Abnormal number identification method, device and electronic equipment
By combining the matching rule model and the number identification model on the terminal device, identifying whether the calling number requested by the call is an exception number, the problem of not identifying uncalibrated exception numbers in the prior art is solved, and the accuracy of exception number recognition is improved and user experience.
Patent Information
- Application Number
- CN202110713306.1
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-06-25
- Publication Date
- 2024-11-01
- Estimated Expiration
- 2041-06-25
AI Technical Summary
In the prior art, only pre-calibrated numbers can be identified by installing security software on terminal devices to identify abnormal numbers, and uncalibrated abnormal numbers cannot be identified, which reduces the accuracy of identification and affects the user experience.
Use the matching rule model and the pre-trained number recognition model to obtain call request information in real time and match based on the preset matching rule model. If the match is successful, input it to the number recognition model for identification. If the recognition result is abnormal, the call request will be intercepted. , improve the accuracy of abnormal number recognition.
By combining the matching rule model with the number recognition model, it is possible to accurately identify whether the calling number requested by the call is an exception number, improving the accuracy and user experience of exception number recognition.
Smart Images

Figure CN115529596B_ABST
Abstract
Description
Technical Field
[0001] The embodiments of the present invention relate to the field of artificial intelligence technology, and in particular to an abnormal number recognition method, device and electronic device. Background Art
[0002] With the advent of the big data era, the rapid development of mobile communications has brought many conveniences to people's work, study and life. But at the same time, various sales, advertising, fraud and other harassing calls are increasing, which not only disrupts the normal order of telecommunications operations, but also brings great trouble to users.
[0003] In the prior art, the most common method to prevent harassing calls is to install security software on the terminal device. The security software can pre-determine whether the number is an abnormal number through the marks of other users, and mark the number. When the user receives an unfamiliar call, if the number is a pre-marked abnormal number, the security software can give a security reminder for the number, and the user can decide whether to answer or reject the call.
[0004] However, the method of identifying abnormal numbers by installing security software on the terminal device can only identify numbers pre-marked by other users, and cannot identify unmarked abnormal numbers, which reduces the accuracy of abnormal number identification and affects the user experience. Summary of the invention
[0005] The embodiments of the present invention provide a method, a device and an electronic device for identifying an abnormal number, so as to improve the accuracy of identifying an abnormal number.
[0006] In a first aspect, an embodiment of the present invention provides a method for identifying an abnormal number, comprising:
[0007] Get call request information in real time;
[0008] Matching the call request information based on a preset matching rule model to obtain a matching result;
[0009] If the matching result is a successful match, the call request information is input into a pre-trained number recognition model for recognition to obtain a recognition result;
[0010] If the identification result is that the number is abnormal, the call request corresponding to the call request information is intercepted.
[0011] Optionally, matching the call request information based on a preset matching rule model to obtain a matching result includes:
[0012] The call request information's call frequency, call duration, and historical call record information are matched based on a preset matching rule model to obtain a matching result.
[0013] Optionally, before matching the call request information based on a preset matching rule model to obtain a matching result, the method further includes:
[0014] Determine whether the calling number corresponding to the call request information is in a preset blacklist library or a whitelist library;
[0015] If the calling number corresponding to the call request information is in the preset whitelist library, continue to call the called number corresponding to the call request information;
[0016] If the calling number corresponding to the call request information is in a preset blacklist library, the call request corresponding to the call request information is intercepted.
[0017] Optionally, after matching the call request information based on a preset matching rule model to obtain a matching result, the method further includes:
[0018] If the matching result is a matching failure, the called number corresponding to the call request information continues to be called.
[0019] Optionally, if the matching result is a successful match, the call request information is input into a pre-trained number recognition model for recognition, and before obtaining the recognition result, the method further includes:
[0020] Obtain a preset binary classification model and a call request information training set;
[0021] Inputting the call request information training set into the preset binary classification model for training to obtain a first classification model;
[0022] Get the preset gradient boosting decision tree model;
[0023] Inputting the request information training set into the preset gradient boosting decision tree model for training to obtain a second classification model;
[0024] According to a preset weight allocation rule, weight values are allocated to the first classification model and the second classification model to obtain a number recognition model.
[0025] Optionally, before obtaining the preset binary classification model and the call request information training set, the method further includes:
[0026] Obtain network data in the operation domain and user data in the business domain;
[0027] Perform data cleaning on the network data of the operation domain and the user data of the business domain according to preset cleaning rules to obtain a positive sample training set;
[0028] Obtaining a negative sample training set according to the positive sample training set and the full data set corresponding to the network data of the operation domain and the user data of the business domain;
[0029] The positive sample training set and the negative sample training set are standardized to obtain a call request information training set.
[0030] In a second aspect, an embodiment of the present invention provides an abnormal number identification device, comprising:
[0031] An acquisition module is used to acquire call request information in real time;
[0032] A processing module, used for matching the call request information based on a preset matching rule model to obtain a matching result;
[0033] The processing module is further configured to input the call request information into a pre-trained number recognition model for recognition to obtain a recognition result if the matching result is a successful match;
[0034] The processing module is further configured to intercept a call request corresponding to the call request information if the identification result is that the number is abnormal.
[0035] In a third aspect, an embodiment of the present invention provides an electronic device, including: at least one processor and a memory;
[0036] The memory stores computer-executable instructions;
[0037] The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the abnormal number identification method as described in any one of the first aspects.
[0038] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer-executable instructions. When a processor executes the computer-executable instructions, the abnormal number identification method as described in any one of the first aspects is implemented.
[0039] In a fifth aspect, an embodiment of the present invention provides a computer program product, including a computer program. When the computer program is executed by a processor, it implements the abnormal number identification method described in the first aspect and various possible designs of the first aspect.
[0040] The embodiments of the present invention provide a method, device and electronic device for identifying abnormal numbers. After adopting the above scheme, call request information can be obtained in real time, and then the call request information is matched based on a preset matching rule model to obtain a matching result. If the matching result is a successful match, the call request information is input into a pre-trained number recognition model for recognition to obtain a recognition result. If the recognition result is that the number is abnormal, the call request corresponding to the call request information is intercepted. By combining the matching rule model with the number recognition model to determine whether the calling number corresponding to the call request information is an abnormal number, the accuracy of abnormal number recognition is improved, thereby improving the user experience. BRIEF DESCRIPTION OF THE DRAWINGS
[0041] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.
[0042] Figure 1 A schematic diagram of an application of abnormal user marking in the prior art;
[0043] Figure 2 A schematic diagram of the architecture of an application system of the abnormal number identification method provided by an embodiment of the present invention;
[0044] Figure 3 A schematic diagram of a flow chart of an abnormal number identification method provided by an embodiment of the present invention;
[0045] Figure 4 A schematic diagram of the principle of the abnormal number identification method provided by an embodiment of the present invention;
[0046] Figure 5 A schematic diagram of the structure of an abnormal number identification device provided by an embodiment of the present invention;
[0047] Figure 6 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention. DETAILED DESCRIPTION
[0048] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0049] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchangeable where appropriate, so that the embodiments of the present invention described herein can also include other sequential instances in addition to those illustrated or described. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.
[0050] The calling numbers of harassing calls such as sales promotion, advertising, fraud, etc. can all be defined as abnormal numbers. In order to build a green and safe communication network, the trouble to users can be reduced by identifying abnormal numbers. In the prior art, abnormal users can be identified by installing security software on the terminal device. Specifically, the security software can pre-determine whether the number is a harassing call based on the marks of other users, and mark the number. When the user receives an unfamiliar call, if the number is a pre-marked number, the security software can give a security reminder for the number, and the user can decide whether to answer or reject the call. For example, Figure 1 The schematic diagram of the application of abnormal user marking in the prior art is as follows: Figure 1 As shown, in this embodiment, the number 123456789 has been marked as an abnormal number by other users in advance. When the number 123456789 is used as the caller, in addition to the answer and hang up controls, the incoming call interface also displays a "nuisance call" text prompt, thereby reminding the user that the number is a nuisance call. However, the method of identifying abnormal numbers by installing security software on the terminal device can only identify numbers pre-marked by other users, and cannot identify abnormal users who have not been marked, which reduces the accuracy of abnormal number identification and affects the user experience.
[0051] Based on the above technical problems, the present application determines whether the calling number corresponding to the call request information is an abnormal number by combining the matching rule model with the number recognition model, thereby achieving the technical effect of both improving the accuracy of abnormal number recognition and improving the user experience.
[0052] Figure 2 A schematic diagram of the architecture of an application system of the abnormal number identification method provided by an embodiment of the present invention, such as Figure 2As shown, in this embodiment, the application system includes: a database 201, a server 202, a first terminal device 203 and a second terminal device 204. The database 201 may store a call request information training set, and the server 202 may obtain the call request information training set from the database 201, and then perform training according to the call request information training set to obtain a number recognition model. Then, the number recognition model may be deployed in the first terminal device 203 and the second terminal device 204. The first terminal device 203 may be used as a device corresponding to the calling number, and the second terminal device 204 may be used as a device corresponding to the called number. In the process of the first terminal device 203 calling the second terminal device 204, the calling number corresponding to the first terminal device 203 may be identified by the number recognition model to determine whether it is an abnormal number.
[0053] The technical solution of the present invention is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.
[0054] Figure 3 Schematic diagram of the process of the abnormal number identification method provided by the embodiment of the present invention. The method of this embodiment can be executed by the second terminal device 204 or by a server with a number identification model deployed thereon. Figure 3 As shown, the method of this embodiment may include:
[0055] S301: Acquire call request information in real time.
[0056] In this embodiment, whether there is a call request can be monitored in real time. If there is a call request, call request information corresponding to the call request can be obtained in real time.
[0057] The call request information may include information such as the calling number and the called number.
[0058] S302: Match the call request information based on a preset matching rule model to obtain a matching result.
[0059] In this embodiment, after the call request information is acquired, the call request information may be matched based on a preset matching rule model to obtain a matching result.
[0060] Further, the call request information is matched based on a preset matching rule model to obtain a matching result, which may specifically include:
[0061] The call request information's call frequency, call duration, and historical call record information are matched based on a preset matching rule model to obtain a matching result.
[0062] Specifically, a matching rule model may be pre-built, and then the call request information may be matched using the matching rule model. The specific construction process may be:
[0063] Based on the number pool filed by the relevant departments, we conducted feature analysis and user profiling of abnormal numbers, and analyzed call data features, APP usage habits, regional distribution, age distribution, terminal distribution, etc. We found that abnormal numbers mainly have the following characteristics: high call frequency, short call duration, extremely high proportion of callers, dialing numbers from the same segment in turn, mainly in remote areas or developed cities, and more "one-ring" call behaviors, etc. Then we can confirm the corresponding critical value based on the above characteristics, and then build a matching rule model.
[0064] After the matching rule model is built, the call frequency of the calling number of the call request information, the call duration of each historical call record, and the historical dial record information can be matched according to the matching rule model, wherein the historical dial record information may include the proportion of callers, whether the dialed numbers are the same number segment and dialed in sequence, whether the dialed areas are mainly remote areas or developed cities, whether there are many "one-ring" call behaviors, etc. If there are behaviors exceeding the preset number of item thresholds in the above matching items, the matching result is a successful match, otherwise the matching result is a failed match. The item number threshold can be customized according to the actual application scenario. For example, the item number threshold can be any value from 2 to 4.
[0065] S303: If the matching result is successful, the call request information is input into the pre-trained number recognition model for recognition to obtain a recognition result.
[0066] In this embodiment, if the matching result is a successful match, it indicates that the calling number corresponding to the call request information has the risk of being an abnormal number. In order to improve the accuracy of determining the abnormal number, the call request information can be input into a pre-trained number recognition model for recognition to obtain a recognition result.
[0067] If the matching result is a match failure, the called number corresponding to the call request information can continue to be called.
[0068] S304: If the identification result is that the number is abnormal, intercepting the call request corresponding to the call request information.
[0069] In this embodiment, if the identification result is that the number is abnormal, indicating that the calling number is a harassing call, the call request corresponding to the call request information can be directly intercepted, and the called number corresponding to the call request information will no longer be called, thereby reducing the trouble to the user.
[0070] After adopting the above scheme, call request information can be obtained in real time, and then the call request information is matched based on the preset matching rule model to obtain a matching result. If the matching result is a successful match, the call request information is input into the pre-trained number recognition model for recognition to obtain a recognition result. If the recognition result is that the number is abnormal, the call request corresponding to the call request information is intercepted. By combining the matching rule model with the number recognition model to determine whether the calling number corresponding to the call request information is an abnormal number, the accuracy of abnormal number recognition is improved, thereby improving the user experience.
[0071] based on Figure 3 The method, the examples of this specification also provide some specific implementation plans of the method, which are described below.
[0072] In addition, in another embodiment, before S302, the method may further include:
[0073] Determine whether the calling number corresponding to the call request information is in a preset blacklist library or whitelist library.
[0074] If the calling number corresponding to the call request information is in the preset white list library, then continue to call the called number corresponding to the call request information.
[0075] If the calling number corresponding to the call request information is in a preset blacklist library, the call request corresponding to the call request information is intercepted.
[0076] In this embodiment, a blacklist library and a whitelist library can be determined in advance based on the calibration of other users. Blacklist numbers are stored in the blacklist library, and whitelist numbers are stored in the whitelist library. After obtaining the call request information, it is possible to first determine whether the calling number corresponding to the call request information is in the blacklist library or the whitelist library, and then further process it based on the determination result, thereby improving the efficiency and accuracy of abnormal number identification.
[0077] Figure 4 A schematic diagram of the principle of the abnormal number identification method provided by the embodiment of the present invention is shown as follows: Figure 4 As shown, in this embodiment, it may include:
[0078] First, obtain the call request record information, and then check whether the caller number corresponding to the call request record information is in the blacklist and whitelist database. If it is in the whitelist database, continue to call. If it is in the blacklist database, perform real-time interception processing. If it is neither in the whitelist database nor in the blacklist database, match the rule model. If the match is not successful, continue to call; if the match is successful, identify it through the number recognition model. If the recognition result is an abnormal number, it will be intercepted in real time, otherwise continue to call.
[0079] In addition, in another embodiment, before S303, the method may further include:
[0080] Get the preset binary classification model and call request information training set.
[0081] The call request information training set is input into the preset binary classification model for training to obtain a first classification model.
[0082] Get the preset gradient boosting decision tree model.
[0083] The request information training set is input into the preset gradient boosting decision tree model for training to obtain a second classification model.
[0084] According to a preset weight allocation rule, weight values are allocated to the first classification model and the second classification model to obtain a number recognition model.
[0085] In this embodiment, the preset binary classification model may be an XGBoost binary classification model, and the preset gradient boosting decision tree model may be a LightGBM binary classification model. The preset weight distribution rule may be customized according to the actual application scenario, for example, it may be 5:5 or 4:6, which is not specifically limited here.
[0086] Furthermore, before obtaining the preset binary classification model and the call request information training set, the method may further include:
[0087] Obtain network data of the O domain (operation support system) and user data of the B domain (business support system).
[0088] The network data in the operation domain and the user data in the business domain are cleaned according to the preset cleaning rules to obtain a positive sample training set.
[0089] A negative sample training set is obtained according to the positive sample training set and the full data set corresponding to the network data of the operation domain and the user data of the business domain.
[0090] The positive sample training set and the negative sample training set are standardized to obtain a call request information training set.
[0091] Specifically, the network data in the O domain may include: XDR signaling and MR data. After obtaining the network data in the O domain and the user data in the B domain, the data may be preliminarily cleaned according to preset cleaning rules to generate positive and negative sample data, namely, a positive sample training set and a negative sample training set.
[0092] The specific cleaning process may be: firstly collect all sample data during the statistical period (for example, half a year or a year), filter the call data, roaming data, terminal data, and basic information data of all users, and then perform data cleaning on the filtered target data, for example:
[0093] Screen out the numbers of fraudulent users that have been shut down, the numbers in the fraudulent phone pool (including but not limited to the numbers registered with the relevant departments), and the numbers of users with high probability of suspected fraud in key high-risk provinces. Most fraudulent users have the characteristics of one ring, short call duration, high call frequency, and mainly outgoing calls. Combined with the user-related information data in the above steps, form a preliminary positive sample training set. Remove fraudulent users from all users and regard them as normal non-fraudulent users to form a preliminary negative sample training set.
[0094] Then, the two types of data are processed for missing values, such as mean filling, 0 filling, mode filling, etc., outlier processing, such as eliminating negative values, eliminating values that are not in the enumeration values, etc., logarithmic transformation and other methods are used to clean the preliminary positive and negative example samples into standard positive and negative examples, and label the two types of users, and split them into a call request information training set and a call request information test set according to a predetermined ratio (for example, 7:3 or 8:2).
[0095] After the number recognition model training is completed, the number recognition model can be tested by requesting an information test set, and the weight value of the number recognition model can be reversely updated according to the test results until the recognition accuracy of the number recognition model reaches a preset accuracy threshold.
[0096] This application establishes temporary established model rules by assigning unique feature values to single features of the number recognition model. Multi-user single-threaded analysis and judgment of all roaming users within the sample data cycle time is based on established rules, and a list of suspected roaming fraud users is output. For the machine learning classification model, real-time classification prediction is performed on users across the entire network, and a list of users with high probability of suspected roaming fraud is also output. The judgment result for the same user depends on the weighted average of the classification results of the two models. Combined with the output list and sample users, regression verification of model integrity and accuracy, combined with the precision and recall of the fusion model output results, the model feature weight values are reduced and increased, and the rule model and machine learning model are further iterated and optimized.
[0097] Based on the same idea, the embodiments of this specification also provide a device corresponding to the above method. Figure 5 A schematic diagram of the structure of an abnormal number identification device provided by an embodiment of the present invention, such as Figure 5 As shown, it may include:
[0098] The acquisition module 501 is used to acquire call request information in real time.
[0099] The processing module 502 is used to match the call request information based on a preset matching rule model to obtain a matching result.
[0100] In this embodiment, the processing module 502 is further used for:
[0101] The call request information's call frequency, call duration, and historical call record information are matched based on a preset matching rule model to obtain a matching result.
[0102] The processing module 502 is further configured to input the call request information into a pre-trained number recognition model for recognition to obtain a recognition result if the matching result is a successful match.
[0103] In this embodiment, the processing module 502 is further used for:
[0104] If the matching result is a matching failure, the called number corresponding to the call request information continues to be called.
[0105] The processing module 502 is further configured to intercept a call request corresponding to the call request information if the identification result is that the number is abnormal.
[0106] In addition, in another embodiment, the processing module 502 is further configured to:
[0107] Determine whether the calling number corresponding to the call request information is in a preset blacklist library or a whitelist library.
[0108] If the calling number corresponding to the call request information is in the preset white list library, the called number corresponding to the call request information continues to be called.
[0109] If the calling number corresponding to the call request information is in a preset blacklist library, the call request corresponding to the call request information is intercepted.
[0110] In addition, in another embodiment, the processing module 502 is further configured to:
[0111] Get the preset binary classification model and call request information training set.
[0112] The call request information training set is input into the preset binary classification model for training to obtain a first classification model.
[0113] Get the preset gradient boosting decision tree model.
[0114] The request information training set is input into the preset gradient boosting decision tree model for training to obtain a second classification model.
[0115] According to a preset weight allocation rule, weight values are allocated to the first classification model and the second classification model to obtain a number recognition model.
[0116] Furthermore, the processing module 502 is further configured to:
[0117] Obtain network data in the operation domain and user data in the business domain.
[0118] The network data in the operation domain and the user data in the business domain are cleaned according to preset cleaning rules to obtain a positive sample training set.
[0119] A negative sample training set is obtained according to the positive sample training set and the full data set corresponding to the network data of the operation domain and the user data of the business domain.
[0120] The positive sample training set and the negative sample training set are standardized to obtain a call request information training set.
[0121] The device provided by the embodiment of the present invention can achieve the above Figure 3 The implementation principle and technical effect of the method in the embodiment shown are similar and will not be described in detail here.
[0122] Figure 6 A schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention is shown in FIG. Figure 6 As shown, the device 600 provided in this embodiment includes: at least one processor 601 and a memory 602. The processor 601 and the memory 602 are connected via a bus 603.
[0123] In a specific implementation process, at least one processor 601 executes the computer-executable instructions stored in the memory 602, so that at least one processor 601 executes the method in the above method embodiment.
[0124] The specific implementation process of the processor 601 can be found in the above method embodiment, and its implementation principle and technical effect are similar, so this embodiment will not be repeated here.
[0125] In the above Figure 6In the illustrated embodiment, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.
[0126] The memory may include a high-speed RAM memory, and may also include a non-volatile storage NVM, such as at least one disk storage.
[0127] The bus may be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus may be divided into an address bus, a data bus, a control bus, etc. For ease of representation, the bus in the drawings of the present application is not limited to only one bus or one type of bus.
[0128] An embodiment of the present invention further provides a computer-readable storage medium, in which computer-executable instructions are stored. When a processor executes the computer-executable instructions, the abnormal number identification method of the above method embodiment is implemented.
[0129] An embodiment of the present invention further provides a computer program product, including a computer program, and when the computer program is executed by a processor, the abnormal number identification method as described above is implemented.
[0130] The computer-readable storage medium mentioned above can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special-purpose computer.
[0131] An exemplary readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). Of course, the processor and the readable storage medium can also exist in the device as discrete components.
[0132] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, disk or optical disk and other media that can store program codes.
[0133] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for identifying abnormal numbers, characterized in that: include: Get call request information in real time; Matching the call request information based on a preset matching rule model to obtain a matching result; If the matching result is a successful match, the call request information is input into a pre-trained number recognition model for recognition to obtain a recognition result; If the identification result is that the number is abnormal, intercepting the call request corresponding to the call request information; After matching the call request information based on the preset matching rule model to obtain a matching result, the method further includes: If the matching result is a match failure, continue to call the called number corresponding to the call request information; If the matching result is a successful match, the call request information is input into a pre-trained number recognition model for recognition, and before obtaining the recognition result, the method further includes: Obtain a preset binary classification model and a call request information training set; Inputting the call request information training set into the preset binary classification model for training to obtain a first classification model; Get the preset gradient boosting decision tree model; Inputting the request information training set into the preset gradient boosting decision tree model for training to obtain a second classification model; Allocate weight values to the first classification model and the second classification model according to a preset weight allocation rule to obtain a number recognition model; After allocating weight values to the first classification model and the second classification model according to a preset weight allocation rule to obtain a number recognition model, the method further includes: The number recognition model is tested through a call request information test set, and the weight value of the number recognition model is reversely updated according to the test result until the recognition accuracy of the number recognition model reaches a preset accuracy threshold.
2. The method according to claim 1, characterized in that The matching of the call request information based on a preset matching rule model to obtain a matching result includes: The call frequency, call duration and historical call record information of the call request information are matched based on a preset matching rule model to obtain a matching result.
3. The method according to claim 1, characterized in that Before matching the call request information based on the preset matching rule model to obtain a matching result, the method further includes: Determine whether the calling number corresponding to the call request information is in a preset blacklist library or a whitelist library; If the calling number corresponding to the call request information is in the preset whitelist library, continue to call the called number corresponding to the call request information; If the calling number corresponding to the call request information is in a preset blacklist library, the call request corresponding to the call request information is intercepted.
4. The method according to claim 1, characterized in that: Before obtaining the preset binary classification model and the call request information training set, the method further includes: Obtain network data in the operation domain and user data in the business domain; Perform data cleaning on the network data of the operation domain and the user data of the business domain according to preset cleaning rules to obtain a positive sample training set; Obtaining a negative sample training set according to the positive sample training set and the full data set corresponding to the network data of the operation domain and the user data of the business domain; The positive sample training set and the negative sample training set are standardized to obtain a call request information training set.
5. An abnormal number identification device, characterized in that: include: An acquisition module is used to acquire call request information in real time; A processing module, used for matching the call request information based on a preset matching rule model to obtain a matching result; The processing module is further configured to input the call request information into a pre-trained number recognition model for recognition to obtain a recognition result if the matching result is a successful match; The processing module is further configured to intercept the call request corresponding to the call request information if the identification result is that the number is abnormal; The processing module is further configured to continue calling the called number corresponding to the call request information if the matching result is a matching failure; The processing module is further used to obtain a preset binary classification model and a call request information training set; input the call request information training set into the preset binary classification model for training to obtain a first classification model; Obtain a preset gradient boosting decision tree model; input the request information training set into the preset gradient boosting decision tree model for training to obtain a second classification model; assign weight values to the first classification model and the second classification model according to a preset weight assignment rule to obtain a number recognition model; The processing module is further used to test the number recognition model through a call request information test set, and reversely update the weight value of the number recognition model according to the test result until the recognition accuracy of the number recognition model reaches a preset accuracy threshold.
6. An electronic device, characterized in that: include: at least one processor and memory; The memory stores computer-executable instructions; The at least one processor executes the computer-executable instructions stored in the memory, so that the at least one processor executes the abnormal number identification method according to any one of claims 1 to 4.
7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and when the processor executes the computer-executable instructions, the abnormal number identification method according to any one of claims 1 to 4 is implemented.
8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the abnormal number identification method according to any one of claims 1 to 4 is implemented.
