Abnormal number identification method and device, equipment, medium and product

By applying noise filtering conditions in abnormal number recognition, filtering misjudgment numbers based on the target group's unique attributes and behavioral habits, the problem of insufficient accuracy in abnormal number recognition in traditional methods is solved, and the recognition accuracy is improved.

CN118945263APending Publication Date: 2024-11-12CHINA MOBILE GROUP ANHUI +2
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410929600.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-11
Publication Date
2024-11-12

AI Technical Summary

Technical Problem

Traditional abnormal number recognition methods are prone to misjudging normal numbers as abnormal numbers, and the recognition accuracy needs to be improved.

Method used

By obtaining the target information of the number to be identified, input the pre-trained exception number identification model, and applying noise filtering conditions in the recognition results, filter out the real exception number based on the unique number attributes and behavioral habits of the target group.

Benefits of technology

It improves the accuracy of abnormal number recognition, reduces misjudgment of normal number, and ensures the accuracy of identification results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118945263A_ABST
    Figure CN118945263A_ABST
Patent Text Reader

Abstract

The invention discloses an abnormal number identification method and device, equipment, a medium and a product, which are applied to the technical field of communication, and the method comprises the following steps: training according to target information of historical abnormal numbers in advance to obtain an abnormal number identification model, and establishing a noise filtering condition at least based on specific number attributes and / or specific behavior habits of a target group; the number of the target group belongs to a normal number. Under the condition that the identification result of the to-be-identified number represents that the to-be-identified number is a suspected abnormal number, performing further judgment based on a noise filtering condition; and determining the suspected abnormal number which does not meet the noise filtering condition as a target abnormal number. The embodiment of the invention considers that the accuracy of the recognition result output by the abnormal number recognition model is relatively low due to the fact that the actual scene is possibly interfered by some specific target groups. Therefore, further judgment is performed through the noise filtering condition, and the target group is filtered, so that the accuracy of abnormal number recognition is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the field of communication technology, and in particular, relates to a method, device, equipment, medium and product for identifying abnormal numbers. Background Art

[0002] Traditional abnormal number identification and early warning methods usually obtain user signaling and call record data of known abnormal numbers as training data, and then train the corresponding recognition model based on the deep learning algorithm. The trained recognition model is then used to analyze the number to be identified, thereby identifying the abnormal number.

[0003] However, traditional recognition schemes easily misjudge some normal numbers as abnormal numbers, and the accuracy of recognition needs to be improved. Summary of the invention

[0004] The embodiments of the present application provide a method, device, equipment, medium and product for identifying abnormal numbers, which can improve the accuracy of abnormal number identification.

[0005] On the one hand, an embodiment of the present application provides a method for identifying an abnormal number, comprising:

[0006] Obtain target information of the number to be identified, the target information including at least any one of the following or any combination thereof: number attributes, user signaling, and call record data;

[0007] Inputting the target information of the number to be identified into a pre-trained abnormal number identification model to obtain an identification result output by the abnormal number identification model; the abnormal number identification model is trained based on the target information of historical abnormal numbers;

[0008] In the case where the identification result indicates that the number to be identified includes a suspected abnormal number, determining whether the suspected abnormal number satisfies a noise filtering condition; the noise filtering condition is established based at least on a unique number attribute and / or unique behavior habit of a target group, and the numbers of the target group are normal numbers;

[0009] The suspected abnormal number that does not meet the noise filtering condition is determined as the target abnormal number.

[0010] On the other hand, there are multiple target groups;

[0011] Before determining whether the suspected abnormal number meets the noise filtering condition, the method further includes:

[0012] Obtain target information of numbers corresponding to each target group;

[0013] Based on the target information of the numbers corresponding to the target groups, determining the unique number attributes and / or unique behavior habits corresponding to the target groups;

[0014] The corresponding noise filtering conditions are respectively constructed based on the unique number attributes and / or unique behavior habits corresponding to each of the target groups.

[0015] On the other hand, the target group includes at least any one or any combination of the following: the elderly, students, unemployed persons and home workers;

[0016] Number attributes include at least the account opening age range and / or package level;

[0017] The behavior habits include at least any one or any combination of the following: the number of calls, the number of calls, the continuous moving time period, the location of the permanent base station, the type of commonly used software, the traffic usage and the target LAN usage.

[0018] On the other hand, the unique number attributes of the elderly include at least:

[0019] The age in the account opening certificate is greater than the first preset age;

[0020] The unique behavioral habits of the elderly include at least one of the following or any combination:

[0021] In the first preset historical time period, the proportion of the traffic used in shopping software and video software to the total traffic used exceeds the preset ratio;

[0022] During the second preset historical time period, the average daily number of calls is greater than the first preset number;

[0023] The earliest continuous movement time in a single day is earlier than the first preset time in the morning of that day.

[0024] On the other hand, the unique number attribute of the student includes at least one of the following or any combination thereof:

[0025] The age of the account opening certificate is less than the second preset age; the second preset age is less than the first preset age;

[0026] The package level is lower than the preset level;

[0027] The student’s unique behavioral habits include at least one of the following or any combination:

[0028] The top n base stations in terms of dwell time on weekdays are the base stations around the school; n is a positive integer;

[0029] Third, there was behavior of using the campus network within the preset historical time period.

[0030] On the other hand, the unique number attributes of unemployed persons include at least:

[0031] The age in the account opening certificate is greater than the third preset age and less than the fourth preset age;

[0032] The unique behavioral habits of unemployed persons include at least one of the following or any combination thereof:

[0033] In the fourth preset historical time period, the average daily number of calls is less than the second preset number;

[0034] The daily traffic usage is greater than the preset traffic usage, and the frequency of using the following software is higher than the first frequency: video software, game software and communication software;

[0035] There is overlap between the top x base stations in terms of daytime residence time and the top y base stations in terms of nighttime residence time; x and y are positive integers.

[0036] On the other hand, the unique behavioral habits of home workers include at least one or any combination of the following:

[0037] The base stations with the highest i-th longest residence time during the day and the base stations with the highest j-th longest residence time at night overlap; i and j are positive integers;

[0038] In a fifth preset historical time period, the average daily number of calls is greater than a third preset number and less than a fourth preset number; and the third preset number is greater than the fourth preset number.

[0039] On the other hand, before inputting the target information of the number to be identified into the pre-trained abnormal number identification model, the method further includes:

[0040] Acquire target information of a first historical abnormal number to construct a first information set;

[0041] extracting a feature label based on target information of each of the first historical abnormal numbers in the first information set;

[0042] With the goal of hitting more of the first historical abnormal numbers in the first information set, determining a preset number of valid tags from the feature tags;

[0043] Building an initial abnormal number recognition model based on the valid tags;

[0044] The initial abnormal number recognition model is trained according to the target information of the second historical abnormal number and the historical normal number to obtain the abnormal number recognition model.

[0045] On the other hand, the initial abnormal number recognition model includes a label extraction network, a label matching network, a weight calculation network and a recognition result output network;

[0046] The training of the initial abnormal number recognition model to obtain the abnormal number recognition model includes:

[0047] Acquire target information of the second historical abnormal number and the historical normal number to construct a second information set;

[0048] Using the label extraction network to extract labels from the target information of the second information set to obtain a first target label;

[0049] Using the label matching network to match the first target label with the valid label, and filtering out a second target label belonging to the valid label from the first target label;

[0050] Based on the weights of the valid tags, the weight calculation network is used to perform weight calculation on each of the second target tags to obtain a first abnormal number suspicion value of each number in the second information set;

[0051] Using the recognition result output network, outputting an initial recognition result according to the first abnormal number suspicion value;

[0052] When the initial recognition result does not meet the preset condition, adjusting each of the valid labels and the weights thereof, and returning to the step of extracting labels from the target information of the second information set using the label extraction network to obtain a first target label;

[0053] When the initial recognition result satisfies the preset conditions, the initial abnormal number recognition model under the current weight is determined to be the abnormal number recognition model; the preset conditions include: in the initial recognition result, the number of the second historical abnormal numbers identified as suspected abnormal numbers is higher than the first preset threshold, and the number of the historical normal numbers identified as suspected abnormal numbers is less than the second preset threshold.

[0054] On the other hand, the step of inputting the target information of the number to be identified into a pre-trained abnormal number identification model to obtain the identification result output by the abnormal number identification model includes:

[0055] Using the label extraction network to extract labels from the target information of the number to be identified, to obtain a third target label;

[0056] Using the label matching network to match the third target label with the valid label, and screening out a fourth target label belonging to the valid label from the third target label;

[0057] Based on the weights of the valid tags, the weight calculation network is used to perform weight calculation on each of the fourth target tags to obtain a second abnormal number suspicion value of the number to be identified;

[0058] The recognition result output network is used to output the recognition result according to the second abnormal number suspicion value.

[0059] On the other hand, an embodiment of the present application provides a device for identifying an abnormal number, the device comprising:

[0060] An acquisition module, used to acquire target information of the number to be identified, the target information including at least any one of the following or any combination thereof: number attributes, user signaling, and call record data;

[0061] The recognition result determination module is used to input the target information of the number to be recognized into a pre-trained abnormal number recognition model to obtain the recognition result output by the abnormal number recognition model; the abnormal number recognition model is trained according to the target information of historical abnormal numbers;

[0062] A judgment module, configured to judge whether the suspected abnormal number satisfies a noise filtering condition when the recognition result indicates that the number to be recognized includes a suspected abnormal number; the noise filtering condition is established based on at least a unique number attribute and / or a unique behavior habit of a target group, and the number of the target group is a normal number;

[0063] The determination module is used to determine the suspected abnormal number that does not meet the noise filtering condition as the target abnormal number.

[0064] In another aspect, an embodiment of the present application provides a device for identifying abnormal numbers, the device comprising: a processor and a memory storing computer program instructions;

[0065] When the processor executes the computer program instructions, the method for identifying the abnormal number as described above is implemented.

[0066] On the other hand, an embodiment of the present application provides a computer-readable storage medium, on which computer program instructions are stored. When the computer program instructions are executed by a processor, the method for identifying abnormal numbers as described above is implemented.

[0067] On the other hand, an embodiment of the present application provides a computer program product. When instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the method for identifying abnormal numbers as described above.

[0068] The method for identifying abnormal numbers in the embodiment of the present application is to obtain an abnormal number recognition model by training the target information of historical abnormal numbers in advance, and to establish a noise filtering condition based on at least the unique number attributes and / or unique behavior habits of the target group; the numbers of the above target group belong to normal numbers. Then, the target information of the number to be identified is obtained to input into the abnormal number recognition model, so as to obtain the recognition result. In the case where the recognition result indicates that the number to be identified is a suspected abnormal number, the suspected abnormal number is filtered based on the noise filtering condition; and the suspected abnormal number that does not meet the noise filtering condition is determined as the target abnormal number. The embodiment of the present application takes into account that in actual application scenarios, when performing abnormal number recognition, it may be interfered by some specific target groups, resulting in a low accuracy of the recognition result output by the abnormal number recognition model. Therefore, after the abnormal number recognition model outputs the recognition result, it is also filtered based on the noise filtering condition, the target group is filtered out, and the real target abnormal number is screened out, thereby improving the accuracy of abnormal number recognition. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solution of the embodiments of the present application, the following is a brief introduction to the drawings required for use in the embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0070] Figure 1 A schematic diagram showing a flow chart of a method for identifying an abnormal number provided by an embodiment of the present application;

[0071] Figure 2 A complete flow chart for identifying abnormal numbers provided by an embodiment of the present application is shown;

[0072] Figure 3 A schematic diagram of the structure of an abnormal number identification device provided in an embodiment of the present application is shown;

[0073] Figure 4 A schematic diagram of the hardware structure of an abnormal number identification device provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0074] The features and exemplary embodiments of various aspects of the present application will be described in detail below. In order to make the purpose, technical solutions and advantages of the present application clearer, the present application will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are only intended to explain the present application, rather than to limit the present application. For those skilled in the art, the present application can be implemented without the need for some of these specific details. The following description of the embodiments is only to provide a better understanding of the present application by illustrating the examples of the present application.

[0075] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device. In the absence of further restrictions, the elements defined by the statement "include..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0076] The traditional method of identifying abnormal numbers uses deep learning algorithms to train corresponding models for analysis and identification. This may be interfered by similar people, resulting in misjudgment of abnormal numbers and causing inconvenience to these normal users.

[0077] In order to solve the problems of the traditional solution, the embodiment of the present application provides a method, device, equipment, medium and product for identifying an abnormal number. The following first introduces the method for identifying an abnormal number provided by the embodiment of the present application. Figure 1 FIG. 1 is a flow chart showing a method for identifying an abnormal number provided by an embodiment of the present application; Figure 1 As shown, the method includes the following steps S101 to S104:

[0078] S101: Obtain target information of the number to be identified.

[0079] The above target information includes at least any one of the following or any combination: number attributes, user signaling, and call bill data; number attributes may include account opening age range and package level, etc., user signaling includes location signaling and Internet log information, and call bill data includes detailed call information, such as caller or called, time, base station, and duration, etc. This information can be used as a basis for determining whether the number is abnormal. The acquisition method can be selected according to the actual situation, and the number to be identified can be one or more; usually, a large number of numbers need to be identified at the same time; since data from historical periods is required during identification, the historical data of these numbers to be identified can be temporarily stored in the database and exported during identification.

[0080] S102: Inputting target information of the number to be identified into a pre-trained abnormal number identification model to obtain an identification result output by the abnormal number identification model.

[0081] Before performing abnormal number recognition, it is necessary to obtain the target information of historical abnormal numbers and train an abnormal number recognition model. First, a large amount of target information of historical abnormal numbers can be obtained, and then feature tags can be extracted based on these target information. Then, with the goal of hitting more first historical abnormal numbers in the first information set, valid tags representing number abnormalities are determined from the feature tags. By assigning corresponding weights to these valid tags, an initial abnormal number recognition model is constructed.

[0082] After the model is built, the effective labels and corresponding weights obtained by the above solution may not be ideal. Then the model is trained through the second information set containing the target information of the second historical abnormal number and the historical normal number. Specifically, the goal is to make the suspected abnormal numbers output by the model hit the second historical abnormal number more often and hit the historical normal number less often, and adjust each effective label and its weight to make the recognition effect of the model better.

[0083] The specific structure of the model is not limited, and may include a label extraction network, a label matching network, a weight calculation network, and a recognition result output network. After obtaining the target information of the number to be identified, it is input into the abnormal number recognition model. The model uses the label extraction network to extract labels from the target information of the number to be identified, and then uses the label matching network to match the extracted labels with valid labels. Based on the matching results and the weights of the valid labels, the weight calculation network is used to perform weight calculation to obtain the abnormal number suspicion value of the number to be identified; finally, the recognition result output network is used to output the recognition result according to the second abnormal number suspicion value.

[0084] S103: When the identification result indicates that the number to be identified includes a suspected abnormal number, it is determined whether the suspected abnormal number satisfies a noise filtering condition.

[0085] In actual applications, there are many numbers to be identified, which may include both abnormal numbers and normal numbers; correspondingly, the identification results may also include both suspected abnormal numbers and normal numbers. The model will identify all abnormal numbers, but there may be misjudged normal numbers. In this case, these normal numbers need to be filtered out, and the remaining suspected normal numbers are determined to be the real abnormal numbers, that is, the target abnormal numbers.

[0086] The embodiment of the present application specifically filters normal numbers by constructing noise filtering conditions, and the noise filtering conditions can be established based on the unique number attributes and unique behavior habits of the target group. The numbers of the above target group are normal numbers, and the specific type of the target group can be set according to the actual situation. For example, the behaviors of the elderly, students, unemployed people, and home workers are easily misjudged as abnormal numbers by the above model, so the unique number attributes and unique behavior habits of these groups can be counted, and then the noise filtering conditions can be constructed to determine the corresponding numbers as normal numbers.

[0087] The specific content of number attributes and behavior habits can be determined according to actual conditions. For example, number attributes may include the age range and package level of account opening, and behavior habits may include: the number of calls, the number of calls, continuous moving time periods, permanent base station locations, common software types, traffic usage, and target LAN usage. Through this information, it can be confirmed that the number is used by the corresponding target group.

[0088] S104: Determine the suspected abnormal number that does not meet the noise filtering condition as the target abnormal number.

[0089] After filtering out the numbers of the target group, the remaining suspected abnormal numbers in the identification results are determined to be target abnormal numbers that are truly abnormal. These target abnormal numbers can be marked, and calls made by target abnormal numbers will be given risk warnings on the answering device, thereby preventing the personal and property safety of users of normal numbers from being violated.

[0090] The method for identifying abnormal numbers in the embodiment of the present application is to obtain an abnormal number recognition model by training the target information of historical abnormal numbers in advance, and to establish a noise filtering condition based on at least the unique number attributes and / or unique behavioral habits of the target group; the numbers of the above target group belong to normal numbers. Then, the target information of the number to be identified is obtained to be input into the abnormal number recognition model, so as to obtain the recognition result. In the case where the recognition result indicates that the number to be identified is a suspected abnormal number, the suspected abnormal number is filtered based on the noise filtering condition; and the suspected abnormal number that does not meet the noise filtering condition is determined as the target abnormal number. The embodiment of the present application takes into account that in actual application scenarios, when performing abnormal number recognition, it may be interfered by some specific target groups, resulting in a low accuracy of the recognition result output by the abnormal number recognition model. Therefore, after the abnormal number recognition model outputs the recognition result, it is filtered once based on the noise filtering condition to filter out the target group and screen out the real target abnormal number, thereby improving the accuracy of abnormal number recognition.

[0091] In practical applications, there are usually multiple target groups, and the usage habits of these groups can easily cause their numbers to be misjudged as abnormal by the model. However, the usage habits and number attributes of each target group are different, so it is necessary to determine corresponding noise filtering conditions for each target group. The embodiment of the present application provides a specific implementation scheme. Before determining whether the suspected abnormal number meets the noise filtering condition, the target information of the corresponding number of each target group is obtained; and based on the target information of the corresponding number of each target group, the unique number attributes and / or unique behavioral habits corresponding to each target group are determined. Finally, corresponding noise filtering conditions are constructed based on the unique number attributes and / or unique behavioral habits corresponding to each target group.

[0092] The embodiment of the present application provides a specific solution for determining noise filtering conditions. Based on different target groups, corresponding noise filtering conditions are determined respectively, so that the numbers of the corresponding target groups can be accurately filtered.

[0093] The above embodiment mentioned that the usage habits of the target group can easily cause their numbers to be misjudged as abnormal by the model. The embodiment of the present application provides a specific strength. The target group includes at least any one of the following or any combination: the elderly, students, unemployed people, and home workers. In order to accurately distinguish these target groups, the noise filtering conditions can be determined based on their unique number attributes and unique behavioral habits. The number attributes include at least the age range and / or package level of the account opening; the behavioral habits include at least any one of the following or any combination: the number of calls, the number of calls, the continuous moving time period, the location of the permanent base station, the type of commonly used software, the traffic usage, and the target LAN usage.

[0094] Based on the number attributes and behavioral habits provided in the embodiments of the present application, more accurate noise filtering conditions can be constructed to accurately filter the target group.

[0095] The following is an explanation of each target group, as well as its corresponding unique number attributes and unique behavioral habits.

[0096] The unique number attributes of the elderly include at least: the age of the account opening certificate is greater than the first preset age. The unique behavior habits of the elderly include at least any one of the following or any combination: 1. In the first preset historical time period, the proportion of the traffic used in shopping software and video software to the total traffic used exceeds the preset ratio; 2. In the second preset historical time period, the average number of daily calls is greater than the first preset number; 3. The earliest continuous movement time in a single day is earlier than the first preset time in the morning of the same day.

[0097] Here, the first preset age, the first preset historical time period, the preset ratio, the second preset historical time period, the first preset number of times and the first preset time can all be set according to actual conditions. For example, the first preset age can be set to 60 years old, the first preset historical time period and the second preset historical time period can both be set to the past 15 days, the preset ratio can be 90%, the first preset number of times can be set to 8, and the first preset time can be set to 7:30 in the morning.

[0098] The noise filtering conditions constructed by the unique number attributes and unique behavioral habits of the elderly provided in this application can accurately screen out the numbers used by the elderly group from suspected abnormal numbers, thereby avoiding misjudgment.

[0099] The unique number attributes of students include at least one of the following or any combination: 1. The age of the account opening certificate is less than the second preset age; the second preset age is less than the first preset age; 2. The package level is lower than the preset level. The unique behavior habits of students include at least one of the following or any combination: 1. The top n base stations for the length of stay on weekdays are base stations around the school; n is a positive integer; 2. There is behavior of using the campus network in the third preset historical time period.

[0100] Here, the second preset age, preset level, third preset historical time period and the value of n can all be set according to the actual situation. For example, the second preset age can be set to 22 years old, the preset level can be determined according to the package amount, and can mainly refer to the main package. The amount of the main package used by students is relatively small, and the specific amount can be selected according to the actual situation; the third preset historical time period can be set to the past 15 days. The value of n can be set to 3, then the unique behavior habit corresponds to the top 3 base stations with the longest residence time on weekdays are the base stations around the school.

[0101] The noise filtering conditions constructed by the unique number attributes and unique behavioral habits of students provided in this application can accurately screen out the numbers used by the student group among the suspected abnormal numbers, thereby avoiding misjudgment.

[0102] The unique number attributes of unemployed persons include at least: the age of the account opening certificate is greater than the third preset age and less than the fourth preset age. The unique behavioral habits of unemployed persons include at least any one of the following or any combination: 1. During the fourth preset historical time period, the average daily number of calls is less than the second preset number; 2. The daily traffic usage is greater than the preset traffic, and the frequency of using the following software is higher than the first frequency: video software, game software and communication software; 3. The base stations with the top x daytime residence time and the base stations with the top y nighttime residence time overlap; x and y are positive integers.

[0103] Here, the third preset age, the fourth preset age, the fourth preset historical time period, the second preset number of times, the preset flow rate, the first frequency, and the values ​​of x and y can all be set according to actual conditions. For example, the third preset age can be set to 22 years old, the fourth preset age can be set to 55 years old, the fourth preset historical time period can be set to the past 15 days, and the second preset number of times can be set to 5. The preset flow rate can be set to 300mb, and the first frequency can be set to the average daily usage times. The values ​​of x and y can both be set to 3, then the unique behavior habits correspond to the top 3 base stations with the longest residence time during the day and the top 3 base stations with the longest residence time at night overlap.

[0104] By using the noise filtering conditions constructed by the unique number attributes and unique behavioral habits of the unemployed provided in this application, the numbers used by the unemployed group among the suspected abnormal numbers can be accurately screened out, thereby avoiding misjudgment.

[0105] The unique behavioral habits of home workers include at least any one of the following or any combination: 1. The base stations with the top i in terms of daytime residence time and the base stations with the top j in terms of nighttime residence time overlap; i and j are positive integers; 2. During the fifth preset historical time period, the average daily number of calls is greater than the third preset number and less than the fourth preset number; the third preset number is greater than the fourth preset number.

[0106] Here, the values ​​of i and j, the fifth preset historical time period, the third preset number of times, and the fourth preset number of times can all be set according to actual conditions. For example, i and j can be set to 3, and the unique behavior habit corresponds to the top 3 base stations with the longest residence time during the day and the top 3 base stations with the longest residence time at night. The fifth preset historical time period can be set to the past 15 days, the third preset number of times can be set to 5, and the fourth preset number of times can be set to 3.

[0107] The noise filtering conditions constructed by the unique number attributes and unique behavioral habits of home workers provided in this application can accurately screen out the numbers used by the home worker group from suspected abnormal numbers, thereby avoiding misjudgment.

[0108] The above embodiment mentioned that the abnormal number recognition model is obtained by training with historical abnormal numbers. Before training, it is necessary to first construct an initial abnormal number recognition model. This embodiment provides a specific construction method. Before inputting the target information of the number to be identified into the pre-trained abnormal number recognition model, the target information of the first historical abnormal number is also obtained to construct a first information set. There are a large number of first historical abnormal numbers in the first information set. Based on the target information of each first historical abnormal number in the first information set, feature labels are extracted, and a large number of feature labels can be obtained at this time. With the goal of hitting more first historical abnormal numbers in the first information set, a preset number of valid labels is determined from the feature labels, that is, a number belongs to the corresponding valid label, indicating that the number may be abnormal. Finally, an initial abnormal number recognition model is constructed based on the valid labels, and then the initial abnormal number recognition model is trained to obtain an abnormal number recognition model.

[0109] The embodiment of the present application extracts feature labels from the target information of the first historical abnormal number in the first information set, and selects valid labels to construct a model, so that the abnormal number recognition model can recognize abnormal numbers.

[0110] The above embodiment mentioned that after constructing the initial abnormal number recognition model, the initial abnormal number recognition model is trained to obtain the abnormal number recognition model. This embodiment specifically describes the training process. The initial abnormal number recognition model includes a label extraction network, a label matching network, a weight calculation network, and a recognition result output network.

[0111] The steps of training the initial abnormal number recognition model and obtaining the abnormal number recognition model specifically include: obtaining target information of the second historical abnormal number and the historical normal number to construct a second information set; using a label extraction network to extract labels from the target information of the second information set to obtain a first target label; using a label matching network to match the first target label with a valid label, and screening out a second target label belonging to a valid label from the first target label.

[0112] Then, based on the weights of the valid labels, the weight calculation network is used to calculate the weights of the second target labels to obtain the first abnormal number suspicion value of each number in the second information set. In the above, when the initial abnormal number recognition model is constructed based on the valid labels, a weight corresponding to each valid label can be assigned, and the weight calculation module performs calculations based on the weights. For example, if a number belongs to multiple valid labels, the weights corresponding to these valid labels are added together to obtain the corresponding abnormal number suspicion value.

[0113] Finally, the recognition result output network is used to output the initial recognition result according to the suspected value of the first abnormal number. Here, the recognition result can be determined specifically according to the suspected value of the first abnormal number. For example, if the suspected value of the first abnormal number is higher than the set threshold, the corresponding number is determined to be a suspected abnormal number.

[0114] When the initial recognition result does not meet the preset conditions, the valid labels and their weights are adjusted, and the label extraction network is used to extract the target information of the second information set to obtain the first target label, so as to repeat the above training process.

[0115] When the initial recognition result meets the preset conditions, the initial abnormal number recognition model under the current weight is determined to be the abnormal number recognition model. The preset conditions include: in the initial recognition result, the number of second historical abnormal numbers identified as suspected abnormal numbers is higher than the first preset threshold, and the number of historical normal numbers identified as suspected abnormal numbers is less than the second preset threshold. That is, the training goal is that the recognition result of the model hits more second historical abnormal numbers and hits fewer historical normal numbers.

[0116] Through the training method provided in the embodiment of the present application, each valid label and its weight are continuously adjusted, so that the abnormal number recognition model finally obtained can more accurately judge whether the identified number is an abnormal number.

[0117] After obtaining the abnormal number recognition model, the target information of the number to be recognized can be input into the pre-trained abnormal number recognition model to obtain the recognition result output by the abnormal number recognition model. The specific process includes: using the label extraction network to extract the label of the target information of the number to be recognized to obtain the third target label; using the label matching network to match the third target label with the valid label, and screening the fourth target label belonging to the valid label from the third target label. The fourth target label of a number to be recognized is the label that coincides with the valid label of the number.

[0118] Based on the weights of each valid label, the weight calculation network is used to calculate the weights of each fourth target label to obtain the second abnormal number suspicion value of the number to be identified. The specific calculation process can be to add the weights corresponding to these valid labels (fourth target labels) respectively, that is, to obtain the corresponding abnormal number suspicion value. Finally, the recognition result output network is used to output the recognition result according to the second abnormal number suspicion value.

[0119] Through the abnormal number recognition model provided in the embodiment of the present application, an accurate recognition result of the number to be recognized can be obtained.

[0120] Figure 2FIG. 1 shows a complete flow chart of identifying abnormal numbers provided by an embodiment of the present application, such as Figure 2 As shown in the figure, the complete steps of model construction, training and application are as follows:

[0121] S201: Construct a first information set based on target information of a first historical abnormal number.

[0122] For example, user signaling and voice call record data of 1,000 identified abnormal numbers may be obtained as the first information set.

[0123] S202: Extracting feature tags based on target information of the first historical abnormal number in the first information set.

[0124] For example, the first information set can extract more than 100 feature tags, including the number of calls on the same day, the number of called, the proportion of calls, the dispersion of calls, the number of cities where calls are made, the proportion of new callers, the number of calls in the past 7 days, the number of called in the past 7 days, the dispersion of calls in the past 7 days, the proportion of new callers in the past 7 days, the number of terminals switched in the past 7 days, the base station lacci during the call, the account opening package, the age of the number owner, and the time online. The tags obtained by initially decomposing the user's call bill, signaling and other information are shown in the following table:

[0125] Table 1 Feature label table

[0126]

[0127]

[0128]

[0129] S203: With the goal of hitting more first historical abnormal numbers in the first information set, a preset number of valid tags are determined from the feature tags, and an initial model is constructed.

[0130] This step is mainly used to determine the valid labels and corresponding weights, and preliminarily determine the model. Weights are assigned to each parameter so that the rule model can hit as many training set users as possible. The model is considered to be initially reliable at this time. The valid labels and weights at this time are as follows:

[0131] Table 2 Valid labels and corresponding weights

[0132]

[0133]

[0134] S204: Construct a second information set based on the target information of the second historical abnormal number and the historical non-abnormal number.

[0135] S205: Based on the second information set, adjust each valid label and its weight to obtain an abnormal number recognition model.

[0136] That is, import the second information set to adjust the labels and weights. For example, you can import the 1,000 most recently determined abnormal numbers and 1,000 normal numbers, adjust the weight of each label, introduce new labels, and delete invalid labels, so that as many abnormal numbers in the second information set as possible are hit, while as few normal numbers as possible are mislabeled, thereby improving the reliability of the model.

[0137] S206: Run the model to obtain a suspected abnormal number.

[0138] You can import target information such as daily user signaling and call record data, and run the model to obtain suspected abnormal numbers.

[0139] S207: Perform noise filtering according to the noise filtering condition.

[0140] Noise filtering conditions can be established by combining traffic call records with DPI technology, call behavior, user information, and location behavior to profile groups such as the elderly, students, the unemployed, and those working from home, and corresponding models can be built for filtering.

[0141] S208: Output the target abnormal number.

[0142] After filtering, the suspected abnormal numbers that do not meet the noise filtering conditions are considered to be real abnormal numbers, and the labels, weights and weighted scores of such numbers are output.

[0143] This model has a prominent target group portrait and high accuracy. The model is refined by analyzing the location characteristics and call behavior characteristics of historical abnormal numbers. At the same time, it combines the elderly model, student model, unemployed model and home worker model to exclude specific interference user groups, identify abnormal behaviors to the greatest extent, and minimize misjudgment of users.

[0144] In addition, the hardware requirements are relatively low, and the model data collection, user identification and shutdown speed are fast. The user account opening information, call bills, and signaling tracks are collected in real time, and the data is stored in real time. The user is weighted according to the minute, hour, and day model. The abnormal user data is pushed to the instruction sending system. From model identification to user shutdown, it takes minutes, which is fast.

[0145] The model is also highly extensible. It supports adding different labels such as the number of roaming cities, the number of terminal changes, the number of silent days, the account opening grid, the total traffic in the past 15 days, the number of unanswered calls in the past 15 days, the proportion of unanswered calls in the past 15 days, the number of International Mobile Equipment Identity (IMEI) codes used in the past 15 days, the average call duration, and adjusting the weights based on this model, thereby deriving other proprietary abnormal scenario recognition models.

[0146] In order to solve the above technical problems, the present application also provides an abnormal number identification device. Figure 3 FIG. 2 shows a schematic diagram of the structure of an abnormal number identification device provided in an embodiment of the present application; Figure 3 As shown, the device includes the following modules:

[0147] The acquisition module 301 is used to acquire target information of the number to be identified, where the target information includes at least one of the following items or any combination thereof: number attributes, user signaling, and call record data;

[0148] The recognition result determination module 302 is used to input the target information of the number to be recognized into the pre-trained abnormal number recognition model to obtain the recognition result output by the abnormal number recognition model; the abnormal number recognition model is trained based on the target information of the historical abnormal numbers;

[0149] The judgment module 303 is used to judge whether the suspected abnormal number meets the noise filtering condition when the recognition result indicates that the number to be recognized includes a suspected abnormal number; the noise filtering condition is established based on at least the unique number attributes and / or unique behavior habits of the target group, and the number of the target group is a normal number;

[0150] The determination module 304 is used to determine the suspected abnormal number that does not meet the noise filtering condition as the target abnormal number.

[0151] In some embodiments, the abnormal number identification device further includes: a condition building module, which is used to obtain target information of numbers corresponding to each target group before determining whether the suspected abnormal number meets the noise filtering condition;

[0152] Based on the target information of the numbers corresponding to each target group, determine the unique number attributes and / or unique behavior habits corresponding to each target group;

[0153] Corresponding noise filtering conditions are respectively constructed based on the unique number attributes and / or unique behavioral habits corresponding to each target group.

[0154] In some embodiments, the abnormal number identification device further includes:

[0155] An initial model building module, used to obtain the target information of the first historical abnormal number before inputting the target information of the number to be identified into the pre-trained abnormal number recognition model to build a first information set;

[0156] Extracting feature tags based on target information of each first historical abnormal number in the first information set;

[0157] With the goal of hitting more first historical abnormal numbers in the first information set, determining a preset number of valid tags from the feature tags;

[0158] Build an initial abnormal number recognition model based on valid labels;

[0159] The model training module is used to train the initial abnormal number recognition model to obtain the abnormal number recognition model.

[0160] In some embodiments, the model training module is specifically used to obtain target information of the second historical abnormal number and the historical normal number to construct a second information set;

[0161] Using a label extraction network to extract labels from target information of the second information set, to obtain a first target label;

[0162] Using a label matching network to match the first target label with the valid label, and filtering out a second target label that is a valid label from the first target label;

[0163] Based on the weights of the valid labels, a weight calculation network is used to calculate the weights of the second target labels to obtain a first abnormal number suspicion value of each number in the second information set;

[0164] Outputting an initial recognition result according to the suspected degree value of the first abnormal number using the recognition result output network;

[0165] When the initial recognition result does not meet the preset conditions, the valid labels and their weights are adjusted, and the label extraction network is used to extract the label of the target information of the second information set to obtain the first target label.

[0166] When the initial recognition result meets the preset conditions, the initial abnormal number recognition model under the current weight is determined to be the abnormal number recognition model; the preset conditions include: in the initial recognition result, the number of second historical abnormal numbers identified as suspected abnormal numbers is higher than the first preset threshold, and the number of historical normal numbers identified as suspected abnormal numbers is less than the second preset threshold.

[0167] In some embodiments, the recognition result determination module 302 is specifically configured to extract a label from the target information of the identification number using a label extraction network to obtain a third target label;

[0168] Using a label matching network to match the third target label with the valid label, and screening out a fourth target label that is a valid label from the third target label;

[0169] Based on the weights of the valid labels, a weight calculation network is used to calculate the weights of the fourth target labels to obtain a second abnormal number suspicion value of the number to be identified;

[0170] The recognition result output network is used to output the recognition result according to the suspected degree value of the second abnormal number.

[0171] The device provided in the embodiment of the present application is the same as the method in the above embodiment, so both have the same embodiments and beneficial effects, which will not be repeated here.

[0172] Figure 4 The hardware structure diagram of the abnormal number identification device provided in the embodiment of the present application is shown; Figure 4 As shown, the abnormal number identification device may include a processor 401 and a memory 402 storing computer program instructions.

[0173] Specifically, the processor 401 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application.

[0174] Memory 402 may include a large capacity memory for data or instructions. By way of example and not limitation, memory 402 may include a hard disk drive (HDD), a floppy disk drive, a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive or a combination of two or more of these. In appropriate cases, memory 402 may include a removable or non-removable (or fixed) medium. In appropriate cases, memory 402 may be inside or outside of an integrated gateway disaster recovery device. In a specific embodiment, memory 402 is a non-volatile solid-state memory.

[0175] The memory 402 may include a read-only memory (ROM), a random access memory (RAM), a magnetic disk storage medium device, an optical storage medium device, a flash memory device, an electrical, optical or other physical / tangible memory storage device. Thus, in general, the memory includes one or more tangible (non-transitory) computer-readable storage media (e.g., a memory device) encoded with software including computer-executable instructions, and when the software is executed (e.g., by one or more processors), it is operable to perform the operations described with reference to the method according to an aspect of the present disclosure.

[0176] The processor 401 implements any one of the abnormal number identification methods in the above embodiments by reading and executing the computer program instructions stored in the memory 402 .

[0177] In one example, the abnormal number identification device may further include a communication interface 403 and a bus 404. The processor 401, the memory 402, and the communication interface 403 are connected via the bus 404 and communicate with each other.

[0178] The communication interface 403 is mainly used to implement communication between various modules, devices, units and / or equipment in the embodiments of the present application.

[0179] Bus 404 includes hardware, software or both, and the parts of the identification device of abnormal number are coupled to each other.For example, but not limitation, bus may include accelerated graphics port (AGP) or other graphics bus, enhanced industrial standard architecture (EISA) bus, front side bus (FSB), hypertransport (HT) interconnection, industrial standard architecture (ISA) bus, infinite bandwidth interconnection, low pin count (LPC) bus, memory bus, micro channel architecture (MCA) bus, peripheral component interconnection (PCI) bus, PCI-Express (PCI-X) bus, serial advanced technology attachment (SATA) bus, video electronics standard association local (VLB) bus or other suitable bus or two or more of these combinations. In suitable cases, bus 404 may include one or more buses. Although the present application embodiment describes and shows specific bus, the application considers any suitable bus or interconnection.

[0180] In addition, in combination with the abnormal number identification method in the above embodiment, the embodiment of the present application can provide a computer storage medium for implementation. The computer storage medium stores computer program instructions; when the computer program instructions are executed by the processor, any abnormal number identification method in the above embodiment is implemented.

[0181] An embodiment of the present application also provides a computer program product, including a computer program, which, when executed, implements any one of the abnormal number identification methods in the above embodiments.

[0182] It should be clear that the present application is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present application is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present application.

[0183] The functional blocks shown in the above-described block diagram can be implemented as hardware, software, firmware or a combination thereof. When implemented in hardware, it can be, for example, an electronic circuit, an application specific integrated circuit (ASIC), appropriate firmware, a plug-in, a function card, etc. When implemented in software, the elements of the present application are programs or code segments that are used to perform the required tasks. The program or code segment can be stored in a machine-readable medium, or transmitted on a transmission medium or a communication link by a data signal carried in a carrier wave. "Machine-readable medium" can include any medium capable of storing or transmitting information. Examples of machine-readable media include electronic circuits, semiconductor memory devices, ROM, flash memory, erasable ROM (EROM), floppy disks, CD-ROMs, optical disks, hard disks, optical fiber media, radio frequency (RF) links, etc. The code segment can be downloaded via a computer network such as the Internet, an intranet, etc.

[0184] It should also be noted that the exemplary embodiments mentioned in this application describe some methods or systems based on a series of steps or devices. However, this application is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiment, or in a different order from the embodiment, or several steps can be performed simultaneously.

[0185] Aspects of the present disclosure are described above with reference to the flowchart and / or block diagram of a method, device, equipment, medium and product for identifying an abnormal number according to an embodiment of the present disclosure. It should be understood that each box in the flowchart and / or block diagram and the combination of each box in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine so that these instructions executed by the processor of the computer or other programmable data processing device enable the implementation of the function / action specified in one or more boxes of the flowchart and / or block diagram. Such a processor can be, but is not limited to, a general-purpose processor, a special-purpose processor, a special application processor or a field programmable logic circuit. It can also be understood that each box in the block diagram and / or flowchart and the combination of boxes in the block diagram and / or flowchart can also be implemented by dedicated hardware that performs a specified function or action, or can be implemented by a combination of dedicated hardware and computer instructions.

[0186] The above is only a specific implementation of the present application. Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, modules and units described above can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. It should be understood that the protection scope of the present application is not limited to this. Any technician familiar with the technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in this application, and these modifications or replacements should be included in the protection scope of this application.

Claims

1. A method for identifying abnormal numbers, It is characterized in that include: Obtain target information of the number to be identified, the target information including at least any one of the following or any combination thereof: number attributes, user signaling, and call record data; Inputting the target information of the number to be identified into a pre-trained abnormal number identification model to obtain an identification result output by the abnormal number identification model; the abnormal number identification model is trained based on the target information of historical abnormal numbers; In the case where the identification result indicates that the number to be identified includes a suspected abnormal number, determining whether the suspected abnormal number satisfies a noise filtering condition; The noise filtering condition is established based at least on the unique number attributes and / or unique behavior habits of the target group, and the numbers of the target group are normal numbers; The suspected abnormal number that does not meet the noise filtering condition is determined as the target abnormal number.

2. The method for identifying abnormal numbers according to claim 1, It is characterized in that There are multiple target groups; Before determining whether the suspected abnormal number meets the noise filtering condition, the method further includes: Obtain target information of numbers corresponding to each target group; Based on the target information of the numbers corresponding to the target groups, determining the unique number attributes and / or unique behavior habits corresponding to the target groups; The corresponding noise filtering conditions are respectively constructed based on the unique number attributes and / or unique behavior habits corresponding to each of the target groups.

3. The method for identifying abnormal numbers according to claim 2, It is characterized in that The target groups include at least any one or any combination of the following: the elderly, students, unemployed persons and home workers; Number attributes include at least the account opening age range and / or package level; The behavior habits include at least any one or any combination of the following: the number of calls, the number of calls, the continuous moving time period, the location of the permanent base station, the type of commonly used software, the traffic usage and the target LAN usage.

4. The method for identifying abnormal numbers according to claim 3, It is characterized in that The unique number attributes of the elderly include at least: The age in the account opening certificate is greater than the first preset age; The unique behavioral habits of the elderly include at least one of the following or any combination: In the first preset historical time period, the proportion of the traffic used in shopping software and video software to the total traffic used exceeds the preset ratio; During the second preset historical time period, the average daily number of calls is greater than the first preset number; The earliest continuous movement time in a single day is earlier than the first preset time in the morning of that day.

5. The method for identifying abnormal numbers according to claim 3, It is characterized in that The student's unique number attributes include at least one of the following or any combination: The age of the account opening certificate is less than the second preset age; the second preset age is less than the first preset age; The package level is lower than the preset level; The student’s unique behavioral habits include at least one of the following or any combination: The top n base stations in terms of dwell time on weekdays are the base stations around the school; n is a positive integer; Third, there was behavior of using the campus network within the preset historical time period.

6. The method for identifying abnormal numbers according to claim 3, It is characterized in that The unique number attributes of unemployed persons include at least: The age in the account opening certificate is greater than the third preset age and less than the fourth preset age; The unique behavioral habits of unemployed persons include at least one of the following or any combination thereof: In the fourth preset historical time period, the average daily number of calls is less than the second preset number; The daily traffic usage is greater than the preset traffic usage, and the frequency of using the following software is higher than the first frequency: video software, game software and communication software; There is overlap between the top x base stations in terms of daytime residence time and the top y base stations in terms of nighttime residence time; x and y are positive integers.

7. The method for identifying abnormal numbers according to claim 3, It is characterized in that The unique behavioral habits of home workers include at least one of the following or any combination of the following: The base stations with the highest i-th longest residence time during the day and the base stations with the highest j-th longest residence time at night overlap; i and j are positive integers; In a fifth preset historical time period, the average daily number of calls is greater than a third preset number and less than a fourth preset number; and the third preset number is greater than the fourth preset number.

8. The method for identifying abnormal numbers according to any one of claims 1 to 7, It is characterized in that Before inputting the target information of the number to be identified into the pre-trained abnormal number identification model, the method further includes: Acquire target information of a first historical abnormal number to construct a first information set; extracting a feature label based on target information of each of the first historical abnormal numbers in the first information set; With the goal of hitting more of the first historical abnormal numbers in the first information set, determining a preset number of valid tags from the feature tags; Building an initial abnormal number recognition model based on the valid tags; The initial abnormal number recognition model is trained according to the target information of the second historical abnormal number and the historical normal number to obtain the abnormal number recognition model.

9. The method for identifying abnormal numbers according to claim 8, It is characterized in that The initial abnormal number recognition model includes a label extraction network, a label matching network, a weight calculation network and a recognition result output network; The training of the initial abnormal number recognition model to obtain the abnormal number recognition model includes: Acquire target information of the second historical abnormal number and the historical normal number to construct a second information set; Using the label extraction network to extract labels from the target information of the second information set to obtain a first target label; Using the label matching network to match the first target label with the valid label, and filtering out a second target label belonging to the valid label from the first target label; Based on the weights of the valid tags, the weight calculation network is used to perform weight calculation on each of the second target tags to obtain a first abnormal number suspicion value of each number in the second information set; Using the recognition result output network, outputting an initial recognition result according to the first abnormal number suspicion value; When the initial recognition result does not meet the preset condition, adjusting each of the valid labels and the weights thereof, and returning to the step of extracting labels from the target information of the second information set using the label extraction network to obtain a first target label; When the initial recognition result satisfies the preset conditions, the initial abnormal number recognition model under the current weight is determined to be the abnormal number recognition model; the preset conditions include: in the initial recognition result, the number of the second historical abnormal numbers identified as suspected abnormal numbers is higher than the first preset threshold, and the number of the historical normal numbers identified as suspected abnormal numbers is less than the second preset threshold.

10. The method for identifying abnormal numbers according to claim 9, It is characterized in that The step of inputting the target information of the number to be identified into a pre-trained abnormal number identification model to obtain the identification result output by the abnormal number identification model includes: Using the label extraction network to extract labels from the target information of the number to be identified, to obtain a third target label; Using the label matching network to match the third target label with the valid label, and screening out a fourth target label belonging to the valid label from the third target label; Based on the weights of the valid tags, the weight calculation network is used to perform weight calculation on each of the fourth target tags to obtain a second abnormal number suspicion value of the number to be identified; The recognition result output network is used to output the recognition result according to the second abnormal number suspicion value.

11. A device for identifying abnormal numbers, It is characterized in that The device comprises: An acquisition module, used to acquire target information of the number to be identified, the target information including at least any one of the following or any combination thereof: number attributes, user signaling, and call record data; The recognition result determination module is used to input the target information of the number to be recognized into a pre-trained abnormal number recognition model to obtain the recognition result output by the abnormal number recognition model; the abnormal number recognition model is trained according to the target information of historical abnormal numbers; A judgment module, configured to judge whether the suspected abnormal number satisfies a noise filtering condition when the recognition result indicates that the number to be recognized includes a suspected abnormal number; the noise filtering condition is established based on at least a unique number attribute and / or a unique behavior habit of a target group, and the number of the target group is a normal number; The determination module is used to determine the suspected abnormal number that does not meet the noise filtering condition as the target abnormal number.

12. A device for identifying abnormal numbers, It is characterized in that The device comprises: a processor and a memory storing computer program instructions; When the processor executes the computer program instructions, the method for identifying an abnormal number according to any one of claims 1 to 10 is implemented.

13. A computer-readable storage medium, It is characterized in that The computer-readable storage medium stores computer program instructions, and when the computer program instructions are executed by a processor, the method for identifying an abnormal number according to any one of claims 1 to 10 is implemented.

14. A computer program product, It is characterized in that When the instructions in the computer program product are executed by a processor of an electronic device, the electronic device executes the abnormal number identification method as described in any one of claims 1 to 10.