Abnormal call detection method, device, electronic device and storage medium
Through the pre-trained decision tree model combined with call record parameters, it automatically determines whether the call record is an abnormal call, solving the problem of low manual review efficiency, and achieving accurate and efficient identification of abnormal calls.
Patent Information
- Application Number
- CN202210394119.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-04-14
- Publication Date
- 2024-11-26
- Estimated Expiration
- 2042-04-14
AI Technical Summary
In the prior art, when determining whether the call record is an abnormal call through manual review, it is less efficient and it is difficult to effectively identify and handle abnormal calls.
The pre-trained decision tree model is used to combine the various parameters of the call record, and through the scoring mechanism of the decision tree model, it automatically determines whether the call record is an abnormal call, and improves the recognition efficiency.
It realizes accurate and automatic identification of abnormal calls, improves the recognition efficiency of abnormal calls, and provides timely handling measures for abnormal calls.
Smart Images

Figure CN114979369B_ABST
Abstract
Description
Technical Field
[0001] The present application relates to the field of artificial intelligence technology, and more specifically, to an abnormal call detection method, device, electronic device and storage medium. Background Art
[0002] Currently, in the fields of service, sales, after-sales, etc., customer service personnel are usually required to communicate with customers through Internet calls.
[0003] There are usually many abnormal calls in the call records between customer service personnel and customers, such as no one answering the call for a long time, the call being hung up soon after answering, etc. In the related technology, the seat determines whether the call record is an abnormal call and reports it to the background server.
[0004] In the related art, whether there is an abnormal call is determined manually, which results in low efficiency in detecting abnormal calls. Summary of the invention
[0005] Embodiments of the present application provide an abnormal call detection method, device, electronic device, and storage medium.
[0006] In a first aspect, an embodiment of the present application provides an abnormal call detection method, the method comprising: obtaining a call parameter set of a call record, the call parameter set comprising at least one call parameter; scoring the call record based on the call parameter set and a decision tree model to obtain a score of the call record, the decision tree model comprising a plurality of node trees, the node trees being obtained by dividing the historical call records according to the call parameter set of the historical call records, the score of the call record representing the probability that the call record is an abnormal call; when the score of the call record is within a preset interval, determining that the call record is an abnormal call.
[0007] In the second aspect, the embodiment of the present application provides an abnormal call detection device, which includes: a parameter acquisition module, a scoring module and a detection module. The parameter acquisition module is used to obtain a call parameter set of a call record, and the call parameter set includes at least one call parameter; the scoring module is used to score the call record based on the call parameter set and the decision tree model to obtain the score of the call record, and the decision tree model includes multiple node trees, and the node trees are obtained by dividing the historical call records according to the call parameter set of the historical call records. The score of the call record represents the probability that the call record is an abnormal call; the detection module is used to determine that the call record is an abnormal call when the score of the call record is within a preset range.
[0008] In a third aspect, an embodiment of the present application provides a training method for a decision tree model, wherein the decision tree model includes multiple node trees, and the method includes: obtaining a historical call record data set and a call parameter data set of the historical call record data set, wherein the historical call record data set includes multiple historical call records; iteratively training the decision tree model until the difference judgment rate is less than a preset threshold, and a trained decision tree model is obtained, specifically including: determining the node corresponding to the target historical call record as the root node of the node tree, wherein the target historical call record is any historical call record in the historical call record data set; determining the node of each layer of the node tree according to the call parameter data set of the call record data set. Hyperplane division basis; determining the position information of nodes corresponding to other historical call records in the node tree according to the hyperplane division basis of nodes at each layer of the node tree and the root node of the node tree, where other historical call records are historical call records other than the target historical call records in the historical call record data set; obtaining the scores of other historical call records based on the position information of nodes corresponding to other historical call records in the node tree; determining whether other historical call records are abnormal calls based on the scores of other historical call records; obtaining the error rate, which is the ratio of the number of errors in judging other historical call records as abnormal calls to the total number of other historical call records.
[0009] In a fourth aspect, an embodiment of the present application provides a training device for a decision tree model, the device comprising: a data acquisition module, used to acquire a historical call record data set and a call parameter data set of the historical call record data set, the historical call record data set comprising a plurality of historical call records; a processing module, used to iteratively train the decision tree model until the discrepancy rate is less than a preset threshold, thereby obtaining a trained decision tree model. A processing module is specifically used to determine the node corresponding to the target historical call record as the root node of the node tree, where the target historical call record is any historical call record in the historical call record data set; determine the hyperplane division basis of each layer of nodes in the node tree according to the call parameter data set of the call record data set; determine the position information of the nodes corresponding to other historical call records in the node tree according to the hyperplane division basis of each layer of nodes in the node tree and the root node of the node tree, where the other historical call records are historical call records in the historical call record data set other than the target historical call record; obtain the scores of other historical call records based on the position information of the nodes corresponding to other historical call records in the node tree; determine whether other historical call records are abnormal calls based on the scores of other historical call records; and obtain the error judgment rate, which is the ratio between the number of errors in judging other historical call records as abnormal calls and the total number of other historical call records.
[0010] In a fifth aspect, an embodiment of the present application provides an electronic device, which includes a processor and a memory, wherein the memory stores computer program instructions, and the computer program instructions are called by the processor to execute the abnormal call detection method as described above, or the training method of the decision tree model.
[0011] In a sixth aspect, an embodiment of the present application provides a computer-readable storage medium, characterized in that a program code is stored in the computer-readable storage medium, and the program code is called by a processor to execute the abnormal call detection method as described above, or the training method of a decision tree model.
[0012] In a seventh aspect, an embodiment of the present application provides a computer program product, which, when executed, is used to implement the abnormal call detection method as described above, or the training method of a decision tree model.
[0013] The embodiment of the present application provides an abnormal call detection method, which quantifies the probability that a call record is an abnormal call through a pre-trained decision tree model and various parameters of the call record. Since the decision tree model is a model that classifies historical call records as instances, when the decision tree model is trained through various parameters of a large number of historical call records, the decision tree model can accurately determine whether it is an abnormal call based on the various parameters of the call record. Therefore, it can realize automatic identification of abnormal calls on the basis of ensuring the recognition accuracy of abnormal call records, improve the recognition efficiency of abnormal calls, so that corresponding processing measures can be provided for abnormal calls in a timely manner later. BRIEF DESCRIPTION OF THE DRAWINGS
[0014] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For those skilled in the art, other drawings can be obtained based on these drawings without creative work.
[0015] Figure 1 It is a schematic diagram of the implementation environment provided by the embodiment of the present application.
[0016] Figure 2 This is a flowchart of an abnormal call detection method provided by an embodiment of the present application.
[0017] Figure 3 This is a flowchart of an abnormal call detection method provided by an embodiment of the present application.
[0018] Figure 4 This is a flowchart for obtaining call record scores provided by an embodiment of the present application.
[0019] Figure 5It is a schematic diagram of a node tree provided by an embodiment of the present application.
[0020] Figure 6 It is a flowchart of a method for training a decision tree model provided by an embodiment of the present application.
[0021] Figure 7 It is a block diagram of an abnormal call detection device provided by an embodiment of the present application.
[0022] Figure 8 It is a block diagram of a device for constructing a decision tree model provided by an embodiment of the present application.
[0023] Fig. 9 It is a block diagram of an electronic device provided by an embodiment of the present application.
[0024] Fig.10 It is a block diagram of a computer-readable storage medium provided by an embodiment of the present application. DETAILED DESCRIPTION
[0025] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be understood as limiting the present application.
[0026] In order to enable those skilled in the art to better understand the solutions of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present application.
[0027] The inventor found that there is a problem of low efficiency when judging whether a call record is an abnormal call through manual review. Based on this, the inventor designed a solution to automatically judge whether a call record is an abnormal call record. Specifically, the probability of a call record being an abnormal call is quantified through a pre-trained decision tree model and various parameters of the call record. Since the decision tree model is a model that classifies historical call records as examples, when the decision tree model is trained through various parameters of a large number of historical call records, the decision tree model can accurately judge whether it is an abnormal call based on the various parameters of the call record. Therefore, it can realize the automatic identification of abnormal calls on the basis of ensuring the recognition accuracy of abnormal call records, improve the recognition efficiency of abnormal calls, so that corresponding treatment measures can be provided for abnormal calls in a timely manner in the future.
[0028] Figure 1 1 is a schematic diagram of an implementation environment provided by an embodiment of the present application. The implementation environment includes an Internet calling system, and the Internet calling system includes a plurality of first terminals 110 , a server 120 , and a second terminal 130 .
[0029] The first terminal 110 provides an Internet phone service, and a customer service staff (hereinafter referred to as an agent) communicates with a customer through the Internet phone service provided by the first terminal 110. The first terminal 110 may be a fixed telephone, a personal computer, a mobile terminal, etc. For example, if the first terminal 110 is a personal computer, the first terminal 110 may be installed with an application that provides an Internet phone service, and the agent conducts Internet phone calls with the customer through the above application.
[0030] It should be noted that when the agent communicates with the customer through the first terminal 110, the first terminal 110 will record the call parameters of the call record, including but not limited to: ringing duration, call duration, called number, call recording, hang-up direction, gateway signaling value, etc. Ringing duration refers to the length of time between the agent initiating the Internet call and the customer answering the Internet call. Call duration refers to the length of time between the Internet call being connected and hanging up. The called number refers to the number corresponding to the customer's SIM card. Call recording refers to the recording file obtained by recording the Internet call, which usually includes the agent's speech content, the customer's speech content, environmental voice signals, etc. Hang-up direction refers to the party that hangs up the Internet call, which can be the agent or the customer. The gateway signaling value represents the network quality during the Internet call.
[0031] The second terminal 130 provides a call service, and the customer communicates with the agent through the call service provided by the second terminal 130. In some embodiments, the second terminal 130 is installed with a call application, and the customer communicates with the agent through the call service provided by the call application. The second terminal 130 can be a smart phone, a landline phone, etc.
[0032] The server 120 is used to determine whether the call record is an abnormal call. In an embodiment of the present application, the server 120 is a pre-trained decision tree model, which scores the call record through the decision tree model and the call parameter set of the call record, and determines whether the call record is an abnormal call according to the range of the score. Among them, the algorithm used by the server to train the decision tree model can be an isolation forest algorithm. Optionally, the server 120 can be a background server corresponding to a specified application installed by the first terminal 110. The server 120 can be a single server, a server cluster consisting of multiple servers, or a cloud computing service center.
[0033] The first terminal 110 and the server 120 may establish a communication connection via a wired network or a wireless network. The first terminal 110 and the second terminal 130 may establish a communication connection via a wireless network.
[0034] The embodiment of the present application provides an abnormal call detection method, which quantifies the probability that a call record is an abnormal call through a pre-trained decision tree model and various parameters of the call record. Since the decision tree model is a model that classifies historical call records as instances, when the decision tree model is trained through various parameters of a large number of historical call records, the decision tree model can accurately determine whether it is an abnormal call based on the various parameters of the call record. Therefore, it can realize automatic identification of abnormal calls on the basis of ensuring the recognition accuracy of abnormal call records, improve the recognition efficiency of abnormal calls, so that corresponding processing measures can be provided for abnormal calls in a timely manner later.
[0035] refer to Figure 2 , which shows a flowchart of abnormal call detection provided by an embodiment of the present application. The server inputs the call record into a pre-trained decision tree model, obtains the score of the call record according to the position of each node tree of the target node representing the call record in the decision tree model, and then determines whether it is an abnormal call based on the score.
[0036] The technical solution provided by the embodiment of the present application can be applied in scenarios where there is a need for voice communication with users, such as telephone sales, telephone reminders, after-sales follow-up visits, etc. Taking telephone reminders as an example, after the agent communicates with the customer by phone, the technical solution provided by the embodiment of the present application is used to determine whether the call record is an abnormal call, and then it can be timely checked that the overdue user refuses to answer the call, the overdue user is unwilling to communicate, etc., so as to further determine the overdue user's willingness to return resources, and formulate corresponding processing measures based on the above willingness to return resources.
[0037] Figure 3 is a flow chart of an abnormal call detection method provided by an embodiment of the present application. The execution subject of the method may be Figure 1 The server in the implementation environment, the method comprises the following steps:
[0038] Step 301, obtaining a call parameter set of a call record.
[0039] The call parameter set includes at least one call parameter, which includes but is not limited to: call time, call duration, ringing duration, hang-up direction, call recording, called number, gateway signaling value, etc.
[0040] In some embodiments, the server obtains the call parameter set of the call record from the first terminal. During the process of making an Internet call with the second terminal, the first terminal usually records the call parameter set of the call record and reports the call parameter set to the server after the call ends. That is, the server can realize real-time detection of the call record. In other embodiments, the server obtains the call parameter set of the call record from the database server. After the call ends, the first terminal reports the call parameter set of the call record to the database server, and the database server classifies and stores the call records, such as classifying them according to the call time, classifying them according to the number of the second terminal, etc.
[0041] It should be noted that, when the server obtains the call parameter set of the call record from the database, the server may perform the above step 301 during a preset abnormal call detection period. For example, the above preset abnormal call detection period may be 22:00-24:00.
[0042] In some embodiments, after step 201, the server first detects whether the call record meets the invalid call condition. If the call record does not meet the invalid call condition, the subsequent scoring step is performed. Among them, the invalid call condition includes at least one of the following: the ringing duration is less than the specified duration, and the call period is in an abnormal time period. The specified duration and the abnormal time period are set according to actual needs. For example, the specified duration is 1 second, and the abnormal event period is [23:00-7:00]. When the ringing duration is too short, or the call period is at night, there is a high probability that the call record is triggered unconsciously by the user. Through the above method, the above unconsciously triggered call records can be excluded, avoiding scoring them to obtain scores that have no reference value, thereby saving server processing resources.
[0043] Step 302: Score the call record based on the call parameter set of the call record and the pre-trained decision tree model to obtain the score of the call record.
[0044] The decision tree model includes multiple node trees. The node tree is obtained by dividing the historical call records according to the call parameter set of the historical call records. The node tree includes a root node and at least one leaf node, and the root node and the leaf node both represent the historical call records. In some embodiments, the electronic device trains multiple historical call records based on the isolation forest algorithm to obtain a decision tree model. The isolation forest is an algorithm for detecting outliers in the category of unsupervised learning, which determines whether the sample is isolated by the density of the sample. The specific training process will be described in the following embodiments.
[0045] The score of a call record represents the probability that the call record is an abnormal call. The closer the score of a call record is to 1, the greater the probability that the call record is an abnormal call record. The closer the score of a call record is to zero, the smaller the probability that the call record is an abnormal call record.
[0046] Step 303: When the score of the call record is within a preset range, determine that the call record is an abnormal call.
[0047] The preset interval is set based on experiments or experience, and the present application embodiment does not limit this. Exemplarily, the preset interval is [0.5, 1]. In one example, the score of call record A is 0.62. In the above preset interval, call record A is determined to be an abnormal call.
[0048] In summary, the technical solution provided by the embodiment of the present application quantifies the probability that a call record is an abnormal call through a pre-trained decision tree model and various parameters of the call record. Since the decision tree model is a model that classifies historical call records as instances, when the decision tree model is trained through various parameters of a large number of historical call records, the decision tree model can accurately determine whether it is an abnormal call based on the various parameters of the call record. Therefore, it can realize automatic identification of abnormal calls on the basis of ensuring the recognition accuracy of abnormal call records, improve the recognition efficiency of abnormal calls, so that corresponding processing measures can be provided for abnormal calls in a timely manner in the future.
[0049] Figure 4 A flowchart of obtaining a call record score provided by an embodiment of the present application is shown. In some embodiments, Figure 3 Step 302 in the above may specifically include Figure 4 Step 3021 to step 3024 in step 302. Step 302 specifically includes:
[0050] Step 3021, mapping processing is performed on each call parameter in the call parameter set to obtain the standard value corresponding to each call parameter.
[0051] Mapping processing refers to mapping the call parameters to standard values according to the specified mapping method. The standard value corresponding to the call parameter is the first preset value or the second preset value. The first preset value is less than the second preset value. The first preset value and the second preset value are both pre-set. In a specific example, the first preset value is 1 and the second preset value is 0. Different call parameters correspond to different mapping methods. The mapping methods corresponding to each parameter are described below.
[0052] When the call parameters include the average ringing duration, if the average ringing duration is greater than the first preset duration, the server maps it to a first value, and if the average ringing duration is less than the first preset duration, the server maps it to a second value. The first preset duration, the first value, and the second value are all set based on experiments or experience. Exemplarily, the first preset duration is 3000 milliseconds, the first value is 1, and the second value is 0. That is, the specified mapping method corresponding to the average ringing duration can be expressed as: Average ringing duration (milliseconds): {[0-3000], standard value 0; [3000-infinity], standard value 1}.
[0053] When the call parameters include call duration, if the call duration is greater than the second preset duration, the server maps it to a first value, and if the call duration is less than the second preset duration, the server maps it to a second value. The second preset duration is set based on experiments or experience. Exemplarily, the second preset duration is 5 seconds. That is, the specified mapping method corresponding to the call duration can be expressed as: call duration (seconds): {[0-5], standard value 0; [5-infinity], standard value 1}.
[0054] When the call parameter includes the on-hook direction, if the on-hook direction is the client side, the server maps it to the first value, and if the on-hook direction is the agent side, the server maps it to the second value. That is, the designated mapping method corresponding to the on-hook direction can be expressed as: on-hook direction: {agent side, standard value 0; client side, standard value 1}.
[0055] When the call parameter includes the number of calls, if the number of calls is greater than the first preset value, the server maps it to a first value, and if the number of calls is less than or equal to the first preset value, the server maps it to a second value. The first preset value is set based on experiments or experience. Exemplarily, the first preset value is 1. That is, the specified mapping method corresponding to the number of calls can be expressed as: number of calls: {1, standard value 0; [1-infinity], standard value 1}.
[0056] When the call parameter includes a call period, if the call period belongs to the first period, the server maps it to the first value, and if the call period belongs to the second period, the server maps it to the second value. The first period and the second period are set according to actual needs. The first period is 9:00-12:00, and the second period is 1:30-9:00. That is, the specified mapping method corresponding to the call period can be expressed as: call period: {[9:00-12:00], standard value 0; [1:30-9:00], standard value 1}.
[0057] When the call parameter includes whether to say that the signal is bad after replying to something, if so, the server maps it to the first value, and if not, the server maps it to the second value.
[0058] When the network parameter includes a gateway signaling value, if the gateway signaling value is a specified character, the server maps it to a first value, and if it is not a specified character, the server maps it to a second value. The specified character is set according to actual needs, and illustratively, the specified character is 200OK.
[0059] The call parameters may also include other parameters, and the designated mapping method of other parameters in the embodiment of the present application is not limited. Through the above method, each call parameter is mapped into a standard value for subsequent unified operation.
[0060] It should be noted that the same call parameter may also correspond to different designated mapping methods. In some embodiments, the server determines the overdue duration of the user corresponding to the call record, and determines the designated mapping method of the call parameter based on the overdue duration. Taking the call parameter as the number of calls as an example, the longer the overdue duration, the greater the first preset duration, and the shorter the overdue duration, the smaller the first preset duration. The following Table-1 shows the corresponding relationship between the overdue duration and the designated mapping relationship when the call parameter is the average ringing duration.
[0061]
[0062] Table-1
[0063] Among them, M1 refers to the overdue period within one quarter, M2 refers to the overdue period between one quarter and two quarters, M3 refers to the overdue period between two quarters and three quarters, and M4 refers to the overdue period between three quarters and four quarters.
[0064] Taking the call parameter as the average ringing duration as an example, the longer the overdue duration is, the smaller the first preset duration is, and the shorter the overdue duration is, the larger the first preset duration is. The following Table-2 shows the correspondence between the overdue duration and the specified mapping relationship when the call parameter is the average ringing duration.
[0065]
[0066]
[0067] Table-2
[0068] Taking the call parameter as the call duration as an example, the longer the overdue duration is, the larger the second preset duration is, and the shorter the overdue duration is, the smaller the second preset duration is. The following Table-3 shows the correspondence between the overdue duration and the specified mapping relationship when the call parameter is the call duration.
[0069]
[0070] Table-3
[0071] Through the above method, the call parameters are mapped into corresponding standard values in combination with the overdue duration corresponding to the user corresponding to the call record, so that the division of the standard values can be more in line with the psychological state of the user with overdue bills, thereby making the detection of abnormal calls more accurate.
[0072] Step 3022: Determine the location information of the target node in each node tree based on the standard values corresponding to each call parameter.
[0073] The target node is used to represent the call record. In the embodiment of the present application, the call record is represented by the target node so that it can be processed by the pre-trained decision tree model. The target node is usually a leaf node.
[0074] The node tree includes a root node and a leaf node. The root node refers to the topmost node of the node tree, and the leaf node refers to the node in the node tree other than the root node. In an embodiment of the present application, the node tree includes multiple layers, each layer includes one or more nodes, and the one or more nodes are distributed in the layer according to the hyperplane division basis of the layer. The hyperplane division basis of each layer is a call parameter in the call parameter set, and the hyperplane division basis of different layers is different. In a specific example, the node tree includes four layers, and the hyperplane division basis from the first layer to the fourth layer is respectively call duration, ringing duration, call time, and hang-up direction.
[0075] Optionally, the node tree includes n layers of nodes. For each node tree, step 3022 may be implemented as follows:
[0076] Obtain the demarcation value of the hyperplane division basis of the i+1th layer node of the node tree, wherein the target call parameter in the call record is used as the hyperplane division basis of the i+1th layer node of the node tree, the target call parameter is any call parameter in the call parameter set of the call record, n is an integer greater than 1, i is a positive integer, and i is less than or equal to n-1; if the standard value corresponding to the target call parameter is greater than or equal to the demarcation value of the hyperplane division basis of the i+1th layer, then the position information of the target node in the node tree is determined to be the first branch of the i-th layer node, and the first branch of the i-th layer node is determined to be the upper node of the i+2th layer node; if the standard value corresponding to the target call parameter is less than the demarcation value of the hyperplane division basis of the i+1th layer, then the position information of the target node in the node tree is determined to be the second branch of the i-th layer node, and the second branch of the i-th layer node is determined to be the upper node of the i+2th layer node. Wherein, the values of i are 1 to n-1.
[0077] The cutoff value for the hyperplane division is preset, which is usually greater than the first preset value and less than the second preset value. Exemplarily, the cutoff value for the hyperplane division is 0.5. The first branch and the second branch are different. In a specific example, the first branch of the i-th layer node is the left branch of the i-th layer node, and the second branch of the i-th layer node is the right branch of the i-th layer node.
[0078] See also Figure 5 , Figure 5 The schematic diagram of the node tree provided by an embodiment of the present application is shown. The node tree includes four layers, and the hyperplane division basis of each layer to the fourth layer is ringing duration, call duration, call time, and hang-up direction. The dividing value of the hyperplane division basis of each layer is 0.5. Figure 5 In the diagram, a dotted line is used to represent the first branch on the left side of the dotted line and a second branch on the right side of the dotted line. The standard value corresponding to the ringing duration of the target node is 1, the standard value corresponding to the call duration is 1, the standard value corresponding to the call time is 0, and the standard value corresponding to the hang-up direction is 0. For the first layer, since the standard value corresponding to the ringing duration of the call record is greater than the demarcation value, the target node is on the right side of the dotted line of the root node; for the second layer, since the standard value corresponding to the call duration of the call record is greater than the demarcation value, the target node is on the right side of node 2 (the first branch of the first layer); for the third layer, since the standard value corresponding to the call time of the call record is less than the demarcation value, the target node is on the left side of node 4 (the first branch of the second layer); for the fourth layer, since the standard value corresponding to the hang-up of the call record is less than the demarcation value, the target node is on the left side of node 5 (the second branch of the third layer).
[0079] Step 3023: Determine the score of the target node in each node tree based on the position information of the target node in each node tree.
[0080] In some embodiments, step 3023 may include the following sub-steps. The following steps are performed for each node tree. The number of nodes between the target node and the root node of the node tree is obtained based on the position information of the target node in the node tree; the average number of nodes of the node tree is obtained, and the score of the target node in the node tree is determined based on the number of nodes between the target node and the root node of the node tree and the average number of nodes of the node tree.
[0081] In some embodiments, the server determines the serial number of the layer where the target node is located in the node tree based on the position information of the target node in the node tree, and determines the difference between the serial number and 1 as the number of nodes between the target node and the root node of the node tree. Figure 5 , the target node is at the fourth level of the node tree, then there are three nodes between the target node and the root node of the node tree, namely node 2, node 4 and node 5.
[0082] The average number of nodes in a node tree refers to the average number of nodes between each node in the node tree and the root node.
[0083] There is a negative correlation between the score of the target node in the node tree and the number of nodes between the target node and the root node of the node tree. That is, the fewer the number of nodes between the target node and the root node of the node tree, the greater the score of the target node in the node tree; the more the number of nodes between the target node and the root node of the node tree, the smaller the score of the target node in the node tree.
[0084] There is a negative correlation between the score of the target node under the node tree and the average number of nodes in the node tree. That is, the larger the average number of nodes in the node tree, the smaller the score of the target node under the node tree, and the smaller the average number of nodes in the node tree, the larger the score of the target node under the node tree. In some embodiments, the server calculates the score of the target node under the node tree using the following formula.
[0085] S(x,n)=2((e(h(x)) / c(n)).
[0086] Wherein, h(x) is the number of nodes between the target node and the root node, c(n) represents the average number of nodes in the node tree, n is the number of nodes included in the node tree, and x represents the target node. c(n)=2H(n-1)*(2(n-1) / n), H(n) is a constant, and illustratively, H(n) is 0.577.
[0087] Step 3024, based on the score of the target node in each node tree, obtain the score of the call record.
[0088] Optionally, the server determines the average of the scores of the target node in each node tree as the score of the call record. In some embodiments, step 303 may specifically include:
[0089] Step 3031: weight the scores of the call records based on the user attributes corresponding to the call records to obtain the weighted scores.
[0090] The user attribute includes at least one of the following: the overdue duration corresponding to the call record, and the region to which the number corresponding to the call record belongs.
[0091] In the case where the user attributes include the overdue duration corresponding to the call record, the server first determines the weighting coefficient corresponding to the overdue duration corresponding to the call record, and then multiplies the weighting coefficient by the score of the call record to obtain the weighted score. Among them, there is a positive correlation between the overdue duration and the weighting coefficient, that is, the longer the overdue duration, the larger the weighting coefficient. The following Table-4 exemplifies the corresponding relationship between the overdue duration and the weighting coefficient.
[0092] Overdue duration Weighting coefficient M1 0.80 M2 0.85 M3 0.90 M4 0.95
[0093] Table-4
[0094] In the case where the user attributes include the region to which the number corresponding to the call record belongs, the server first determines whether the region is a high-risk financial region. If the region is a high-risk financial region, the weighting coefficient is set to the first coefficient, and then the score of the call record is multiplied by the first coefficient to obtain the weighted score. If the region is not a high-risk financial region, no weighting is required. Among them, the high-risk financial region can be obtained through manual evaluation.
[0095] Through the above method, the scores of call records are weighted based on user attributes, so that the actual scores of call records are associated with user attributes, which is more in line with the actual situation and makes the scores of call records more accurate.
[0096] Step 3032: When the weighted score is within a preset range, the call record is determined to be an abnormal call.
[0097] It should be noted that after step 3032, the electronic device can also obtain a verification result. If the verification result indicates that the result of the call record being an abnormal call record is inaccurate, the hyperplane division basis of at least one node tree is adjusted. The verification result is used to characterize whether the result of determining that the call record is an abnormal call record is accurate. The verification result can be obtained by manual review. In the above manner, the accuracy of the node tree in the decision tree model can be improved, thereby improving the accuracy of determining whether the call record is an abnormal call record.
[0098] The technical solution provided by the embodiment of the present application determines the position of the target node representing the call record in each node tree in the decision tree model, calculates the score based on the position of the target node in each node tree, and determines the score of the call record based on the score of the target node under each node tree. Since the decision tree model is a model that classifies historical call records as instances, when the decision tree model is trained by various parameters of a large number of historical call records, the decision tree model can accurately determine whether it is an abnormal call based on the various parameters of the call record, thereby ensuring the recognition accuracy of abnormal call records. The score of the call record is also weighted based on the user attributes, so that the actual score of the call record is associated with the user attributes, which is more in line with the actual situation and makes the score of the call record more accurate.
[0099] See also Figure 6 , which shows a flow chart of a method for training a decision tree model provided by an embodiment of the present application. The method comprises the following steps:
[0100] Step 601: Acquire a historical call record data set and a call parameter data set of the historical call record data set.
[0101] The historical call record data set includes multiple historical call records. The server can obtain the historical call record data set from the call records reported by each first terminal, or can obtain the historical call record set from a database. Each historical call record in the historical call record data set corresponds to a call parameter set.
[0102] Step 602, iteratively train the decision tree model until the discrepancy rate is less than a preset threshold, thereby obtaining a trained decision tree model.
[0103] The misjudgment rate is the ratio of the number of errors in judging other historical call records as abnormal calls to the total number of other historical call records. The server determines whether other historical call records are abnormal calls based on the scores of other historical call records, and then compares the judgment results with the manual annotation results to determine whether this judgment is a success or failure. Finally, the misjudgment rate is determined based on the ratio between the number of failed judgments and the total number of historical call records.
[0104] The preset threshold is actually set according to the accuracy requirement. The higher the accuracy requirement, the smaller the preset threshold. For example, the preset threshold is one thousandth.
[0105] In some embodiments, step 602 may include the following sub-steps.
[0106] Step 602a: determine the node corresponding to the target historical call record as the root node of the node tree.
[0107] The target historical call record is any historical call record in the historical call record data set. For each node tree, a historical call record is randomly selected from the historical call record data set as a root node, and the historical call records represented by the root nodes of different node trees are different.
[0108] Step 602b, determining a hyperplane division basis for nodes at each layer of the node tree according to the call parameter data set of the call record data set.
[0109] The basis for the hyperplane division of each layer of the node tree is any call parameter in the call parameter data set. The basis for the hyperplane division of different layers is different. It should be noted that for different node trees, the basis for the hyperplane division at the same layer is usually different. For example, the basis for the hyperplane division of the first layer of node tree A is the call duration, and the basis for the hyperplane division of the first layer of node tree B is the ringing duration.
[0110] Step 602c, determining the position information of the nodes corresponding to other historical call records in the node tree according to the hyperplane division basis of the nodes at each layer of the node tree and the root node of the node tree.
[0111] Other historical call records are historical call records in the historical call record dataset except for the target historical call records. Optionally, the node tree includes n layers of nodes. For each node corresponding to a historical call record, step 602c can be implemented as follows: obtaining a demarcation value of a hyperplane division basis for the i+1th layer node of the node tree, wherein a target call parameter in the historical call record is used as a hyperplane division basis for the i+1th layer node of the node tree, the target call parameter is any call parameter in a call parameter set of the historical call record, n is an integer greater than 1, i is a positive integer, and i is less than or equal to n-1; if the standard value corresponding to the target call parameter is greater than or equal to the demarcation value of the hyperplane division basis for the i+1th layer, then the position information of the node corresponding to the historical call record in the node tree is determined to be the first branch of the i-th layer node, and the first branch of the i-th layer node is used as the upper-level node of the i+2th layer node; if the standard value corresponding to the target call parameter is less than the demarcation value of the hyperplane division basis for the i+1th layer, then the position information of the node corresponding to the historical call record in the node tree is determined to be the second branch of the i-th layer node, and the second branch of the i-th layer node is used as the upper-level node of the i+2th layer node. The value of i ranges from 1 to n-1.
[0112] Step 602d: based on the position information of the nodes corresponding to the other historical call records in the node tree, obtain the scores of the other historical call records.
[0113] Optionally, in some embodiments, step 602d may include the following sub-steps. The following steps are performed for each node tree. The number of nodes between the node corresponding to the historical call record and the root node of the node tree, and the average number of nodes of the node tree are obtained, and the score of the node corresponding to the historical call record in the node tree is determined based on the number of nodes between the node corresponding to the historical call record and the root node of the node tree and the average number of nodes of the node tree. Finally, the server obtains the score of the historical call record according to the score of the node corresponding to the historical call record in each node tree.
[0114] Step 602e: Determine whether the other historical call records are abnormal calls based on the scores of the other historical call records.
[0115] When the scores of other historical call records belong to a preset range, the other historical call records are determined to be abnormal calls.
[0116] Step 602f, obtaining the different judgment rate.
[0117] If the discrepancy rate is greater than the preset threshold, the relevant parameters of the node tree are adjusted until the discrepancy rate is less than the preset threshold, and a decision tree model is obtained.
[0118] The relevant parameters include at least one of the following: the number of node trees, the number of nodes included in the node tree, the number of layers of nodes included in the node tree, and the basis for hyperplane division of nodes in each layer of the node tree.
[0119] The number of node trees refers to the number of node trees included in the decision tree model, that is, n_estimators. The number of nodes included in the node tree refers to the number of historical call records for constructing the node tree, that is, max_samples. The number of node layers included in the node number refers to the number of call parameters used when constructing the node tree, that is, max_features. Among them, adjusting the hyperplane division basis of the nodes at each layer of the node tree refers to swapping the order of the call parameters corresponding to each layer of the node tree. In a specific example, the node tree includes four layers, and the hyperplane division basis of each layer to the fourth layer is ringing duration, call duration, call time, and hang-up direction, respectively. The hyperplane division basis of the nodes at each layer of the node tree is swapped, and the hyperplane division basis of the first layer to the fourth layer of the adjusted node tree is call duration, call time, hang-up direction, and ringing duration, respectively.
[0120] The technical solution provided in the embodiment of the present application trains a decision tree model through a large number of unlabeled historical call records, and can subsequently score real-time call records through the trained decision tree model, thereby realizing automatic identification of abnormal calls and improving the efficiency of identifying abnormal calls, so that corresponding processing measures can be provided for abnormal calls in a timely manner in the future.
[0121] See also Figure 7 , which shows a structural block diagram of an abnormal call detection device 700 provided in an embodiment of the present application. The device 700 includes: a parameter acquisition module 710, used to obtain a call parameter set of a call record, the call parameter set including at least one call parameter. A scoring module 720, used to obtain a score of a call record based on the call parameter set of the call record and a pre-trained decision tree model, the decision tree model including a plurality of node trees, the node trees are obtained by dividing the historical call records according to the call parameter set of the historical call records, and the score of the call record represents the probability that the call record is an abnormal call. A detection module 730, used to determine that a call record is an abnormal call when the score of the call record is within a preset interval.
[0122] The embodiment of the present application provides an abnormal call detection device, which quantifies the probability that a call record is an abnormal call through a pre-trained decision tree model and various parameters of the call record. Since the decision tree model is a model that classifies historical call records as examples, when the decision tree model is trained through various parameters of a large number of historical call records, the decision tree model can accurately determine whether it is an abnormal call based on the various parameters of the call record. Therefore, it can realize automatic identification of abnormal calls on the basis of ensuring the recognition accuracy of abnormal call records, improve the recognition efficiency of abnormal calls, so that corresponding processing measures can be provided for abnormal calls in a timely manner later.
[0123] In some embodiments, the scoring module 720 is used to map each call parameter in the call parameter set respectively to obtain a standard value corresponding to each call parameter; the standard value corresponding to the call parameter is a first preset value or a second preset value, and the first preset value is less than the second preset value; based on the standard value corresponding to each call parameter, determine the position information of the target node corresponding to the call record in each node tree; based on the position information of the target node in each node tree, determine the score of the target node in each node tree; based on the score of the target node in each node tree, obtain the score of the call record.
[0124] In some embodiments, the scoring module 720 is used to perform the following steps for each node tree: obtaining the number of nodes between the target node and the root node of the node tree based on the position information of the target node in the node tree; obtaining the average number of nodes of the node tree, the average number of nodes of the node tree refers to the average number of nodes between each node in the node tree and the root node; determining the score of the target node in the node tree based on the number of nodes between the target node and the root node of the node tree and the average number of nodes of the node tree, the score of the target node in the node tree is negatively correlated with the number of nodes between the target node and the root node of the node tree, and the score of the target node in the node tree is negatively correlated with the average number of nodes in the node tree.
[0125] In some embodiments, the node tree includes n layers of nodes; the scoring module 720 is used to perform the following steps on each node tree: obtaining the demarcation value of the hyperplane division basis of the i+1th layer nodes of the node tree, wherein the target call parameter in the call record is used as the hyperplane division basis of the i+1th layer nodes of the node tree, the target call parameter is any call parameter in the call parameter set of the call record, n is an integer greater than 1, i is a positive integer, and i is less than or equal to n-1; if the standard value corresponding to the target call parameter is greater than or equal to the demarcation value of the hyperplane division basis of the i+1th layer, then the position information of the target node in the node tree is determined to be the first branch of the i-th layer node, and the first branch of the i-th layer node is determined to be the upper-level node of the i+2th layer node; if the standard value corresponding to the target call parameter is less than the demarcation value of the hyperplane division basis of the i+1th layer, then the position information of the target node in the node tree is determined to be the second branch of the i-th layer node, and the second branch of the i-th layer node is determined to be the upper-level node of the i+2th layer node, wherein the value of i is 1 to n-1.
[0126] In some embodiments, the device further includes: a result acquisition module and a first adjustment module (not shown in the figure). The result acquisition module is used to obtain a verification result, and the verification result is used to indicate whether the result of determining that the call record is an abnormal call record is accurate. The first adjustment module is used to adjust the hyperplane division basis of at least one node tree if the verification result indicates that the result of determining that the call record is an abnormal call record is inaccurate.
[0127] In some embodiments, the apparatus further comprises: a weighting module ( Figure 7 The weighting module is used to perform weighted processing on the score of the call record based on the user attribute corresponding to the call record to obtain the weighted score, and the user attribute includes at least one of the following: the overdue time corresponding to the call record, and the region to which the number corresponding to the call record belongs. The detection module 730 is used to determine that the call record corresponds to an abnormal call when the weighted score belongs to a preset interval.
[0128] See also Figure 8 , which shows a structural block diagram of a training device 800 for a decision tree model provided in an embodiment of the present application. The device 800 includes: a data acquisition module 810 and a processing module 820.
[0129] The data acquisition module 810 is used to acquire a historical call record data set and a call parameter data set of the historical call record data set, wherein the historical call record data set includes a plurality of historical call records. The iterative training module 820 is used to iteratively train the decision tree model until the discrepancy rate is less than a preset threshold, thereby obtaining a trained decision tree model. The processing module 820 is specifically used to determine the node corresponding to the target historical call record as the root node of the node tree, where the target historical call record is any historical call record in the historical call record data set; determine the hyperplane division basis of each layer of the node tree according to the call parameter data set of the call record data set; determine the position information of the nodes corresponding to other historical call records in the node tree according to the hyperplane division basis of each layer of the node tree and the root node of the node tree, where the other historical call records are historical call records in the historical call record data set other than the target historical call record; obtain the scores of other historical call records based on the position information of the nodes corresponding to other historical call records in the node tree; determine whether other historical call records are abnormal calls based on the scores of other historical call records; and obtain the error judgment rate, which is the ratio between the number of errors in judging other historical call records as abnormal calls and the total number of other historical call records.
[0130] An embodiment of the present application provides a training device for a decision tree model, which trains the decision tree model through a large number of unlabeled historical call records. The trained decision tree model can subsequently be used to score real-time call records, thereby realizing automatic identification of abnormal calls and improving the efficiency of identifying abnormal calls, so that corresponding processing measures can be provided for abnormal calls in a timely manner.
[0131] 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 modules can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here.
[0132] In several embodiments provided in the present application, the coupling between modules may be electrical, mechanical or other forms of coupling.
[0133] In addition, each functional module in each embodiment of the present application can be integrated into a processing module, or each module can exist physically separately, or two or more modules can be integrated into one module. The above integrated modules can be implemented in the form of hardware or software functional modules.
[0134] like Fig. 9 As shown, the present application example further provides an electronic device 900, which may be a server, and includes a processor 910 and a memory 920. The memory 920 stores computer program instructions.
[0135] The processor 910 may include one or more processing cores. The processor 910 uses various interfaces and lines to connect various parts of the entire battery management system, and executes various functions and processes data of the battery management system by running or executing instructions, programs, code sets or instruction sets stored in the memory 920, and calling data stored in the memory 920. Optionally, the processor 910 can be implemented in at least one hardware form of digital signal processing (DSP), field programmable gate array (FPGA), and programmable logic array (PLA). The processor 910 can integrate one or more combinations of a central processing unit 910 (CPU), a graphics processing unit 910 (GPU), and a modem. Among them, the CPU mainly processes the operating system, user interface, and application programs; the GPU is responsible for rendering and drawing display content; and the modem is used to process wireless communications. It can be understood that the above-mentioned modem may not be integrated into the processor 910, but may be implemented separately through a communication chip.
[0136] The memory 920 may include a random access memory 920 (Random Access Memory, RAM), and may also include a read-only memory 920 (Read-Only Memory). The memory 920 may be used to store instructions, programs, codes, code sets, or instruction sets. The memory 920 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for implementing at least one function (such as a touch function, a sound playback function, an image playback function, etc.), instructions for implementing the following various method examples, etc. The data storage area may also store data created by the electronic device during use (such as a phone book, audio and video data, chat record data), etc.
[0137] See also Fig.10 , which shows that an embodiment of the present application also provides a computer-readable storage medium 1000, in which computer program instructions 1010 are stored. The computer program instructions 1010 can be called by a processor to execute the method described in the above embodiment.
[0138] The computer-readable storage medium 1000 may be an electronic memory such as a flash memory, an EEPROM (electrically erasable programmable read-only memory), an EPROM, a hard disk or a ROM. Optionally, the computer-readable storage medium 1000 includes a non-transitory computer-readable storage medium. The computer-readable storage medium 1000 has storage space for computer program instructions 1010 for executing any method steps of the above method. These computer program instructions 1010 may be read from or written to one or more computer program products. The computer program instructions 1010 may be compressed in an appropriate form.
[0139] The above are only preferred examples of the present application and are not intended to limit the present application in any form. Although the present application has been disclosed as the preferred examples above, they are not intended to limit the present application. Any technical personnel in this field can make some changes or modify the technical contents disclosed above into equivalent examples with equivalent changes without departing from the scope of the technical solution of the present application. However, any brief modifications, equivalent changes and modifications made to the above examples based on the technical essence of the present application without departing from the content of the technical solution of the present application are still within the scope of the technical solution of the present application.
Claims
1. A method for detecting abnormal calls, characterized in that: The method comprises: Acquire a call parameter set of the call record, where the call parameter set includes at least one call parameter; Scoring the call record based on the call parameter set of the call record and the pre-trained decision tree model to obtain the score of the call record, wherein the decision tree model is obtained by iterative training based on the historical call record data set, and the decision tree model includes multiple node trees, and each historical call record corresponds to a node tree; for each node tree, the hyperplane division basis of each layer of nodes of the node tree is based on the difference judgment rate of judging whether other historical call records are abnormal calls is less than a preset threshold, and the hyperplane division basis of each layer of nodes of the node tree is based on a call parameter corresponding to the target historical call record, and the target historical call record corresponds to the node tree; the score of the call record is used to characterize the probability that the call record is an abnormal call; When the score of the call record is within a preset range, determining that the call record is an abnormal call; The abnormal judgment rate is the ratio between the number of abnormal call errors in other historical call records and the total number of other historical call records.
2. The method according to claim 1, characterized in that: Scoring the call record based on the call parameter set of the call record and the pre-trained decision tree model to obtain the score of the call record includes: Mapping each call parameter in the call parameter set to obtain a standard value corresponding to each call parameter; the standard value corresponding to the call parameter is a first preset value or a second preset value, and the first preset value is less than the second preset value; Based on the standard values corresponding to the call parameters, respectively, determining the position information of the target node in each of the node trees, wherein the target node is used to represent the call record; Determine the score of the target node in each of the node trees based on the position information of the target node in each of the node trees; The score of the call record is obtained based on the score of the target node in each of the node trees.
3. The method according to claim 2, characterized in that The node tree includes n layers of nodes; The determining, based on the standard values corresponding to the call parameters, the position information of the target node in each of the node trees comprises: The following steps are performed for each of the node trees: Obtaining a demarcation value of a hyperplane division basis for the i+1th layer node of the node tree, wherein a target call parameter in the call record is used as a hyperplane division basis for the i+1th layer node of the node tree, the target call parameter is any call parameter in a call parameter set of the call record, n is an integer greater than 1, i is a positive integer, and i is less than or equal to n-1; If the standard value corresponding to the target call parameter is greater than or equal to the demarcation value based on which the hyperplane of the i+1 layer is divided, the position information of the target node in the node tree is determined to be the first branch of the i-th layer node, and the first branch of the i-th layer node is determined to be the upper-level node of the i+2-th layer node; If the standard value corresponding to the target call parameter is smaller than the demarcation value based on which the hyperplane division of the i+1 layer is based, the position information of the target node in the node tree is determined to be the second branch of the i-th layer node, and the second branch of the i-th layer node is determined to be the upper-level node of the i+2-th layer node, where the value of i is 1 to n-1 respectively.
4. The method according to claim 2, characterized in that The determining, based on the position information of the target node in each of the node trees, a score of the target node in each of the node trees, comprises: The following steps are performed for each of the node trees: Based on the position information of the target node in the node tree, obtaining the number of nodes between the target node and the root node of the node tree; Obtaining an average number of nodes in the node tree, where the average number of nodes in the node tree refers to an average value of the number of nodes between each node in the node tree and the root node; Based on the number of nodes between the target node and the root node of the node tree and the average number of nodes in the node tree, the score of the target node in the node tree is determined, the score of the target node in the node tree is negatively correlated with the number of nodes between the target node and the root node of the node tree, and the score of the target node in the node tree is negatively correlated with the average number of nodes in the node tree.
5. The method according to claim 1, characterized in that After determining that the call record is an abnormal call, the method further includes: Obtaining a verification result, wherein the verification result is used to indicate whether the result of determining that the call record is an abnormal call record is accurate; If the verification result indicates that the call record is an abnormal call record and the result is inaccurate, a hyperplane division basis of at least one node tree is adjusted.
6. The method according to claim 1, characterized in that After obtaining the score of the call record, the method further includes: The score of the call record is weighted based on the user attribute corresponding to the call record to obtain the weighted score, wherein the user attribute includes at least one of the following: the overdue duration corresponding to the call record and the region to which the number corresponding to the call record belongs.
7. A training method for a decision tree model, characterized in that: The decision tree model includes a plurality of node trees, and the method includes: Acquire a historical call record data set and a call parameter data set of the historical call record data set, wherein the historical call record data set includes a plurality of historical call records; The decision tree model is iteratively trained until the discrepancy rate is less than a preset threshold, thereby obtaining a trained decision tree model, which specifically includes: Determine a node corresponding to a target historical call record as a root node of the node tree, wherein the target historical call record is any historical call record in the historical call record data set; Determine, according to the call parameter data set of the call record data set, a hyperplane division basis of nodes at each layer of the node tree; determine, according to the hyperplane division basis of nodes at each layer of the node tree and the root node of the node tree, the position information of nodes corresponding to other historical call records in the node tree, wherein the other historical call records are historical call records in the historical call record data set other than the target historical call records; Obtaining scores of other historical call records based on position information of nodes corresponding to other historical call records in the node tree; Determining whether the other historical call records are abnormal calls based on the scores of the other historical call records; The error determination rate is obtained, where the error determination rate is the ratio of the number of other historical call records determined to be abnormal call errors to the total number of other historical call records.
8. An abnormal call detection device, characterized in that: The device comprises: A parameter acquisition module, used to acquire a call parameter set of a call record, wherein the call parameter set includes at least one call parameter; A scoring module, used for scoring the call record based on the call parameter set of the call record and the pre-trained decision tree model to obtain the score of the call record, wherein the decision tree model is obtained by iterative training based on the historical call record data set, and the decision tree model includes multiple node trees, each historical call record corresponds to a node tree; for each node tree, based on the hyperplane division of each layer of nodes of the node tree, the difference judgment rate of judging whether other historical call records are abnormal calls is less than a preset threshold, and the hyperplane division of each layer of nodes of the node tree is based on a call parameter corresponding to a target historical call record, and the target historical call record corresponds to the node tree; the score of the call record is used to characterize the probability that the call record is an abnormal call; The detection module is used to determine that the call record is an abnormal call when the score of the call record is within a preset range, and the abnormal judgment rate is the ratio between the number of abnormal call errors in the other historical call records and the total number of the other historical call records.
9. An electronic device, characterized in that: The electronic device includes a processor and a memory, wherein the memory stores computer program instructions, and the computer program instructions are called by the processor to execute the abnormal call detection method as described in any one of claims 1 to 6, or the training method of the decision tree model as described in claim 7.
10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores program code, and the program code is called by a processor to execute the abnormal call detection method as described in any one of claims 1 to 6, or the decision tree model training method as described in claim 7.
11. A computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the computer program implements the abnormal call detection method according to any one of claims 1 to 6, or the abnormal call detection method according to claim 7.
