic distance is compared with the preset radius based on the abnormal behavior recognition model, so as to determine the abnormal behavior user according to the comparison result. The present invention realizes the identification of abnormal behavior users, and identifies abnormal behavior users through abnormal behavior recognition models and characteristic distances, thereby improving the efficiency and accuracy of identification, thereby realizing the monitoring of user behavior. Compared with the traditional technology of only monitoring the daily behavior of users of each business system of the intranet through entity monitoring rules, the present invention can improve the efficiency, accuracy and comprehensiveness of identifying abnormal user behavior, assist in the handling of security operation risks, avoid the risk of internal intranet users inquiring sensitive information of various business systems within the enterprise, and play a positive role in internal user security, data leakage security and internal user account security. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 This is a schematic diagram of a first embodiment of a method for identifying users with abnormal behavior in an embodiment of the present invention;
[0021] Figure 2 Schematic diagram of a second embodiment of the abnormal behavior user identification method in an embodiment of the present invention;
[0022] Figure 3 Schematic diagram of a third embodiment of the abnormal behavior user identification method in an embodiment of the present invention;
[0023] Figure 4 Schematic diagram of a fourth embodiment of the abnormal behavior user identification method according to an embodiment of the present invention;
[0024] Figure 5 A schematic diagram of an embodiment of a device for identifying abnormal behavior users in an embodiment of the present invention;
[0025] Figure 6 It is a schematic diagram of another embodiment of the abnormal behavior user identification device in an embodiment of the present invention;
[0026] Figure 7 The figure is a schematic diagram of an embodiment of a device for identifying users with abnormal behavior in an embodiment of the present invention. DETAILED DESCRIPTION
[0027] The embodiments of the present invention provide a method, device, equipment and storage medium for identifying users with abnormal behavior, which selects a target user by acquiring the dimensional features of the behavior data of each user querying business system data, calculates the characteristic distance between the target user and other users based on the dimensional features, and compares the characteristic distance with a preset radius based on an abnormal behavior recognition model, thereby determining the abnormal behavior user based on the comparison result. The present invention realizes the identification of users with abnormal behavior, and identifies users with abnormal behavior through abnormal behavior recognition models and characteristic distances, thereby improving the efficiency and accuracy of identification, thereby realizing the monitoring of user behavior.
[0028] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0029] For ease of understanding, the specific contents of the embodiments of the present invention are described below. Figure 1 The first embodiment of the abnormal behavior user identification method in the embodiment of the present invention includes:
[0030] 101, obtaining behavior data of each user querying business system data, and extracting dimension features of each dimension in the behavior data;
[0031] The server obtains the behavior data of each user querying the business system data, and extracts the dimensional features of each dimension in the behavior data. In this embodiment, the dimensions of the dimensional features include the query time dimension, the query volume dimension, and the abnormal behavior dimension; the query time dimension refers to the time difference between the query time of business data and the abnormal state of business data, the time difference between the business data complaint time and the query time of business data, the time difference between the newly generated business data and the abnormal state of business data, etc.; the query volume dimension refers to the ratio of the user's personal query volume in the current month to the personal average query volume in the previous three months, the ratio of the personal query volume in the current month (such as the period from 9 to 12 o'clock and 14 to 18 o'clock) to the personal average query volume in the previous three months, the ratio of the personal query volume in the non-working period of the current month to the personal average query volume in the non-working period of the previous three months, the personal average monthly query volume and the monthly average query volume of the user's colleagues in the same organization and the same rank and position ratio, etc.; the abnormal behavior dimension refers to multiple inquiries into a single business data, the amount of inquiries into off-site business data, the proportion of inquiries into off-site business data, the amount and proportion of inquiries into business data in an abnormal state, etc.
[0032] In addition, embodiments of the present invention can acquire and process user behavior data based on artificial intelligence technology. Among them, artificial intelligence (AI) is the theory, method, technology and application system that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results. Basic artificial intelligence technologies generally include technologies such as sensors, dedicated artificial intelligence chips, cloud computing, distributed storage, big data processing technology, operation / interaction systems, mechatronics, etc. Artificial intelligence software technology mainly includes computer vision technology, robotics technology, biometrics technology, speech processing technology, natural language processing technology, and machine learning / deep learning.
[0033] 102, randomly selecting a target user from all users, and calculating the feature distance between the target user and other users based on the dimensional features of each user;
[0034] The server uses the user identification number as the primary key for all users' behavior data of querying business system data to construct a wide table, that is, the wide table records the behavior data of all users. The number of records in the wide table corresponds to the number of all users, and one record corresponds to the behavior data of one user.
[0035] The server randomly selects a target user from all users, and calculates the characteristic distance between the target user and other users based on the dimensional characteristics of each user. The number of target users is at least one; when calculating the characteristic distance between the target user and other users, the historical data of the target user in the wide table is first extracted and marked as the first data point, and the historical data of one user among the remaining users in the wide table is extracted and marked as the second data point; then the distance between the two data points is calculated. The characteristic distance between the target user and other users can be calculated by calculating the Euclidean distance, Manhattan distance, Mahalanobis distance, etc. between the target user and other users, which is not limited here.
[0036] The Euclidean metric (also known as the Euclidean distance) is a commonly used definition of distance, which refers to the true distance between two data points in m-dimensional space, or the natural length of a vector (i.e., the distance from the point to the origin). The Euclidean distance in two-dimensional and three-dimensional space is the actual distance between two data points. The Manhattan distance is the sum of the distances of the projections of the line segment formed by the two data points on the axis in the fixed rectangular coordinate system of the Euclidean space. The Mahalanobis distance can also be defined as the degree of difference between two random variables whose data points follow the same distribution and whose covariance matrix is Σ. If the covariance matrix is the identity matrix, then the Mahalanobis distance is simplified to the Euclidean distance. If the covariance matrix is a diagonal matrix, it can also be called the normalized Euclidean distance.
[0037] 103, inputting the characteristic distance into a preset abnormal behavior recognition model, and comparing the characteristic distance with a preset radius;
[0038] 104 , determining users with abnormal behavior according to the comparison result.
[0039] The server inputs the calculated characteristic distance into the preset abnormal behavior recognition model, and uses the abnormal behavior recognition model to compare the characteristic distance with the preset radius to obtain a comparison result. The preset abnormal behavior recognition model can identify users who abnormally query sensitive information of the business system, i.e., abnormal behavior users, based on the result of comparing the characteristic distance between the target user and other users with the preset radius.
[0040] When the characteristic distance between the target user and other users is not less than the radius, the number of corresponding other users is counted, and it is determined whether the number of users is less than the preset number threshold (min_sample). If the number of users is less than min_sample, the target user is regarded as an abnormal behavior user.
[0041] In the embodiment of the present invention, by obtaining the behavior data of each user querying the business system data, selecting the target user and calculating the characteristic distance between the target user and other users, the abnormal behavior user is identified according to the abnormal behavior identification model and the characteristic distance. The embodiment of the present invention realizes the identification of abnormal behavior users, and the abnormal behavior users are identified according to the behavior data of each user and the preset abnormal behavior identification model, thereby improving the efficiency and accuracy of identification.
[0042] See also Figure 2 , a second embodiment of the abnormal behavior user identification method in the embodiment of the present invention includes:
[0043] 201, obtaining behavior data of each user querying business system data, and extracting dimension features of each dimension in the behavior data;
[0044] The server obtains the behavior data of each user querying the business system data, and extracts the dimensional features of each dimension in the behavior data. In this embodiment, the dimensions of the dimensional features include the query time dimension, the query volume dimension, and the abnormal behavior dimension; the query time dimension refers to the time difference between the query time of business data and the abnormal state of business data, the time difference between the business data complaint time and the query time of business data, the time difference between the newly generated business data and the abnormal state of business data, etc.; the query volume dimension refers to the ratio of the user's personal query volume in the current month to the personal average query volume in the previous three months, the ratio of the personal query volume in the current month (such as the period from 9 to 12 o'clock and 14 to 18 o'clock) to the personal average query volume in the previous three months, the ratio of the personal query volume in the non-working period of the current month to the personal average query volume in the non-working period of the previous three months, the personal average monthly query volume and the monthly average query volume of the user's colleagues in the same organization and the same rank and position ratio, etc.; the abnormal behavior dimension refers to multiple inquiries into a single business data, the amount of inquiries into off-site business data, the proportion of inquiries into off-site business data, the amount and proportion of inquiries into business data in an abnormal state, etc.
[0045] 202, counting the number of dimensions of the dimensional features corresponding to the user;
[0046] 203, according to the number of dimensions, calculate the mean and standard deviation of the features of each dimension;
[0047] 204, standardize the features of each dimension according to the mean and standard deviation to obtain standard dimensional features;
[0048] In order to eliminate the impact of dimension, each dimensional feature needs to be standardized. The server counts the number of dimensions of the dimensional features corresponding to all users, and calculates the mean and standard deviation of each dimensional feature based on the number of dimensions. Among them, the dimensional features of each dimension correspond to different dimensions, and the number of dimensions of the dimensional features corresponding to the user is the number of dimensions.
[0049] The server standardizes each dimensional feature according to the calculated mean and standard deviation of each dimensional feature to obtain a standard dimensional feature. In this embodiment, the standard dimensional feature = (dimensional feature - mean) / standard deviation; for example, the standardized value xi1_new of the first dimensional feature of user xi = (xi1 - mean of the dimensional feature of the first dimension) / standard deviation of the dimensional feature of the first dimension). Among them, the mean of the dimensional feature of the first dimension of user xi = sum (feature value of the dimensional feature of the first dimension) / number of users.
[0050] The standard deviation formula is:
[0051]
[0052] For example, the standard deviation of the dimension feature of the first dimension is calculated, where N corresponds to the total number of users of all users, and xi is the dimension feature value of the i-th user corresponding to the dimension feature of the first dimension, and is the mean of the dimension feature values of the first dimension of all users.
[0053] 205, randomly selecting a target user from all users, and obtaining a feature value of each standard dimension feature in the behavior data of each user;
[0054] 206, performing a complete square difference operation on the feature values of the target user and other users belonging to the same dimension to obtain feature difference values of each standard dimension feature;
[0055] The server randomly selects at least one user from all users as the target user, obtains the behavior data of all users, and extracts the feature values of each standard dimension feature from the behavior data. The feature values of the target user and other users belonging to the same dimension are extracted, that is, under the same dimension, the feature values of the standard dimension features of all users in the dimension are extracted, and a complete square difference operation is performed based on the feature values to obtain the feature difference values of the standard dimension features of each user.
[0056] 207, performing a square root operation on the sum of each feature difference to obtain a feature distance between the target user and other users;
[0057] The server performs an arithmetic sum operation on each feature difference to obtain the sum of each feature difference, and performs a square root operation on the sum of each feature difference, that is, calculates the square root of each feature difference, and uses the calculated result as the feature distance between the target user and other users. The feature distance can be calculated using calculation rules such as Euclidean distance, Manhattan distance, and Mahalanobis distance. This embodiment uses Euclidean distance for explanation, and its formula is as follows:
[0058] D(xi, xj)=sqrt((xi1-xj1)^2+(xi2-xj2)^2+…+(xin-xjn)^2);
[0059] Among them, xi and xj refer to two users, namely the target user and another other user, xi1 refers to the feature value of the first standard dimension feature of user xi, xin refers to the feature value of the nth standard dimension feature of employee xi, and n refers to the number of features.
[0060] 208, inputting the characteristic distance into a preset abnormal behavior recognition model, and comparing the characteristic distance with a preset radius;
[0061] 209 , determining the user with abnormal behavior according to the comparison result.
[0062] In the embodiment of the present invention, steps 208-209 are consistent with steps 103-104 in the first embodiment of the above-mentioned abnormal behavior user identification method, and are not described in detail here.
[0063] In an embodiment of the present invention, the number of dimensions of the dimensional features corresponding to each user is counted, and the mean and standard deviation of each dimensional feature are calculated based on the number of dimensions, so as to standardize the dimensional features to obtain standard dimensional features, and the feature distance between users is calculated based on the standard dimensional features, thereby improving the calculation accuracy of the feature distance and thus improving the accuracy of subsequent identification of users with abnormal behavior.
[0064] See also Figure 3 The third embodiment of the abnormal behavior user identification method in the embodiment of the present invention includes:
[0065] 301, obtaining behavior data of each user querying business system data, and extracting dimension features of each dimension in the behavior data;
[0066] 302, randomly selecting a target user from all users, and calculating the feature distance between the target user and other users based on the dimensional features of each user;
[0067] 303, inputting the characteristic distance into a preset abnormal behavior recognition model, and comparing the characteristic distance with a preset radius;
[0068] 304, extracting users whose characteristic distances with the target user are not greater than a preset radius from all users to obtain a first user;
[0069] The server compares the characteristic distance between the target user and other users with a preset radius (eps) to determine whether the characteristic distance is not greater than the preset radius, thereby extracting users whose characteristic distance with the target user is less than the preset radius from all users to obtain the first user.
[0070] In this embodiment, the preset radius is pre-set by the server, and the specific value of the radius can be a default value of 0.5, or can be obtained in the following manner: eps can be obtained by k-distance calculation, and the best corresponding parameter is the obvious inflection point in the k-distance curve. The k-distance refers to a given k-domain parameter k, where k is generally 4. For each point in the data, the corresponding kth nearest neighbor distance is calculated, that is, the distance value of the nearest 4 points of each point is calculated, and the nearest neighbor distances corresponding to all points in the data set are sorted in descending order, and the distance value corresponding to the first lowest point is selected as eps.
[0071] 305, counting the number of users of the first user, and determining whether the number of users is less than a preset number threshold;
[0072] 306, if the number of users is less than the preset number threshold, the target user is regarded as an abnormal behavior user;
[0073] After obtaining the first user, the number of users included in the first user is counted, and the number of users is compared with a preset number threshold (min_sample), that is, it is determined whether the number of users is less than the preset number threshold. When the number of users of the first user is less than the number threshold, the target user is regarded as an abnormal behavior user, wherein the abnormal behavior user is a user who abnormally queries sensitive information of the business system.
[0074] In this embodiment, the preset number threshold (min_sample) is generally taken as min_sample>=dim+1, where dim is the feature dimension value, that is, the feature number of each dimensional feature in the behavior data.
[0075] 307, when the number of users is not less than a preset number threshold, marking the target user as a core user, and assigning a cluster label to the target user;
[0076] When the number of users of the first user is not less than the preset number threshold, the server marks the target user as a core user and determines whether the target user is assigned a cluster label. If the target user is not assigned a cluster label, a cluster label is created and the cluster label is assigned to the target user. One cluster label corresponds to one cluster, and a cluster is defined as the maximum set of densely connected users. Among them, the cluster label is only used for marking, and can be marked with a self-incrementing number (the initial value is set to 0, the next mark is 1, the next mark is 2, and so on). This embodiment does not set the cluster label.
[0077] 308, removing the core user from all users to obtain the second user;
[0078] 309, calculating a characteristic distance of each second user, and comparing the characteristic distance of each second user with the radius;
[0079] 310 , determining users with abnormal behavior according to the comparison result.
[0080] Core users are eliminated from all users, and the remaining users are taken as second users. The characteristic distance of each user in the second user is calculated, that is, a new target user is selected from the second user, and the characteristic distance between the new target user and other users is calculated, and the characteristic distance is compared with the preset radius, so as to extract users whose characteristic distance is not greater than the radius, and determine whether the number of users of these users is less than the preset number threshold. If the number of users is less than the preset number threshold, the new target user is taken as an abnormal behavior user. If the number of users is not less than the preset number threshold, the new target user is taken as a core user. If the new target user is not assigned a cluster label, the newly created cluster label is assigned to the new target user. Then visit all neighbors of the new target user (other users within the distance eps) in turn. If they have not been assigned a cluster label, the newly created cluster label is assigned to them. If they are core samples, their neighbors are visited in turn, and so on. The cluster gradually increases until there are no more core users within the eps distance of the cluster.
[0081] In the embodiment of the present invention, steps 301-303 are consistent with steps 101-103 in the first embodiment of the above-mentioned abnormal behavior user identification method, and are not described in detail here.
[0082] In the embodiment of the present invention, the abnormal behavior users are identified based on the characteristic distance between users and the number of users whose characteristic distance is not greater than a preset radius, thereby improving the efficiency and accuracy of the abnormal behavior users.
[0083] See also Figure 4 , a fourth embodiment of the abnormal behavior user identification method in the embodiment of the present invention includes:
[0084] 401, obtaining the behavior data of each user querying the business system data, and extracting the dimension features of each dimension in the behavior data;
[0085] 402, randomly selecting a target user from all users, and calculating the feature distance between the target user and other users based on the dimensional features of each user;
[0086] 403, obtain historical behavior data of each user;
[0087] 404, using a preset density-based clustering algorithm to perform sample training on the historical behavior data to obtain abnormal behavior data samples;
[0088] 405, using the abnormal behavior data sample to perform model training on a preset abnormal behavior recognition tool to obtain an abnormal behavior recognition model;
[0089] The server obtains the historical behavior data of all users, and uses a preset density-based clustering algorithm (DBSCAN algorithm) to perform sample training on the historical behavior data to obtain abnormal behavior data samples, and then performs model training on a preset abnormal behavior recognition tool based on the abnormal behavior data samples to obtain an abnormal behavior recognition model.
[0090] In this embodiment, the process of using a density-based clustering algorithm to train a model for a preset abnormal behavior identification tool is to arbitrarily select a user corresponding to the historical behavior data, use the user as the starting user, and then find all users whose distance to the user is less than or equal to the preset radius (eps). If the number of users within the radius (eps) from the starting user is less than the preset number threshold (min_samples), then the user is marked as an abnormal behavior user. If the number of users within the distance eps is greater than min_samples, then the user is marked as a core user and a new cluster label is assigned. Then visit all neighbors of the core user (within the distance eps). If they have not been assigned a cluster, then the new cluster label just created is assigned to them. If they are core users, then visit their neighbors in turn, and so on. The cluster gradually increases until there are no more core users within the eps distance of the cluster. Select another user that has not been visited and repeat the same process. Among them, if eps is set very small, it means that no user is a core user, which may cause all users to be marked as abnormal behavior users. If eps is set very large, it may cause all users to form a single cluster. Although you do not need to explicitly set the number of clusters, setting eps can implicitly control the number of eps found. It is sometimes easier to find a good value for eps by scaling all historical behavior data using StandarScaler or MinMaxScaler. This is because using scaling techniques will ensure that all historical behavior data has similar ranges for each dimension.
[0091] 406, inputting the characteristic distance into a preset abnormal behavior recognition model, and comparing the characteristic distance with a preset radius;
[0092] 407 , determine the user with abnormal behavior according to the comparison result.
[0093] In the embodiment of the present invention, steps 401-402 and steps 406-407 are consistent with steps 101-102 and steps 103-104 in the first embodiment of the above-mentioned abnormal behavior user identification method, and are not described in detail here.
[0094] In an embodiment of the present invention, sample training is performed based on the user's historical behavior data to obtain abnormal behavior data samples, and the abnormal behavior data samples are used to perform model training to obtain an abnormal behavior recognition model, so that the trained abnormal behavior recognition model can be used to identify abnormal behavior users, thereby improving the efficiency and accuracy of abnormal behavior user recognition.
[0095] The above describes the abnormal behavior user identification method in the embodiment of the present invention. The following describes the abnormal behavior user identification device in the embodiment of the present invention. Please refer to Figure 5 An embodiment of the abnormal behavior user identification device in the embodiment of the present invention includes:
[0096] The acquisition module 501 is used to acquire the behavior data of each user querying the business system data, and extract the dimension features of each dimension in the behavior data;
[0097] A calculation module 502 is used to arbitrarily select a first target user from all the users, and calculate a feature distance between the first target user and other users according to the dimensional features of each user;
[0098] A comparison module 503 is used to input the characteristic distance into a preset abnormal behavior recognition model and compare the characteristic distance with a preset radius;
[0099] The determination module 504 is used to determine an abnormal behavior user according to the comparison result, wherein the abnormal behavior user is a user who abnormally queries the sensitive information of the business system.
[0100] In the embodiment of the present invention, the dimensional features of the behavior data of each user querying the business system data are obtained through the abnormal behavior user identification device, the target user is selected, the characteristic distance between the target user and other users is calculated based on the dimensional features, and the characteristic distance is compared with the preset radius based on the abnormal behavior identification model, so as to determine the abnormal behavior user according to the comparison result. The present invention realizes the identification of abnormal behavior users, and identifies abnormal behavior users through abnormal behavior identification models and characteristic distances, improves the efficiency and accuracy of identification, and thus can realize the monitoring of user behavior. .
[0101] See also Figure 6 Another embodiment of the abnormal behavior user identification device in the embodiment of the present invention includes:
[0102] The acquisition module 501 is used to acquire the behavior data of each user querying the business system data, and extract the dimension features of each dimension in the behavior data;
[0103] A calculation module 502 is used to arbitrarily select a first target user from all the users, and calculate a feature distance between the first target user and other users according to the dimensional features of each user;
[0104] A comparison module 503 is used to input the characteristic distance into a preset abnormal behavior recognition model, and compare the characteristic distance with a preset radius;
[0105] The determination module 504 is used to determine an abnormal behavior user according to the comparison result, wherein the abnormal behavior user is a user who abnormally queries the sensitive information of the business system.
[0106] The abnormal behavior user identification device further includes a standardization module 505, and the standardization module 505 includes:
[0107] A statistical unit 5051 is used to count the number of dimensional features corresponding to the user;
[0108] A calculation unit 5052 is used to calculate the mean and standard deviation of each of the dimensional features according to the number of features;
[0109] The standardization unit 5053 is used to standardize each of the dimensional features according to the mean and the standard deviation to obtain a standard dimensional feature.
[0110] Wherein, the calculation module 502 includes:
[0111] A first acquisition unit 5021 is used to arbitrarily select a first target user from all the users, and obtain a feature value of each of the standard dimension features in the behavior data of each of the users;
[0112] The first operation unit 5022 is used to perform a complete square difference operation on the feature values of the first target user and other users belonging to the same dimension to obtain a feature difference value of each standard dimensional feature;
[0113] The second operation unit 5023 is used to perform a square root operation on the sum of the feature differences to obtain a feature distance between the first target user and the other users.
[0114] The determining module 504 is specifically used for:
[0115] Extracting users whose characteristic distance with the first target user is not greater than a preset radius from all the users to obtain a first user;
[0116] Counting the number of users of the first user, and determining whether the number of users is less than a preset number threshold;
[0117] If the number of users is less than a preset number threshold, the first target user is regarded as an abnormal behavior user.
[0118] When the number of users is not less than a preset number threshold, the determination module 504 is further specifically configured to:
[0119] marking the first target user as a core user, and assigning a cluster label to the first target user;
[0120] Eliminate the core user from all the users to obtain a second user;
[0121] calculating a characteristic distance of each of the second users, and comparing the characteristic distance of each of the second users with the radius;
[0122] The users with abnormal behaviors are identified based on the comparison results.
[0123] The abnormal behavior user identification device further includes a model training module 506, and the model training module 506 includes:
[0124] The second acquisition unit 5061 is used to acquire the historical behavior data of each user;
[0125] The sample training unit 5062 is used to perform sample training on the historical behavior data using a preset density-based clustering algorithm to obtain abnormal behavior data samples;
[0126] The model training unit 5063 is used to use the abnormal behavior data sample to perform model training on a preset abnormal behavior recognition tool to obtain an abnormal behavior recognition model.
[0127] In an embodiment of the present invention, the behavior data of each user querying business system data is obtained through an abnormal behavior user identification device, a target user is selected and the characteristic distance between the target user and other users is calculated, and the abnormal behavior user is identified based on the characteristic distance between the users, thereby improving the efficiency and accuracy of abnormal behavior user identification and realizing the monitoring of user abnormal behavior.
[0128] See also Figure 7 , an embodiment of the abnormal behavior user identification device in the embodiment of the present invention is described in detail below from the perspective of hardware processing.
[0129] Figure 77 is a schematic diagram of the structure of an abnormal behavior user identification device provided by an embodiment of the present invention. The abnormal behavior user identification device 700 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 710 (for example, one or more processors) and a memory 720, and one or more storage media 730 (for example, one or more mass storage devices) storing application programs 733 or data 732. Among them, the memory 720 and the storage medium 730 can be short-term storage or permanent storage. The program stored in the storage medium 730 may include one or more modules (not shown in the figure), and each module may include a series of instruction operations in the abnormal behavior user identification device 700. Furthermore, the processor 710 can be configured to communicate with the storage medium 730 to execute a series of instruction operations in the storage medium 730 on the abnormal behavior user identification device 700.
[0130] The abnormal behavior user identification device 700 may also include one or more power supplies 740, one or more wired or wireless network interfaces 750, one or more input and output interfaces 760, and or, one or more operating systems 731, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. Those skilled in the art will appreciate that Figure 7 The structure of the abnormal behavior user identification device shown does not constitute a limitation on the abnormal behavior user identification device, and may include more or fewer components than shown in the figure, or combine certain components, or arrange the components differently.
[0131] The server referred to in the present invention can be an independent server or a cloud server that provides basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security services, content delivery networks (CDNs), as well as big data and artificial intelligence platforms.
[0132] The blockchain referred to in this invention is a new application mode of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanism, encryption algorithm, etc. Blockchain is essentially a decentralized database, a string of data blocks generated by cryptographic methods. Each data block contains a batch of network transaction information, which is used to verify the validity of its information (anti-counterfeiting) and generate the next block. Blockchain can include the underlying blockchain platform, platform product service layer, and application service layer.
[0133] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the abnormal behavior user identification method.
[0134] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described devices and units can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0135] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including a number of instructions to enable a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program codes.
[0136] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for identifying users with abnormal behavior, It is characterized in that The abnormal behavior user identification method comprises: Obtaining behavior data of each user querying business system data, and extracting dimensional features of each dimension in the behavior data, wherein the dimensions of the dimensional features include a query time dimension, a query volume dimension, and an abnormal behavior dimension; Counting the number of features of the dimensional features corresponding to the user; According to the number of features, calculate the mean and standard deviation of each dimensional feature; According to the mean and the standard deviation, each of the dimensional features is standardized to obtain a standard dimensional feature; Randomly selecting a first target user from all the users, and calculating a feature distance between the first target user and other users according to the standard dimensional features of each user; The characteristic distance is input into a preset abnormal behavior recognition model, and the characteristic distance is compared with a preset radius, wherein the radius is obtained by calculating the distance values of the nearest K points of each point, and sorting the nearest neighbor distances corresponding to all points in the data set in descending order, and taking the distance value corresponding to the first lowest point as the radius; Determine an abnormal behavior user according to the comparison result, wherein the abnormal behavior user is a user who abnormally queries sensitive information of the business system; Determining the abnormal behavior user according to the comparison result includes: When the number of users within the radius of the first target user is not less than a preset number threshold, the first target user is removed from all the users, and the remaining users are used as second users, and the number threshold is greater than the number of features; calculating a characteristic distance of each of the second users, and comparing the characteristic distance of each of the second users with the radius; The users with abnormal behaviors are identified based on the comparison results.
2. According to the abnormal behavior user identification method of claim 1, the first target user is arbitrarily selected from all the users, and the feature distance between the first target user and other users is calculated based on the standard dimensional features of each user. include: Randomly selecting a first target user from all the users, and obtaining a feature value of each of the standard dimension features in the behavior data of each of the users; Performing a complete square difference operation on the feature values of the first target user and other users belonging to the same dimension to obtain feature difference values of the features of each standard dimension; Perform a square root operation on the sum of the feature differences to obtain the feature distance between the first target user and the other users.
3. The abnormal behavior user identification method according to claim 2, wherein the abnormal behavior user is determined based on the comparison result. include: Extracting users whose characteristic distance with the first target user is not greater than a preset radius from all the users to obtain a first user; Counting the number of users of the first user, and determining whether the number of users is less than a preset number threshold; If so, the first target user is regarded as an abnormal behavior user.
4. According to the abnormal behavior user identification method of claim 3, when the number of users is not less than a preset number threshold, the abnormal behavior user is determined based on the comparison result. include: The first target user is marked as a core user, and a cluster label is assigned to the first target user.
5. The abnormal behavior user identification method according to any one of claims 1 to 4, before inputting the characteristic distance into a preset abnormal behavior identification model and comparing the characteristic distance with a preset radius, further comprising: include: Obtaining historical behavior data of each of the users; Using a preset density-based clustering algorithm to perform sample training on the historical behavior data to obtain abnormal behavior data samples; The abnormal behavior data samples are used to perform model training on a preset abnormal behavior recognition tool to obtain an abnormal behavior recognition model.
6. A device for identifying abnormal user behavior, It is characterized in that The abnormal behavior user identification device comprises: An acquisition module is used to acquire the behavior data of each user querying the business system data, and extract the dimensional features of each dimension in the behavior data, wherein the dimensions of the dimensional features include the query time dimension, the query volume dimension, and the abnormal behavior dimension; A statistical unit, used to count the number of dimensional features corresponding to the user; A calculation unit, used to calculate the mean and standard deviation of each of the dimensional features according to the number of features; A standardization unit, used to perform standardization processing on each of the dimensional features according to the mean and the standard deviation to obtain a standard dimensional feature; A calculation module, used for arbitrarily selecting a first target user from all the users, and calculating a feature distance between the first target user and other users according to the dimensional features of each user; A comparison module is used to input the characteristic distance into a preset abnormal behavior recognition model, and compare the characteristic distance with a preset radius, wherein the radius is obtained by calculating the distance values of the nearest K points of each point, and sorting the nearest neighbor distances corresponding to all points in the data set in descending order, and taking the distance value corresponding to the first lowest point as the radius; A determination module, configured to determine an abnormal behavior user according to the comparison result, wherein the abnormal behavior user is a user who abnormally queries sensitive information of the business system; The determination module is specifically used to, when the number of users of the first target user within the radius is not less than a preset number threshold, eliminate the first target user from all the users and use the remaining users as second users, and the number threshold is greater than the number of features; calculate the feature distance of each second user, and compare the feature distance of each second user with the radius; and determine the abnormal behavior user according to the comparison result.
7. The abnormal behavior user identification device according to claim 6, It is characterized in that The calculation module includes: a first acquisition unit, used to arbitrarily select a first target user from all the users, and obtain the feature value of each standard dimensional feature in the behavior data of each user; a first operation unit, used to perform a complete square difference operation on the feature values of the first target user and other users belonging to the same dimension to obtain the feature difference value of each standard dimensional feature; a second operation unit, used to perform a square root operation on the sum of each feature difference value to obtain the feature distance between the first target user and the other users.
8. The abnormal behavior user identification device according to claim 6, It is characterized in that The determination module is specifically used to: extract users whose characteristic distance with the first target user is not greater than a preset radius from all the users to obtain the first user; count the number of users of the first user, and determine whether the number of users is less than a preset number threshold; if the number of users is less than the preset number threshold, the first target user is regarded as an abnormal behavior user.
9. A device for identifying abnormal user behavior, It is characterized in that The abnormal behavior user identification device comprises: A memory and at least one processor, wherein instructions are stored in the memory, and the memory and the at least one processor are interconnected via a line; The at least one processor calls the instructions in the memory to enable the abnormal behavior user identification device to perform the steps of the abnormal behavior user identification method according to any one of claims 1 to 5.
10. A computer-readable storage medium having instructions stored thereon, It is characterized in that When the instructions are executed by the processor, the steps of the abnormal behavior user identification method according to any one of claims 1 to 5 are implemented.
Citation Information
Patent Citations
Method, apparatus and system for detecting abnormal user behavior
CN108809745A
Information prompting method and device, storage medium and related equipment
CN111391784A
Abnormal user detection method and device based on clustering analysis, equipment and medium
CN111783875A
Retailer identification method, terminal and readable storage medium
CN111798274A
Systems and methods for multivariate anomaly detection in software monitoring
US20200351283A1
